Spineweb_UNet++_smp.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\manjw\\.conda\\envs\\dataEngineering\\Lib\\site-packages\\torchaudio\\backend\\utils.py:74: UserWarning: No audio backend is available.\n",
      "  warnings.warn(\"No audio backend is available.\")\n",
      "c:\\Users\\manjw\\.conda\\envs\\dataEngineering\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import nibabel as nib\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import Dataset\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.transforms import ToTensor\n",
    "import torchmetrics.functional as evafunc\n",
    "import torchvision.transforms as transforms\n",
    "import numpy as np\n",
    "import torchvision.models as models\n",
    "from segmentation_models_pytorch.losses import DiceLoss, FocalLoss\n",
    "import segmentation_models_pytorch as smp\n",
    "import os\n",
    "from PIL import Image\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 自定义数据集类\n",
    "class SpineWeb15(Dataset):\n",
    "    def __init__(self, data_path,transform1=None,transform2=None):\n",
    "        self.data_path = data_path\n",
    "        self.transform1 = transform1\n",
    "        self.transform2 = transform2\n",
    "        \n",
    "    def __len__(self):\n",
    "        file_list = os.listdir(self.data_path)  # 列出文件夹中的所有文件和文件夹\n",
    "        file_count = len(file_list)  # 获取文件数量\n",
    "        return file_count\n",
    "    \n",
    "    def __getitem__(self, index):\n",
    "        metadata = np.load(self.data_path+str(index)+\".npy\")\n",
    "        image = metadata[0]\n",
    "        label = metadata[1]\n",
    "\n",
    "        # 找到数组中的最小值和最大值\n",
    "        min_value = np.min(image)\n",
    "        max_value = np.max(image)\n",
    "\n",
    "        # # 将数组元素限制在0到255之间（类似于归一化）\n",
    "        image = (image - min_value) * (255 / (max_value - min_value))\n",
    "\n",
    "        image_data = Image.fromarray(image)\n",
    "        label_data = Image.fromarray(label)\n",
    "        # image_data = torch.from_numpy(image).float()\n",
    "        # label_data = torch.from_numpy(label).float()\n",
    "        \n",
    "        if self.transform1:\n",
    "            image_data = self.transform1(image_data)\n",
    "        if self.transform2:\n",
    "            label_data = self.transform2(label_data)\n",
    "        return image_data, label_data\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_pixel_accuracy(prediction, target):\n",
    "    # 将预测结果和目标标签进行逐像素比较，并计算正确预测像素的比例\n",
    "    correct_pixels = (prediction == target).sum().item()\n",
    "    total_pixels = target.numel()\n",
    "    return correct_pixels, total_pixels\n",
    "\n",
    "def calculate_iou(prediction, target):\n",
    "    intersection = torch.logical_and(prediction, target).sum()\n",
    "    union = torch.logical_or(prediction, target).sum()\n",
    "    iou = intersection / union\n",
    "    return iou.item()\n",
    "\n",
    "def calculate_dice(prediction, target):\n",
    "    intersection = torch.logical_and(prediction, target).sum()\n",
    "    dice = (2 * intersection) / (prediction.sum() + target.sum())\n",
    "    return dice.item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/200], Avg Loss: 0.02814, Avg IoU: 0.3742, Avg Dice: 0.5239,Test IoU: 0.4615, Test Dice: 0.6144, Train Acc: 0.9259, Test Acc: 0.9572\n",
      "Epoch [2/200], Avg Loss: 0.02791, Avg IoU: 0.5675, Avg Dice: 0.7079,Test IoU: 0.5967, Test Dice: 0.7273, Train Acc: 0.9695, Test Acc: 0.9761\n",
      "Epoch [3/200], Avg Loss: 0.02783, Avg IoU: 0.6742, Avg Dice: 0.7901,Test IoU: 0.6682, Test Dice: 0.7802, Train Acc: 0.9815, Test Acc: 0.9831\n",
      "Epoch [4/200], Avg Loss: 0.02778, Avg IoU: 0.7262, Avg Dice: 0.8269,Test IoU: 0.7045, Test Dice: 0.8067, Train Acc: 0.9859, Test Acc: 0.9863\n",
      "Epoch [5/200], Avg Loss: 0.02775, Avg IoU: 0.7585, Avg Dice: 0.8489,Test IoU: 0.7244, Test Dice: 0.8204, Train Acc: 0.9884, Test Acc: 0.9878\n",
      "Epoch [6/200], Avg Loss: 0.02773, Avg IoU: 0.7795, Avg Dice: 0.8625,Test IoU: 0.7527, Test Dice: 0.8397, Train Acc: 0.9898, Test Acc: 0.9897\n",
      "Epoch [7/200], Avg Loss: 0.02771, Avg IoU: 0.7983, Avg Dice: 0.8746,Test IoU: 0.7612, Test Dice: 0.8453, Train Acc: 0.9911, Test Acc: 0.9903\n",
      "Epoch [8/200], Avg Loss: 0.02770, Avg IoU: 0.8120, Avg Dice: 0.8835,Test IoU: 0.7593, Test Dice: 0.8442, Train Acc: 0.9920, Test Acc: 0.9902\n",
      "Epoch [9/200], Avg Loss: 0.02770, Avg IoU: 0.8202, Avg Dice: 0.8889,Test IoU: 0.7577, Test Dice: 0.8423, Train Acc: 0.9925, Test Acc: 0.9901\n",
      "Epoch [10/200], Avg Loss: 0.02769, Avg IoU: 0.8271, Avg Dice: 0.8928,Test IoU: 0.7723, Test Dice: 0.8524, Train Acc: 0.9929, Test Acc: 0.9911\n",
      "Epoch [11/200], Avg Loss: 0.02769, Avg IoU: 0.8299, Avg Dice: 0.8947,Test IoU: 0.7888, Test Dice: 0.8630, Train Acc: 0.9930, Test Acc: 0.9920\n",
      "Epoch [12/200], Avg Loss: 0.02768, Avg IoU: 0.8378, Avg Dice: 0.8998,Test IoU: 0.7772, Test Dice: 0.8557, Train Acc: 0.9935, Test Acc: 0.9913\n",
      "Epoch [13/200], Avg Loss: 0.02768, Avg IoU: 0.8444, Avg Dice: 0.9035,Test IoU: 0.7703, Test Dice: 0.8515, Train Acc: 0.9939, Test Acc: 0.9909\n",
      "Epoch [14/200], Avg Loss: 0.02768, Avg IoU: 0.8487, Avg Dice: 0.9064,Test IoU: 0.7892, Test Dice: 0.8632, Train Acc: 0.9941, Test Acc: 0.9921\n",
      "Epoch [15/200], Avg Loss: 0.02768, Avg IoU: 0.8545, Avg Dice: 0.9101,Test IoU: 0.7831, Test Dice: 0.8600, Train Acc: 0.9944, Test Acc: 0.9916\n",
      "Epoch [16/200], Avg Loss: 0.02768, Avg IoU: 0.8582, Avg Dice: 0.9122,Test IoU: 0.7876, Test Dice: 0.8624, Train Acc: 0.9946, Test Acc: 0.9919\n",
      "Epoch [17/200], Avg Loss: 0.02767, Avg IoU: 0.8599, Avg Dice: 0.9133,Test IoU: 0.7953, Test Dice: 0.8669, Train Acc: 0.9947, Test Acc: 0.9924\n",
      "Epoch [18/200], Avg Loss: 0.02767, Avg IoU: 0.8635, Avg Dice: 0.9155,Test IoU: 0.8022, Test Dice: 0.8720, Train Acc: 0.9949, Test Acc: 0.9928\n",
      "Epoch [19/200], Avg Loss: 0.02767, Avg IoU: 0.8677, Avg Dice: 0.9181,Test IoU: 0.8094, Test Dice: 0.8761, Train Acc: 0.9951, Test Acc: 0.9932\n",
      "Epoch [20/200], Avg Loss: 0.02767, Avg IoU: 0.8693, Avg Dice: 0.9189,Test IoU: 0.8010, Test Dice: 0.8711, Train Acc: 0.9952, Test Acc: 0.9927\n",
      "Epoch [21/200], Avg Loss: 0.02767, Avg IoU: 0.8709, Avg Dice: 0.9199,Test IoU: 0.8127, Test Dice: 0.8786, Train Acc: 0.9953, Test Acc: 0.9934\n",
      "Epoch [22/200], Avg Loss: 0.02767, Avg IoU: 0.8703, Avg Dice: 0.9195,Test IoU: 0.8017, Test Dice: 0.8714, Train Acc: 0.9952, Test Acc: 0.9928\n",
      "Epoch [23/200], Avg Loss: 0.02767, Avg IoU: 0.8745, Avg Dice: 0.9221,Test IoU: 0.8165, Test Dice: 0.8809, Train Acc: 0.9955, Test Acc: 0.9936\n",
      "Epoch [24/200], Avg Loss: 0.02767, Avg IoU: 0.8744, Avg Dice: 0.9220,Test IoU: 0.7994, Test Dice: 0.8698, Train Acc: 0.9954, Test Acc: 0.9926\n",
      "Epoch [25/200], Avg Loss: 0.02767, Avg IoU: 0.8749, Avg Dice: 0.9224,Test IoU: 0.8054, Test Dice: 0.8737, Train Acc: 0.9955, Test Acc: 0.9929\n",
      "Epoch [26/200], Avg Loss: 0.02767, Avg IoU: 0.8814, Avg Dice: 0.9262,Test IoU: 0.8068, Test Dice: 0.8747, Train Acc: 0.9958, Test Acc: 0.9930\n",
      "Epoch [27/200], Avg Loss: 0.02767, Avg IoU: 0.8807, Avg Dice: 0.9257,Test IoU: 0.8172, Test Dice: 0.8812, Train Acc: 0.9957, Test Acc: 0.9936\n",
      "Epoch [28/200], Avg Loss: 0.02767, Avg IoU: 0.8843, Avg Dice: 0.9278,Test IoU: 0.8220, Test Dice: 0.8840, Train Acc: 0.9959, Test Acc: 0.9939\n",
      "Epoch [29/200], Avg Loss: 0.02767, Avg IoU: 0.8876, Avg Dice: 0.9297,Test IoU: 0.8108, Test Dice: 0.8773, Train Acc: 0.9961, Test Acc: 0.9933\n",
      "Epoch [30/200], Avg Loss: 0.02767, Avg IoU: 0.8904, Avg Dice: 0.9314,Test IoU: 0.8036, Test Dice: 0.8728, Train Acc: 0.9962, Test Acc: 0.9929\n",
      "Epoch [31/200], Avg Loss: 0.02767, Avg IoU: 0.8911, Avg Dice: 0.9319,Test IoU: 0.8123, Test Dice: 0.8779, Train Acc: 0.9962, Test Acc: 0.9934\n",
      "Epoch [32/200], Avg Loss: 0.02767, Avg IoU: 0.8925, Avg Dice: 0.9326,Test IoU: 0.8243, Test Dice: 0.8858, Train Acc: 0.9963, Test Acc: 0.9940\n",
      "Epoch [33/200], Avg Loss: 0.02767, Avg IoU: 0.8875, Avg Dice: 0.9294,Test IoU: 0.7967, Test Dice: 0.8683, Train Acc: 0.9961, Test Acc: 0.9925\n",
      "Epoch [34/200], Avg Loss: 0.02766, Avg IoU: 0.8890, Avg Dice: 0.9306,Test IoU: 0.8089, Test Dice: 0.8760, Train Acc: 0.9962, Test Acc: 0.9932\n",
      "Epoch [35/200], Avg Loss: 0.02766, Avg IoU: 0.8935, Avg Dice: 0.9330,Test IoU: 0.8146, Test Dice: 0.8797, Train Acc: 0.9964, Test Acc: 0.9935\n",
      "Epoch [36/200], Avg Loss: 0.02766, Avg IoU: 0.8958, Avg Dice: 0.9345,Test IoU: 0.8230, Test Dice: 0.8852, Train Acc: 0.9965, Test Acc: 0.9940\n",
      "Epoch [37/200], Avg Loss: 0.02766, Avg IoU: 0.8981, Avg Dice: 0.9358,Test IoU: 0.8163, Test Dice: 0.8807, Train Acc: 0.9966, Test Acc: 0.9936\n",
      "Epoch [38/200], Avg Loss: 0.02766, Avg IoU: 0.8997, Avg Dice: 0.9368,Test IoU: 0.8231, Test Dice: 0.8848, Train Acc: 0.9967, Test Acc: 0.9939\n",
      "Epoch [39/200], Avg Loss: 0.02766, Avg IoU: 0.9014, Avg Dice: 0.9378,Test IoU: 0.8264, Test Dice: 0.8869, Train Acc: 0.9968, Test Acc: 0.9941\n",
      "Epoch [40/200], Avg Loss: 0.02766, Avg IoU: 0.9031, Avg Dice: 0.9389,Test IoU: 0.8263, Test Dice: 0.8867, Train Acc: 0.9968, Test Acc: 0.9941\n",
      "Epoch [41/200], Avg Loss: 0.02766, Avg IoU: 0.8991, Avg Dice: 0.9364,Test IoU: 0.8173, Test Dice: 0.8813, Train Acc: 0.9966, Test Acc: 0.9936\n",
      "Epoch [42/200], Avg Loss: 0.02766, Avg IoU: 0.8993, Avg Dice: 0.9366,Test IoU: 0.8133, Test Dice: 0.8785, Train Acc: 0.9967, Test Acc: 0.9934\n",
      "Epoch [43/200], Avg Loss: 0.02766, Avg IoU: 0.9027, Avg Dice: 0.9386,Test IoU: 0.8243, Test Dice: 0.8857, Train Acc: 0.9968, Test Acc: 0.9940\n",
      "Epoch [44/200], Avg Loss: 0.02766, Avg IoU: 0.9048, Avg Dice: 0.9399,Test IoU: 0.8268, Test Dice: 0.8871, Train Acc: 0.9969, Test Acc: 0.9941\n",
      "Epoch [45/200], Avg Loss: 0.02766, Avg IoU: 0.9074, Avg Dice: 0.9411,Test IoU: 0.8214, Test Dice: 0.8839, Train Acc: 0.9971, Test Acc: 0.9939\n",
      "Epoch [46/200], Avg Loss: 0.02766, Avg IoU: 0.9103, Avg Dice: 0.9430,Test IoU: 0.8274, Test Dice: 0.8877, Train Acc: 0.9972, Test Acc: 0.9942\n",
      "Epoch [47/200], Avg Loss: 0.02766, Avg IoU: 0.9097, Avg Dice: 0.9426,Test IoU: 0.8276, Test Dice: 0.8874, Train Acc: 0.9971, Test Acc: 0.9942\n",
      "Epoch [48/200], Avg Loss: 0.02766, Avg IoU: 0.9114, Avg Dice: 0.9437,Test IoU: 0.8296, Test Dice: 0.8890, Train Acc: 0.9972, Test Acc: 0.9943\n",
      "Epoch [49/200], Avg Loss: 0.02766, Avg IoU: 0.9129, Avg Dice: 0.9445,Test IoU: 0.8365, Test Dice: 0.8932, Train Acc: 0.9973, Test Acc: 0.9946\n",
      "Epoch [50/200], Avg Loss: 0.02766, Avg IoU: 0.9118, Avg Dice: 0.9438,Test IoU: 0.8154, Test Dice: 0.8804, Train Acc: 0.9972, Test Acc: 0.9935\n",
      "Epoch [51/200], Avg Loss: 0.02766, Avg IoU: 0.9116, Avg Dice: 0.9438,Test IoU: 0.8194, Test Dice: 0.8826, Train Acc: 0.9972, Test Acc: 0.9938\n",
      "Epoch [52/200], Avg Loss: 0.02766, Avg IoU: 0.9122, Avg Dice: 0.9440,Test IoU: 0.8276, Test Dice: 0.8877, Train Acc: 0.9973, Test Acc: 0.9942\n",
      "Epoch [53/200], Avg Loss: 0.02766, Avg IoU: 0.9122, Avg Dice: 0.9439,Test IoU: 0.8277, Test Dice: 0.8878, Train Acc: 0.9973, Test Acc: 0.9942\n",
      "Epoch [54/200], Avg Loss: 0.02766, Avg IoU: 0.9138, Avg Dice: 0.9449,Test IoU: 0.8258, Test Dice: 0.8866, Train Acc: 0.9973, Test Acc: 0.9941\n",
      "Epoch [55/200], Avg Loss: 0.02766, Avg IoU: 0.9167, Avg Dice: 0.9466,Test IoU: 0.8272, Test Dice: 0.8876, Train Acc: 0.9975, Test Acc: 0.9941\n",
      "Epoch [56/200], Avg Loss: 0.02766, Avg IoU: 0.9147, Avg Dice: 0.9456,Test IoU: 0.8260, Test Dice: 0.8867, Train Acc: 0.9974, Test Acc: 0.9941\n",
      "Epoch [57/200], Avg Loss: 0.02766, Avg IoU: 0.9163, Avg Dice: 0.9464,Test IoU: 0.8295, Test Dice: 0.8888, Train Acc: 0.9974, Test Acc: 0.9942\n",
      "Epoch [58/200], Avg Loss: 0.02766, Avg IoU: 0.9159, Avg Dice: 0.9462,Test IoU: 0.8313, Test Dice: 0.8898, Train Acc: 0.9974, Test Acc: 0.9944\n",
      "Epoch [59/200], Avg Loss: 0.02766, Avg IoU: 0.9162, Avg Dice: 0.9462,Test IoU: 0.8328, Test Dice: 0.8907, Train Acc: 0.9974, Test Acc: 0.9944\n",
      "Epoch [60/200], Avg Loss: 0.02766, Avg IoU: 0.9191, Avg Dice: 0.9479,Test IoU: 0.8309, Test Dice: 0.8897, Train Acc: 0.9976, Test Acc: 0.9944\n",
      "Epoch [61/200], Avg Loss: 0.02766, Avg IoU: 0.9169, Avg Dice: 0.9468,Test IoU: 0.8319, Test Dice: 0.8902, Train Acc: 0.9975, Test Acc: 0.9944\n",
      "Epoch [62/200], Avg Loss: 0.02766, Avg IoU: 0.9176, Avg Dice: 0.9471,Test IoU: 0.8290, Test Dice: 0.8886, Train Acc: 0.9975, Test Acc: 0.9943\n",
      "Epoch [63/200], Avg Loss: 0.02766, Avg IoU: 0.9196, Avg Dice: 0.9483,Test IoU: 0.8325, Test Dice: 0.8906, Train Acc: 0.9976, Test Acc: 0.9945\n",
      "Epoch [64/200], Avg Loss: 0.02766, Avg IoU: 0.9219, Avg Dice: 0.9496,Test IoU: 0.8335, Test Dice: 0.8913, Train Acc: 0.9977, Test Acc: 0.9945\n",
      "Epoch [65/200], Avg Loss: 0.02766, Avg IoU: 0.9216, Avg Dice: 0.9493,Test IoU: 0.8292, Test Dice: 0.8887, Train Acc: 0.9977, Test Acc: 0.9943\n",
      "Epoch [66/200], Avg Loss: 0.02766, Avg IoU: 0.9232, Avg Dice: 0.9503,Test IoU: 0.8354, Test Dice: 0.8925, Train Acc: 0.9977, Test Acc: 0.9946\n",
      "Epoch [67/200], Avg Loss: 0.02766, Avg IoU: 0.9232, Avg Dice: 0.9504,Test IoU: 0.8305, Test Dice: 0.8896, Train Acc: 0.9977, Test Acc: 0.9943\n",
      "Epoch [68/200], Avg Loss: 0.02766, Avg IoU: 0.9229, Avg Dice: 0.9501,Test IoU: 0.8318, Test Dice: 0.8903, Train Acc: 0.9977, Test Acc: 0.9944\n",
      "Epoch [69/200], Avg Loss: 0.02766, Avg IoU: 0.9236, Avg Dice: 0.9504,Test IoU: 0.8300, Test Dice: 0.8890, Train Acc: 0.9978, Test Acc: 0.9943\n",
      "Epoch [70/200], Avg Loss: 0.02766, Avg IoU: 0.9247, Avg Dice: 0.9511,Test IoU: 0.8349, Test Dice: 0.8923, Train Acc: 0.9978, Test Acc: 0.9946\n",
      "Epoch [71/200], Avg Loss: 0.02766, Avg IoU: 0.9253, Avg Dice: 0.9514,Test IoU: 0.8319, Test Dice: 0.8902, Train Acc: 0.9978, Test Acc: 0.9944\n",
      "Epoch [72/200], Avg Loss: 0.02766, Avg IoU: 0.9250, Avg Dice: 0.9511,Test IoU: 0.8278, Test Dice: 0.8877, Train Acc: 0.9978, Test Acc: 0.9942\n",
      "Epoch [73/200], Avg Loss: 0.02766, Avg IoU: 0.9251, Avg Dice: 0.9513,Test IoU: 0.8358, Test Dice: 0.8925, Train Acc: 0.9978, Test Acc: 0.9946\n",
      "Epoch [74/200], Avg Loss: 0.02766, Avg IoU: 0.9257, Avg Dice: 0.9516,Test IoU: 0.8256, Test Dice: 0.8863, Train Acc: 0.9979, Test Acc: 0.9941\n",
      "Epoch [75/200], Avg Loss: 0.02766, Avg IoU: 0.9252, Avg Dice: 0.9514,Test IoU: 0.8287, Test Dice: 0.8884, Train Acc: 0.9978, Test Acc: 0.9942\n",
      "Epoch [76/200], Avg Loss: 0.02766, Avg IoU: 0.9244, Avg Dice: 0.9509,Test IoU: 0.8360, Test Dice: 0.8927, Train Acc: 0.9978, Test Acc: 0.9946\n",
      "Epoch [77/200], Avg Loss: 0.02766, Avg IoU: 0.9270, Avg Dice: 0.9522,Test IoU: 0.8322, Test Dice: 0.8904, Train Acc: 0.9979, Test Acc: 0.9944\n",
      "Epoch [78/200], Avg Loss: 0.02766, Avg IoU: 0.9266, Avg Dice: 0.9520,Test IoU: 0.8330, Test Dice: 0.8910, Train Acc: 0.9979, Test Acc: 0.9944\n",
      "Epoch [79/200], Avg Loss: 0.02766, Avg IoU: 0.9279, Avg Dice: 0.9528,Test IoU: 0.8363, Test Dice: 0.8931, Train Acc: 0.9979, Test Acc: 0.9946\n",
      "Epoch [80/200], Avg Loss: 0.02766, Avg IoU: 0.9291, Avg Dice: 0.9535,Test IoU: 0.8387, Test Dice: 0.8948, Train Acc: 0.9980, Test Acc: 0.9947\n",
      "Epoch [81/200], Avg Loss: 0.02766, Avg IoU: 0.9282, Avg Dice: 0.9529,Test IoU: 0.8324, Test Dice: 0.8905, Train Acc: 0.9980, Test Acc: 0.9944\n",
      "Epoch [82/200], Avg Loss: 0.02766, Avg IoU: 0.9291, Avg Dice: 0.9534,Test IoU: 0.8355, Test Dice: 0.8923, Train Acc: 0.9980, Test Acc: 0.9946\n",
      "Epoch [83/200], Avg Loss: 0.02766, Avg IoU: 0.9290, Avg Dice: 0.9535,Test IoU: 0.8369, Test Dice: 0.8936, Train Acc: 0.9980, Test Acc: 0.9946\n",
      "Epoch [84/200], Avg Loss: 0.02766, Avg IoU: 0.9288, Avg Dice: 0.9533,Test IoU: 0.8354, Test Dice: 0.8925, Train Acc: 0.9980, Test Acc: 0.9946\n",
      "Epoch [85/200], Avg Loss: 0.02766, Avg IoU: 0.9295, Avg Dice: 0.9537,Test IoU: 0.8376, Test Dice: 0.8938, Train Acc: 0.9980, Test Acc: 0.9947\n",
      "Epoch [86/200], Avg Loss: 0.02766, Avg IoU: 0.9305, Avg Dice: 0.9543,Test IoU: 0.8371, Test Dice: 0.8937, Train Acc: 0.9981, Test Acc: 0.9947\n",
      "Epoch [87/200], Avg Loss: 0.02766, Avg IoU: 0.9309, Avg Dice: 0.9546,Test IoU: 0.8354, Test Dice: 0.8924, Train Acc: 0.9981, Test Acc: 0.9946\n",
      "Epoch [88/200], Avg Loss: 0.02766, Avg IoU: 0.9320, Avg Dice: 0.9550,Test IoU: 0.8313, Test Dice: 0.8898, Train Acc: 0.9981, Test Acc: 0.9943\n",
      "Epoch [89/200], Avg Loss: 0.02766, Avg IoU: 0.9321, Avg Dice: 0.9552,Test IoU: 0.8395, Test Dice: 0.8950, Train Acc: 0.9981, Test Acc: 0.9948\n",
      "Epoch [90/200], Avg Loss: 0.02766, Avg IoU: 0.9330, Avg Dice: 0.9558,Test IoU: 0.8409, Test Dice: 0.8958, Train Acc: 0.9982, Test Acc: 0.9948\n",
      "Epoch [91/200], Avg Loss: 0.02766, Avg IoU: 0.9326, Avg Dice: 0.9554,Test IoU: 0.8353, Test Dice: 0.8928, Train Acc: 0.9981, Test Acc: 0.9945\n",
      "Epoch [92/200], Avg Loss: 0.02766, Avg IoU: 0.9315, Avg Dice: 0.9547,Test IoU: 0.8369, Test Dice: 0.8935, Train Acc: 0.9981, Test Acc: 0.9946\n",
      "Epoch [93/200], Avg Loss: 0.02766, Avg IoU: 0.9326, Avg Dice: 0.9554,Test IoU: 0.8383, Test Dice: 0.8942, Train Acc: 0.9981, Test Acc: 0.9947\n",
      "Epoch [94/200], Avg Loss: 0.02766, Avg IoU: 0.9329, Avg Dice: 0.9556,Test IoU: 0.8362, Test Dice: 0.8932, Train Acc: 0.9981, Test Acc: 0.9946\n",
      "Epoch [95/200], Avg Loss: 0.02766, Avg IoU: 0.9327, Avg Dice: 0.9555,Test IoU: 0.8377, Test Dice: 0.8939, Train Acc: 0.9981, Test Acc: 0.9947\n",
      "Epoch [96/200], Avg Loss: 0.02766, Avg IoU: 0.9346, Avg Dice: 0.9566,Test IoU: 0.8418, Test Dice: 0.8966, Train Acc: 0.9982, Test Acc: 0.9949\n",
      "Epoch [97/200], Avg Loss: 0.02766, Avg IoU: 0.9348, Avg Dice: 0.9567,Test IoU: 0.8380, Test Dice: 0.8942, Train Acc: 0.9982, Test Acc: 0.9947\n",
      "Epoch [98/200], Avg Loss: 0.02766, Avg IoU: 0.9341, Avg Dice: 0.9563,Test IoU: 0.8373, Test Dice: 0.8939, Train Acc: 0.9982, Test Acc: 0.9946\n",
      "Epoch [99/200], Avg Loss: 0.02766, Avg IoU: 0.9351, Avg Dice: 0.9569,Test IoU: 0.8414, Test Dice: 0.8964, Train Acc: 0.9982, Test Acc: 0.9948\n",
      "Epoch [100/200], Avg Loss: 0.02766, Avg IoU: 0.9346, Avg Dice: 0.9565,Test IoU: 0.8320, Test Dice: 0.8903, Train Acc: 0.9982, Test Acc: 0.9944\n",
      "Epoch [101/200], Avg Loss: 0.02766, Avg IoU: 0.9331, Avg Dice: 0.9557,Test IoU: 0.8327, Test Dice: 0.8908, Train Acc: 0.9982, Test Acc: 0.9944\n",
      "Epoch [102/200], Avg Loss: 0.02766, Avg IoU: 0.9378, Avg Dice: 0.9584,Test IoU: 0.8404, Test Dice: 0.8957, Train Acc: 0.9984, Test Acc: 0.9948\n",
      "Epoch [103/200], Avg Loss: 0.02766, Avg IoU: 0.9393, Avg Dice: 0.9592,Test IoU: 0.8401, Test Dice: 0.8953, Train Acc: 0.9984, Test Acc: 0.9948\n",
      "Epoch [104/200], Avg Loss: 0.02766, Avg IoU: 0.9390, Avg Dice: 0.9590,Test IoU: 0.8373, Test Dice: 0.8937, Train Acc: 0.9984, Test Acc: 0.9947\n",
      "Epoch [105/200], Avg Loss: 0.02766, Avg IoU: 0.9386, Avg Dice: 0.9587,Test IoU: 0.8402, Test Dice: 0.8956, Train Acc: 0.9984, Test Acc: 0.9948\n",
      "Epoch [106/200], Avg Loss: 0.02766, Avg IoU: 0.9391, Avg Dice: 0.9590,Test IoU: 0.8406, Test Dice: 0.8959, Train Acc: 0.9984, Test Acc: 0.9948\n",
      "Epoch [107/200], Avg Loss: 0.02766, Avg IoU: 0.9382, Avg Dice: 0.9584,Test IoU: 0.8438, Test Dice: 0.8977, Train Acc: 0.9984, Test Acc: 0.9950\n",
      "Epoch [108/200], Avg Loss: 0.02766, Avg IoU: 0.9399, Avg Dice: 0.9594,Test IoU: 0.8420, Test Dice: 0.8966, Train Acc: 0.9985, Test Acc: 0.9949\n",
      "Epoch [109/200], Avg Loss: 0.02766, Avg IoU: 0.9402, Avg Dice: 0.9595,Test IoU: 0.8425, Test Dice: 0.8970, Train Acc: 0.9985, Test Acc: 0.9949\n",
      "Epoch [110/200], Avg Loss: 0.02766, Avg IoU: 0.9409, Avg Dice: 0.9599,Test IoU: 0.8427, Test Dice: 0.8971, Train Acc: 0.9985, Test Acc: 0.9949\n",
      "Epoch [111/200], Avg Loss: 0.02766, Avg IoU: 0.9408, Avg Dice: 0.9599,Test IoU: 0.8412, Test Dice: 0.8964, Train Acc: 0.9985, Test Acc: 0.9948\n",
      "Epoch [112/200], Avg Loss: 0.02766, Avg IoU: 0.9394, Avg Dice: 0.9591,Test IoU: 0.8426, Test Dice: 0.8970, Train Acc: 0.9984, Test Acc: 0.9949\n",
      "Epoch [113/200], Avg Loss: 0.02766, Avg IoU: 0.9410, Avg Dice: 0.9600,Test IoU: 0.8437, Test Dice: 0.8975, Train Acc: 0.9985, Test Acc: 0.9950\n",
      "Epoch [114/200], Avg Loss: 0.02766, Avg IoU: 0.9426, Avg Dice: 0.9609,Test IoU: 0.8456, Test Dice: 0.8987, Train Acc: 0.9986, Test Acc: 0.9951\n",
      "Epoch [115/200], Avg Loss: 0.02766, Avg IoU: 0.9419, Avg Dice: 0.9606,Test IoU: 0.8365, Test Dice: 0.8932, Train Acc: 0.9985, Test Acc: 0.9946\n",
      "Epoch [116/200], Avg Loss: 0.02766, Avg IoU: 0.9407, Avg Dice: 0.9600,Test IoU: 0.8444, Test Dice: 0.8980, Train Acc: 0.9985, Test Acc: 0.9950\n",
      "Epoch [117/200], Avg Loss: 0.02766, Avg IoU: 0.9405, Avg Dice: 0.9597,Test IoU: 0.8416, Test Dice: 0.8967, Train Acc: 0.9985, Test Acc: 0.9948\n",
      "Epoch [118/200], Avg Loss: 0.02766, Avg IoU: 0.9414, Avg Dice: 0.9602,Test IoU: 0.8445, Test Dice: 0.8984, Train Acc: 0.9985, Test Acc: 0.9950\n",
      "Epoch [119/200], Avg Loss: 0.02766, Avg IoU: 0.9425, Avg Dice: 0.9609,Test IoU: 0.8429, Test Dice: 0.8973, Train Acc: 0.9985, Test Acc: 0.9949\n",
      "Epoch [120/200], Avg Loss: 0.02766, Avg IoU: 0.9429, Avg Dice: 0.9611,Test IoU: 0.8413, Test Dice: 0.8963, Train Acc: 0.9986, Test Acc: 0.9948\n",
      "Epoch [121/200], Avg Loss: 0.02766, Avg IoU: 0.9423, Avg Dice: 0.9608,Test IoU: 0.8388, Test Dice: 0.8949, Train Acc: 0.9985, Test Acc: 0.9947\n",
      "Epoch [122/200], Avg Loss: 0.02766, Avg IoU: 0.9421, Avg Dice: 0.9607,Test IoU: 0.8439, Test Dice: 0.8978, Train Acc: 0.9985, Test Acc: 0.9950\n",
      "Epoch [123/200], Avg Loss: 0.02766, Avg IoU: 0.9425, Avg Dice: 0.9608,Test IoU: 0.8430, Test Dice: 0.8972, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [124/200], Avg Loss: 0.02766, Avg IoU: 0.9429, Avg Dice: 0.9610,Test IoU: 0.8408, Test Dice: 0.8962, Train Acc: 0.9986, Test Acc: 0.9948\n",
      "Epoch [125/200], Avg Loss: 0.02766, Avg IoU: 0.9430, Avg Dice: 0.9611,Test IoU: 0.8427, Test Dice: 0.8974, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [126/200], Avg Loss: 0.02766, Avg IoU: 0.9432, Avg Dice: 0.9612,Test IoU: 0.8475, Test Dice: 0.9003, Train Acc: 0.9986, Test Acc: 0.9951\n",
      "Epoch [127/200], Avg Loss: 0.02766, Avg IoU: 0.9434, Avg Dice: 0.9614,Test IoU: 0.8374, Test Dice: 0.8937, Train Acc: 0.9986, Test Acc: 0.9946\n",
      "Epoch [128/200], Avg Loss: 0.02766, Avg IoU: 0.9431, Avg Dice: 0.9612,Test IoU: 0.8425, Test Dice: 0.8967, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [129/200], Avg Loss: 0.02766, Avg IoU: 0.9430, Avg Dice: 0.9612,Test IoU: 0.8438, Test Dice: 0.8978, Train Acc: 0.9986, Test Acc: 0.9950\n",
      "Epoch [130/200], Avg Loss: 0.02766, Avg IoU: 0.9427, Avg Dice: 0.9610,Test IoU: 0.8462, Test Dice: 0.8995, Train Acc: 0.9986, Test Acc: 0.9951\n",
      "Epoch [131/200], Avg Loss: 0.02766, Avg IoU: 0.9431, Avg Dice: 0.9612,Test IoU: 0.8428, Test Dice: 0.8973, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [132/200], Avg Loss: 0.02766, Avg IoU: 0.9441, Avg Dice: 0.9618,Test IoU: 0.8439, Test Dice: 0.8977, Train Acc: 0.9986, Test Acc: 0.9950\n",
      "Epoch [133/200], Avg Loss: 0.02766, Avg IoU: 0.9434, Avg Dice: 0.9613,Test IoU: 0.8421, Test Dice: 0.8965, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [134/200], Avg Loss: 0.02766, Avg IoU: 0.9436, Avg Dice: 0.9614,Test IoU: 0.8434, Test Dice: 0.8976, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [135/200], Avg Loss: 0.02766, Avg IoU: 0.9442, Avg Dice: 0.9617,Test IoU: 0.8421, Test Dice: 0.8968, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [136/200], Avg Loss: 0.02766, Avg IoU: 0.9443, Avg Dice: 0.9618,Test IoU: 0.8446, Test Dice: 0.8983, Train Acc: 0.9986, Test Acc: 0.9950\n",
      "Epoch [137/200], Avg Loss: 0.02766, Avg IoU: 0.9409, Avg Dice: 0.9598,Test IoU: 0.8453, Test Dice: 0.8987, Train Acc: 0.9985, Test Acc: 0.9950\n",
      "Epoch [138/200], Avg Loss: 0.02766, Avg IoU: 0.9430, Avg Dice: 0.9610,Test IoU: 0.8450, Test Dice: 0.8984, Train Acc: 0.9986, Test Acc: 0.9950\n",
      "Epoch [139/200], Avg Loss: 0.02766, Avg IoU: 0.9440, Avg Dice: 0.9617,Test IoU: 0.8395, Test Dice: 0.8949, Train Acc: 0.9986, Test Acc: 0.9948\n",
      "Epoch [140/200], Avg Loss: 0.02766, Avg IoU: 0.9442, Avg Dice: 0.9618,Test IoU: 0.8426, Test Dice: 0.8968, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [141/200], Avg Loss: 0.02766, Avg IoU: 0.9448, Avg Dice: 0.9621,Test IoU: 0.8458, Test Dice: 0.8988, Train Acc: 0.9987, Test Acc: 0.9951\n",
      "Epoch [142/200], Avg Loss: 0.02766, Avg IoU: 0.9452, Avg Dice: 0.9624,Test IoU: 0.8375, Test Dice: 0.8937, Train Acc: 0.9987, Test Acc: 0.9946\n",
      "Epoch [143/200], Avg Loss: 0.02766, Avg IoU: 0.9449, Avg Dice: 0.9622,Test IoU: 0.8422, Test Dice: 0.8967, Train Acc: 0.9986, Test Acc: 0.9949\n",
      "Epoch [144/200], Avg Loss: 0.02766, Avg IoU: 0.9454, Avg Dice: 0.9624,Test IoU: 0.8458, Test Dice: 0.8989, Train Acc: 0.9987, Test Acc: 0.9951\n",
      "Epoch [145/200], Avg Loss: 0.02766, Avg IoU: 0.9467, Avg Dice: 0.9632,Test IoU: 0.8406, Test Dice: 0.8957, Train Acc: 0.9987, Test Acc: 0.9948\n",
      "Epoch [146/200], Avg Loss: 0.02766, Avg IoU: 0.9449, Avg Dice: 0.9621,Test IoU: 0.8459, Test Dice: 0.8988, Train Acc: 0.9987, Test Acc: 0.9951\n",
      "Epoch [147/200], Avg Loss: 0.02766, Avg IoU: 0.9455, Avg Dice: 0.9625,Test IoU: 0.8443, Test Dice: 0.8978, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [148/200], Avg Loss: 0.02766, Avg IoU: 0.9458, Avg Dice: 0.9626,Test IoU: 0.8427, Test Dice: 0.8970, Train Acc: 0.9987, Test Acc: 0.9949\n",
      "Epoch [149/200], Avg Loss: 0.02766, Avg IoU: 0.9454, Avg Dice: 0.9622,Test IoU: 0.8407, Test Dice: 0.8957, Train Acc: 0.9987, Test Acc: 0.9948\n",
      "Epoch [150/200], Avg Loss: 0.02766, Avg IoU: 0.9453, Avg Dice: 0.9623,Test IoU: 0.8419, Test Dice: 0.8964, Train Acc: 0.9987, Test Acc: 0.9949\n",
      "Epoch [151/200], Avg Loss: 0.02766, Avg IoU: 0.9458, Avg Dice: 0.9626,Test IoU: 0.8444, Test Dice: 0.8981, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [152/200], Avg Loss: 0.02766, Avg IoU: 0.9458, Avg Dice: 0.9626,Test IoU: 0.8452, Test Dice: 0.8982, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [153/200], Avg Loss: 0.02766, Avg IoU: 0.9450, Avg Dice: 0.9621,Test IoU: 0.8437, Test Dice: 0.8977, Train Acc: 0.9987, Test Acc: 0.9949\n",
      "Epoch [154/200], Avg Loss: 0.02766, Avg IoU: 0.9461, Avg Dice: 0.9628,Test IoU: 0.8449, Test Dice: 0.8982, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [155/200], Avg Loss: 0.02766, Avg IoU: 0.9466, Avg Dice: 0.9630,Test IoU: 0.8383, Test Dice: 0.8943, Train Acc: 0.9987, Test Acc: 0.9947\n",
      "Epoch [156/200], Avg Loss: 0.02766, Avg IoU: 0.9459, Avg Dice: 0.9627,Test IoU: 0.8453, Test Dice: 0.8985, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [157/200], Avg Loss: 0.02766, Avg IoU: 0.9465, Avg Dice: 0.9630,Test IoU: 0.8429, Test Dice: 0.8969, Train Acc: 0.9987, Test Acc: 0.9949\n",
      "Epoch [158/200], Avg Loss: 0.02766, Avg IoU: 0.9469, Avg Dice: 0.9632,Test IoU: 0.8414, Test Dice: 0.8961, Train Acc: 0.9987, Test Acc: 0.9948\n",
      "Epoch [159/200], Avg Loss: 0.02766, Avg IoU: 0.9471, Avg Dice: 0.9634,Test IoU: 0.8431, Test Dice: 0.8971, Train Acc: 0.9987, Test Acc: 0.9949\n",
      "Epoch [160/200], Avg Loss: 0.02766, Avg IoU: 0.9466, Avg Dice: 0.9631,Test IoU: 0.8413, Test Dice: 0.8963, Train Acc: 0.9987, Test Acc: 0.9948\n",
      "Epoch [161/200], Avg Loss: 0.02766, Avg IoU: 0.9460, Avg Dice: 0.9627,Test IoU: 0.8462, Test Dice: 0.8990, Train Acc: 0.9987, Test Acc: 0.9951\n",
      "Epoch [162/200], Avg Loss: 0.02766, Avg IoU: 0.9468, Avg Dice: 0.9632,Test IoU: 0.8444, Test Dice: 0.8979, Train Acc: 0.9987, Test Acc: 0.9950\n",
      "Epoch [163/200], Avg Loss: 0.02766, Avg IoU: 0.9480, Avg Dice: 0.9638,Test IoU: 0.8485, Test Dice: 0.9006, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [164/200], Avg Loss: 0.02766, Avg IoU: 0.9479, Avg Dice: 0.9638,Test IoU: 0.8490, Test Dice: 0.9008, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [165/200], Avg Loss: 0.02766, Avg IoU: 0.9484, Avg Dice: 0.9640,Test IoU: 0.8477, Test Dice: 0.8999, Train Acc: 0.9988, Test Acc: 0.9951\n",
      "Epoch [166/200], Avg Loss: 0.02766, Avg IoU: 0.9481, Avg Dice: 0.9639,Test IoU: 0.8456, Test Dice: 0.8985, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [167/200], Avg Loss: 0.02766, Avg IoU: 0.9483, Avg Dice: 0.9640,Test IoU: 0.8486, Test Dice: 0.9006, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [168/200], Avg Loss: 0.02766, Avg IoU: 0.9489, Avg Dice: 0.9643,Test IoU: 0.8429, Test Dice: 0.8970, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [169/200], Avg Loss: 0.02766, Avg IoU: 0.9483, Avg Dice: 0.9640,Test IoU: 0.8430, Test Dice: 0.8970, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [170/200], Avg Loss: 0.02766, Avg IoU: 0.9485, Avg Dice: 0.9641,Test IoU: 0.8439, Test Dice: 0.8976, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [171/200], Avg Loss: 0.02766, Avg IoU: 0.9483, Avg Dice: 0.9640,Test IoU: 0.8430, Test Dice: 0.8969, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [172/200], Avg Loss: 0.02766, Avg IoU: 0.9485, Avg Dice: 0.9641,Test IoU: 0.8432, Test Dice: 0.8974, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [173/200], Avg Loss: 0.02766, Avg IoU: 0.9480, Avg Dice: 0.9638,Test IoU: 0.8485, Test Dice: 0.9004, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [174/200], Avg Loss: 0.02766, Avg IoU: 0.9485, Avg Dice: 0.9641,Test IoU: 0.8425, Test Dice: 0.8967, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [175/200], Avg Loss: 0.02766, Avg IoU: 0.9484, Avg Dice: 0.9640,Test IoU: 0.8407, Test Dice: 0.8955, Train Acc: 0.9988, Test Acc: 0.9948\n",
      "Epoch [176/200], Avg Loss: 0.02766, Avg IoU: 0.9493, Avg Dice: 0.9646,Test IoU: 0.8495, Test Dice: 0.9010, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [177/200], Avg Loss: 0.02766, Avg IoU: 0.9494, Avg Dice: 0.9646,Test IoU: 0.8453, Test Dice: 0.8984, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [178/200], Avg Loss: 0.02766, Avg IoU: 0.9498, Avg Dice: 0.9649,Test IoU: 0.8444, Test Dice: 0.8980, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [179/200], Avg Loss: 0.02766, Avg IoU: 0.9486, Avg Dice: 0.9641,Test IoU: 0.8467, Test Dice: 0.8994, Train Acc: 0.9988, Test Acc: 0.9951\n",
      "Epoch [180/200], Avg Loss: 0.02766, Avg IoU: 0.9482, Avg Dice: 0.9640,Test IoU: 0.8409, Test Dice: 0.8957, Train Acc: 0.9988, Test Acc: 0.9948\n",
      "Epoch [181/200], Avg Loss: 0.02766, Avg IoU: 0.9487, Avg Dice: 0.9642,Test IoU: 0.8434, Test Dice: 0.8972, Train Acc: 0.9988, Test Acc: 0.9949\n",
      "Epoch [182/200], Avg Loss: 0.02766, Avg IoU: 0.9483, Avg Dice: 0.9640,Test IoU: 0.8461, Test Dice: 0.8990, Train Acc: 0.9988, Test Acc: 0.9951\n",
      "Epoch [183/200], Avg Loss: 0.02766, Avg IoU: 0.9480, Avg Dice: 0.9639,Test IoU: 0.8411, Test Dice: 0.8961, Train Acc: 0.9988, Test Acc: 0.9948\n",
      "Epoch [184/200], Avg Loss: 0.02766, Avg IoU: 0.9479, Avg Dice: 0.9638,Test IoU: 0.8441, Test Dice: 0.8979, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [185/200], Avg Loss: 0.02766, Avg IoU: 0.9493, Avg Dice: 0.9646,Test IoU: 0.8480, Test Dice: 0.9001, Train Acc: 0.9988, Test Acc: 0.9952\n",
      "Epoch [186/200], Avg Loss: 0.02766, Avg IoU: 0.9497, Avg Dice: 0.9648,Test IoU: 0.8471, Test Dice: 0.8999, Train Acc: 0.9988, Test Acc: 0.9951\n",
      "Epoch [187/200], Avg Loss: 0.02766, Avg IoU: 0.9490, Avg Dice: 0.9644,Test IoU: 0.8452, Test Dice: 0.8984, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [188/200], Avg Loss: 0.02766, Avg IoU: 0.9491, Avg Dice: 0.9644,Test IoU: 0.8477, Test Dice: 0.8997, Train Acc: 0.9988, Test Acc: 0.9951\n",
      "Epoch [189/200], Avg Loss: 0.02766, Avg IoU: 0.9495, Avg Dice: 0.9646,Test IoU: 0.8446, Test Dice: 0.8980, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [190/200], Avg Loss: 0.02766, Avg IoU: 0.9490, Avg Dice: 0.9644,Test IoU: 0.8450, Test Dice: 0.8983, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [191/200], Avg Loss: 0.02766, Avg IoU: 0.9496, Avg Dice: 0.9647,Test IoU: 0.8436, Test Dice: 0.8974, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [192/200], Avg Loss: 0.02766, Avg IoU: 0.9494, Avg Dice: 0.9646,Test IoU: 0.8452, Test Dice: 0.8984, Train Acc: 0.9988, Test Acc: 0.9950\n",
      "Epoch [193/200], Avg Loss: 0.02766, Avg IoU: 0.9500, Avg Dice: 0.9649,Test IoU: 0.8456, Test Dice: 0.8986, Train Acc: 0.9989, Test Acc: 0.9950\n",
      "Epoch [194/200], Avg Loss: 0.02766, Avg IoU: 0.9499, Avg Dice: 0.9649,Test IoU: 0.8452, Test Dice: 0.8985, Train Acc: 0.9989, Test Acc: 0.9950\n",
      "Epoch [195/200], Avg Loss: 0.02766, Avg IoU: 0.9505, Avg Dice: 0.9652,Test IoU: 0.8465, Test Dice: 0.8994, Train Acc: 0.9989, Test Acc: 0.9951\n",
      "Epoch [196/200], Avg Loss: 0.02766, Avg IoU: 0.9506, Avg Dice: 0.9652,Test IoU: 0.8493, Test Dice: 0.9012, Train Acc: 0.9989, Test Acc: 0.9952\n",
      "Epoch [197/200], Avg Loss: 0.02766, Avg IoU: 0.9500, Avg Dice: 0.9648,Test IoU: 0.8482, Test Dice: 0.9005, Train Acc: 0.9989, Test Acc: 0.9952\n",
      "Epoch [198/200], Avg Loss: 0.02766, Avg IoU: 0.9500, Avg Dice: 0.9647,Test IoU: 0.8436, Test Dice: 0.8975, Train Acc: 0.9989, Test Acc: 0.9949\n",
      "Epoch [199/200], Avg Loss: 0.02766, Avg IoU: 0.9509, Avg Dice: 0.9653,Test IoU: 0.8472, Test Dice: 0.8998, Train Acc: 0.9989, Test Acc: 0.9951\n",
      "Epoch [200/200], Avg Loss: 0.02766, Avg IoU: 0.9507, Avg Dice: 0.9652,Test IoU: 0.8440, Test Dice: 0.8978, Train Acc: 0.9989, Test Acc: 0.9950\n",
      "-----------------------------------------\n",
      "Best Loss: 0.0277, Best Train IoU: 0.9509, Best Train Dice: 0.9653, Best Test IoU: 0.8495, Best Test Dice: 0.9012, Best Train Acc: 0.9989, Best Test Acc: 0.9952\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x800 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# -----------------------------\n",
    "# UNet++ Train and Test Set\n",
    "# -----------------------------\n",
    "\n",
    "# empty the cache for model and training process\n",
    "torch.cuda.empty_cache()\n",
    "\n",
    "# set the hyper parameters and the paths 设置超参数和路径\n",
    "data_path = \"./dataset/Spineweb_dataset15/train/\"\n",
    "test_path = \"./dataset/Spineweb_dataset15/test/\"\n",
    "\n",
    "#hyper parameters set\n",
    "batch_size = 32\n",
    "num_epochs = 200\n",
    "learning_rate = 1e-04\n",
    "confidence = 0.7\n",
    "poster_threshold = 200\n",
    "in_channels = 1  # according to demand to adjust the number of channels 根据实际情况修改通道数\n",
    "out_channels = 1  # according to demand to adjust the number of channels 根据实际情况修改通道数\n",
    "\n",
    "# 创建数据集和数据加载器\n",
    "transform = transforms.Compose([\n",
    "    # transforms.ToPILImage(),  # 转换为 PIL 图像对象\n",
    "    transforms.Resize((128, 128)),  # 调整大小为 128x128\n",
    "    transforms.ToTensor()  # 转换为张量\n",
    "])\n",
    "\n",
    "dataset = SpineWeb15(data_path=data_path, transform1=transform, transform2=transform)\n",
    "dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=True)\n",
    "testset = SpineWeb15(data_path=test_path, transform1=transform, transform2=transform)\n",
    "test_loader = DataLoader(testset, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "# 创建模型和优化器\n",
    "model = smp.UnetPlusPlus(encoder_name='resnet34', encoder_depth=5, encoder_weights=\"imagenet\", decoder_use_batchnorm=True, decoder_channels=(256, 128, 64, 32, 16), decoder_attention_type=None, in_channels=1, classes=1, activation=None, aux_params=None)\n",
    "optimizer = optim.Adam(model.parameters(), lr=learning_rate)\n",
    "# criterion = nn.BCEWithLogitsLoss()  # 二分类任务可以使用BCEWithLogitsLoss\n",
    "# criterion = DiceLoss(mode=\"binary\")   #DiceLoss\n",
    "# criterion = FocalLoss(mode=\"binary\")\n",
    "criterion = smp.losses.TverskyLoss(mode=\"binary\")\n",
    "# criterion = smp.losses.LovaszLoss(mode=\"binary\")\n",
    "\n",
    "# 设置设备\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model = model.to(device)\n",
    "\n",
    "# check the parameters of the model could be trained 检查模型的参数是否设置为可训练\n",
    "# for name, param in model.named_parameters():\n",
    "#     print(f\"Parameter: {name}, Requires Grad: {param.requires_grad}\")\n",
    "\n",
    "best_loss_train = 100.0\n",
    "best_IoU_train = 0.0\n",
    "best_Dice_train = 0.0\n",
    "\n",
    "best_IoU_test = 0.0\n",
    "best_Dice_test = 0.0\n",
    "\n",
    "best_acc_train = 0.0\n",
    "best_acc_test = 0.0\n",
    "\n",
    "loss_train_list = []\n",
    "IoU_train_list = []\n",
    "Dice_train_list = []\n",
    "IoU_test_list = []\n",
    "Dice_test_list = []\n",
    "acc_train_list = []\n",
    "acc_test_list = []\n",
    "\n",
    "# 训练模型\n",
    "for epoch in range(num_epochs):\n",
    "    # torch.cuda.empty_cache()\n",
    "    model.train()\n",
    "    train_loss = 0.0\n",
    "    total_iou = 0.0\n",
    "    total_dice = 0.0\n",
    "    total_iou_test = 0.0\n",
    "    total_dice_test = 0.0\n",
    "    total_acc_train = 0.0\n",
    "    total_acc_test = 0.0\n",
    "    total_correct_pixels = 0\n",
    "    total_pixels = 0\n",
    "    total_correct_pixels_test = 0\n",
    "    total_pixels_test = 0\n",
    "\n",
    "    if epoch > 100:\n",
    "        confidence = 0.9\n",
    "\n",
    "    for images, labels in dataloader:\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "    \n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        outputs = model(images)\n",
    "        # squeeze the dimension that pytorch add for channels\n",
    "        outputs = outputs.squeeze(1)     # tensor size ([16,128,128])\n",
    "        labels = labels.squeeze(1)\n",
    "        labels = labels * 255\n",
    "        labels[labels <= poster_threshold] = 0\n",
    "        labels[labels > poster_threshold] = 1\n",
    "        # apply sigmoid on the outputs\n",
    "        outputs = torch.sigmoid(outputs)\n",
    "        # process the result, bigger than confidence is reliable\n",
    "        predicted = (outputs > confidence).float()     # tensor size ([16,128,128]), True->1.0 and False->0.0\n",
    "        intersection = torch.logical_and(predicted, labels).sum((1, 2))\n",
    "        union = torch.logical_or(predicted, labels).sum((1, 2))\n",
    "        iou = (intersection / (union + 1e-7))  # 每个样本的IoU\n",
    "        dice = (2 * intersection / (predicted.sum((1, 2)) + labels.sum((1, 2))+1e-7))  # 每个样本的Dice Coefficient\n",
    "        # 计算训练集的像素精度\n",
    "        batch_correct_pixels,bacth_total_pixels = calculate_pixel_accuracy(predicted, labels)\n",
    "\n",
    "        loss = criterion(outputs, labels)\n",
    "\n",
    "        train_loss += loss.item()\n",
    "        total_iou += torch.sum(iou).item()\n",
    "        total_dice += torch.sum(dice).item()\n",
    "        total_correct_pixels += batch_correct_pixels\n",
    "        total_pixels += bacth_total_pixels\n",
    "\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "    # train set results\n",
    "    avg_loss = train_loss / len(dataset)\n",
    "    avg_iou = total_iou / len(dataset)\n",
    "    avg_dice = total_dice / len(dataset)\n",
    "    avg_acc_train = total_correct_pixels / total_pixels\n",
    "\n",
    "    # test set results\n",
    "    with torch.no_grad():\n",
    "        for inputs, targets in test_loader:     \n",
    "            inputs = inputs.to(device)\n",
    "            targets = targets.to(device)\n",
    "            targets = targets * 255\n",
    "            targets[targets <= poster_threshold] = 0\n",
    "            targets[targets > poster_threshold] = 1\n",
    "            outputs = model(inputs)\n",
    "            # squeeze the dimension that pytorch add for channels\n",
    "            outputs = outputs.squeeze(1)     # tensor size ([16,128,128])\n",
    "            targets = targets.squeeze(1)\n",
    "            outputs = torch.sigmoid(outputs)\n",
    "            # process the result, bigger than 0.5 is reliable\n",
    "            predicted = (outputs > confidence).float()     # tensor size ([16,128,128])\n",
    "            intersection = torch.logical_and(predicted, targets).sum((1, 2))\n",
    "            union = torch.logical_or(predicted, targets).sum((1, 2))\n",
    "            iou = (intersection / (union + 1e-7))  # 每个样本的IoU\n",
    "            dice = (2 * intersection / (predicted.sum((1, 2)) + targets.sum((1, 2))+1e-7))  # 每个样本的Dice Coefficient\n",
    "\n",
    "\n",
    "            batch_correct_pixels,batch_total_pixels = calculate_pixel_accuracy(predicted, targets)\n",
    "            total_iou_test += torch.sum(iou).item()\n",
    "            total_dice_test += torch.sum(dice).item()\n",
    "            total_correct_pixels_test += batch_correct_pixels\n",
    "            total_pixels_test += batch_total_pixels\n",
    "\n",
    "    test_iou = total_iou_test / len(testset)\n",
    "    test_dice = total_dice_test / len(testset)\n",
    "    test_acc = total_correct_pixels_test / total_pixels_test\n",
    "\n",
    "    # get the best IoU and Dice\n",
    "    best_loss_train = best_loss_train if best_loss_train < avg_loss else avg_loss\n",
    "    best_IoU_train = best_IoU_train if best_IoU_train > avg_iou else avg_iou\n",
    "    best_Dice_train = best_Dice_train if best_Dice_train > avg_dice else avg_dice\n",
    "    best_IoU_test = best_IoU_test if best_IoU_test > test_iou else test_iou\n",
    "    # best_Dice_test = best_Dice_test if best_Dice_test > test_dice else test_dice\n",
    "\n",
    "    if(best_Dice_test <= test_dice):\n",
    "        best_Dice_test = test_dice\n",
    "        torch.save(model.state_dict(), f\"./model/UNet++_checkpoint.pth\")\n",
    "\n",
    "    # get the best accuracy of the trainloader and testloader\n",
    "    best_acc_train = best_acc_train if best_acc_train > avg_acc_train else avg_acc_train\n",
    "    best_acc_test = best_acc_test if best_acc_test > test_acc else test_acc\n",
    "\n",
    "    # store the data in each epoch\n",
    "    loss_train_list.append(avg_loss)\n",
    "    IoU_train_list.append(avg_iou)\n",
    "    Dice_train_list.append(avg_dice)\n",
    "    IoU_test_list.append(test_iou)\n",
    "    Dice_test_list.append(test_dice)\n",
    "    acc_train_list.append(avg_acc_train)\n",
    "    acc_test_list.append(test_acc)\n",
    "\n",
    "    print(f\"Epoch [{epoch+1}/{num_epochs}], Avg Loss: {avg_loss:.5f}, Avg IoU: {avg_iou:.4f}, Avg Dice: {avg_dice:.4f},Test IoU: {test_iou:.4f}, Test Dice: {test_dice:.4f}, Train Acc: {avg_acc_train:.4f}, Test Acc: {test_acc:.4f}\")\n",
    "    \n",
    "print(\"-----------------------------------------\")\n",
    "print(f\"Best Loss: {best_loss_train:.4f}, Best Train IoU: {best_IoU_train:.4f}, Best Train Dice: {best_Dice_train:.4f}, Best Test IoU: {best_IoU_test:.4f}, Best Test Dice: {best_Dice_test:.4f}, Best Train Acc: {best_acc_train:.4f}, Best Test Acc: {best_acc_test:.4f}\")\n",
    "# draw the plot for the loss and dice\n",
    "# 定义epochs的数量，用于x轴坐标\n",
    "epochs = len(loss_train_list)\n",
    "fit,(ax1,ax2) = plt.subplots(2,1,figsize=(8,8))\n",
    "# 绘制训练损失曲线\n",
    "ax1.plot(range(1, epochs + 1), loss_train_list, label='Train Loss')\n",
    "# 设置图例1\n",
    "ax1.legend()\n",
    "ax1.set_xlabel('Epochs')\n",
    "ax1.set_ylabel('Loss')\n",
    "ax1.set_title('Training Loss')\n",
    "ax1.grid(True)\n",
    "# 绘制IoU和Dice\n",
    "ax2.plot(range(1,epochs +1),IoU_train_list,label='Train IoU')\n",
    "ax2.plot(range(1,epochs +1),Dice_train_list,label='Train Dice')\n",
    "# 设置图例2\n",
    "ax2.legend()\n",
    "ax2.set_xlabel('Epochs')\n",
    "ax2.set_ylabel('IoU and Dice')\n",
    "ax2.set_title('Training IoU and Dice')\n",
    "ax2.grid(True)# show the plots\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "\n",
    "# empty the cache for model and training process\n",
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Test IoU: 0.8499, Best Test Dice: 0.9016 Best Test Acc: 0.9953\n"
     ]
    }
   ],
   "source": [
    "# Using the best model to evaluate on the test set\n",
    "\n",
    "model = smp.UnetPlusPlus(encoder_name='resnet34', encoder_depth=5, encoder_weights=\"imagenet\", decoder_use_batchnorm=True, decoder_channels=(256, 128, 64, 32, 16), decoder_attention_type=None, in_channels=1, classes=1, activation=None, aux_params=None)\n",
    "\n",
    "# 设置设备\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model = model.to(device)\n",
    "\n",
    "test_path = \"./dataset/Spineweb_dataset15/test/\"\n",
    "\n",
    "transform = transforms.Compose([\n",
    "    # transforms.ToPILImage(),  # 转换为 PIL 图像对象\n",
    "    transforms.Resize((128, 128)),  # 调整大小为 128x128\n",
    "    transforms.ToTensor()  # 转换为张量\n",
    "])\n",
    "\n",
    "# 加载已保存的模型参数\n",
    "checkpoint_path = \"./model/UNet++_checkpoint.pth\"  # 模型文件路径\n",
    "checkpoint = torch.load(checkpoint_path)\n",
    "model.load_state_dict(checkpoint)\n",
    "\n",
    "# 设置模型为推理模式\n",
    "model.eval()\n",
    "\n",
    "batch_size = 64\n",
    "\n",
    "confidence = 0.9\n",
    "\n",
    "testset = SpineWeb15(data_path=test_path, transform1=transform,transform2=transform)\n",
    "test_loader = DataLoader(testset, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "# data_iter = iter(test_loader)\n",
    "# images, labels = next(data_iter)\n",
    "# print(\"加载的样本数量:\", images.shape[0])\n",
    "\n",
    "total_iou_test = 0.0\n",
    "total_dice_test = 0.0\n",
    "total_acc_test = 0.0\n",
    "total_correct_pixels_test = 0\n",
    "total_pixels_test = 0\n",
    "\n",
    "# test set results\n",
    "with torch.no_grad():\n",
    "    for inputs, targets in test_loader:\n",
    "        inputs = inputs.to(device)\n",
    "        targets = targets.to(device)\n",
    "        targets = targets * 255\n",
    "        targets[targets <= poster_threshold] = 0\n",
    "        targets[targets > poster_threshold] = 1\n",
    "        outputs = model(inputs)\n",
    "        # squeeze the dimension that pytorch add for channels\n",
    "        outputs = outputs.squeeze(1)     # tensor size ([16,128,128])\n",
    "        targets = targets.squeeze(1)\n",
    "        outputs = torch.sigmoid(outputs)\n",
    "        # process the result, bigger than 0.5 is reliable\n",
    "        predicted = (outputs > confidence).float()     # tensor size ([16,128,128])\n",
    "        intersection = torch.logical_and(predicted, targets).sum((1, 2))\n",
    "        union = torch.logical_or(predicted, targets).sum((1, 2))\n",
    "        iou = (intersection / (union + 1e-7))  # 平均每个样本的IoU\n",
    "        dice = (2 * intersection / (predicted.sum((1, 2)) + targets.sum((1, 2))+1e-7))  # 平均每个样本的Dice Coefficient\n",
    "\n",
    "\n",
    "        batch_correct_pixels,batch_total_pixels = calculate_pixel_accuracy(predicted, targets)\n",
    "        total_iou_test += torch.sum(iou).item()\n",
    "        total_dice_test += torch.sum(dice).item()\n",
    "        total_correct_pixels_test += batch_correct_pixels\n",
    "        total_pixels_test += batch_total_pixels\n",
    "        \n",
    "inputs_npimage = inputs.squeeze(1)[0].detach().cpu().numpy()\n",
    "predicted_npimage = predicted[0].detach().cpu().numpy()\n",
    "targets_npimage = targets[0].detach().cpu().numpy()\n",
    "fig, axes = plt.subplots(3, 1)\n",
    "axes[0].imshow(inputs_npimage, cmap='gray')\n",
    "axes[0].set_title('Original')\n",
    "axes[1].imshow(predicted_npimage, cmap='gray')\n",
    "axes[1].set_title('Predicted')\n",
    "axes[2].imshow(targets_npimage, cmap='gray')\n",
    "axes[2].set_title('Target')\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "\n",
    "test_iou = total_iou_test / len(testset)\n",
    "test_dice = total_dice_test / len(testset)\n",
    "test_acc = total_correct_pixels_test / total_pixels_test\n",
    "\n",
    "# output the result, the result may vary a little due to the resize function in transform\n",
    "print(f\"Best Test IoU: {test_iou:.4f}, Best Test Dice: {test_dice:.4f} Best Test Acc: {test_acc:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJYAAAHWCAYAAACVLnA6AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAABTkUlEQVR4nO29e3hU1bk//plL5pKZJJMbuWASAqIIFtAgkfPFe9qA1xZra0pb4CigolVRW2mroO1TerRVKuKtWjlaBOS0Yr2ADyqKWEgVolYQyyVCuIRLQjKTZGaSmVm/P/J7V969k8AkzM5kyPo8zzzJ7L1n7bX3/uz3tt71LpMQQkBBIcYwx7sDCqcnFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxYoQFCxbAZDL16rdLly6FyWTCN998E9tOMXzzzTcwmUxYunSpYefgUMQCsG3bNvz4xz/G4MGDYbfbkZ+fj6lTp2Lbtm3x7lriQgxw/O1vfxM2m03k5uaKX/3qV+L5558Xv/71r0VeXp6w2Wzi73//e1TttLW1Cb/f36s+hEIh4ff7RSQS6dXvo0F1dbUAIF588UXDzsExoIm1a9cukZycLEaMGCGOHDmi2Xf06FExYsQI4XK5xO7du7tto6mpyehuxgR9TawBrQofffRRtLS04LnnnkN2drZmX1ZWFp599lk0NzfjkUceAdBhR23fvh0/+tGPkJ6ejokTJ2r2cfj9fvzsZz9DVlYWUlJScO211+LAgQMwmUxYsGCBPK4rG2vIkCG4+uqrsXHjRowfPx4OhwNDhw7FSy+9pDlHfX097r33XnzrW9+C2+1GamoqJk+ejM8//zyGd6rnsMb17HHGG2+8gSFDhuCiiy7qcv/FF1+MIUOG4K233tJsv+GGGzB8+HD87ne/gzhB1tH06dPx6quv4ic/+QkuvPBCfPjhh7jqqqui7t+uXbvw/e9/HzfddBOmTZuGv/zlL5g+fTpKSkowatQoAMCePXuwevVq3HDDDSguLsbhw4fx7LPP4pJLLsH27duRn58f9fliij6Ri/0QDQ0NAoC47rrrTnjctddeKwAIr9cr5s+fLwCIioqKTsfRPsKWLVsEAHHXXXdpjps+fboAIObPny+3vfjiiwKAqK6ultuKiooEALFhwwa57ciRI8Jut4t77rlHbgsEAiIcDmvOUV1dLex2u3j44Yc126BUofHw+XwAgJSUlBMeR/u9Xq/cdsstt5y0/bVr1wIAbrvtNs32O+64I+o+jhw5UiNNs7OzcfbZZ2PPnj1ym91uh9nc/hjD4TDq6urgdrtx9tlnY+vWrVGfK9YYsMQiwhDBukNXBCwuLj5p+3v37oXZbO507Jlnnhl1HwsLCzttS09Px/Hjx+X3SCSCxx9/HMOHD4fdbkdWVhays7PxxRdfoLGxMepzxRoDllhpaWnIy8vDF198ccLjvvjiCwwePBipqalym9PpNLp7AACLxdLldsHsut/97neYO3cuLr74Yvz1r3/FO++8g3Xr1mHUqFGIRCJ90s+uMKCN96uvvhp//vOfsXHjRundcXz00Uf45ptvMHv27B63XVRUhEgkgurqagwfPlxu37Vr1yn1WY//+7//w2WXXYYXXnhBs72hoQFZWVkxPVdPMGAlFgDcd999cDqdmD17Nurq6jT76uvrccsttyA5ORn33Xdfj9suLy8HADz11FOa7YsXL+59h7uAxWLp5JmuWrUKBw4ciOl5eooBLbGGDx+O//3f/8XUqVPxrW99CzfddBOKi4vxzTff4IUXXsCxY8ewfPlyDBs2rMdtl5SU4Prrr8eiRYtQV1cnww3/+c9/AKDX44p6XH311Xj44YcxY8YM/Nd//Rf+/e9/Y9myZRg6dGhM2u8tBjSxgPaY1IgRI7Bw4UJJpszMTFx22WX45S9/iXPPPbfXbb/00kvIzc3F8uXL8dprr6GsrAwrV67E2WefDYfDEZP+//KXv0RzczNeeeUVrFy5Eueffz7eeust3H///TFpv7cwCb0cVTAUn332Gc477zz89a9/xdSpU+PdHcMwoG0so+H3+zttW7RoEcxmMy6++OI49KjvMOBVoZF45JFHsGXLFlx22WWwWq1Ys2YN1qxZg1mzZqGgoCDe3TMUShUaiHXr1uGhhx7C9u3b0dTUhMLCQvzkJz/Br371K1itp/k73ScDR93gySefFEVFRcJut4vx48eLysrKeHZHIYaIm421cuVKzJ07F/Pnz8fWrVsxZswYlJeX48iRI/HqkkIMETdVWFpaigsuuABPPvkkgPYxr4KCAtxxxx1xd5UVTh1xUfStra3YsmUL5s2bJ7eZzWaUlZVh06ZNJ/19JBLBwYMHkZKSErNAo8LJIYSAz+dDfn6+zKjoDnEh1rFjxxAOh5GTk6PZnpOTgx07dnQ6PhgMIhgMyu8HDhzAyJEjDe+nQteoqanBGWecccJjEsI1WbhwIR566KFO2yl6HYlENF5WcnIyRo0ahWAwiGHDhiE5ORn79u1DbW0tSktL4Xa74ff7EQgE5JvHLQKSguFwGDabDSaTCaFQCKJ9jgCSkpIQDocRDodhsVhgt9uRlJQEu92OUCgEs9mMSCQCIYR8Kex2O5KTkxEOh2V7gUAAqampsFqtsFgs8jcWiwXhcBitra0AgKSkJJjNZk0fw+EwAoEAAMDj8cBkMqGtrQ1CCJhMJthsNgghEA6H5W/a2tpgNptl1gQdG4lEEIlE5H7qB90Hs9mMUCgEv9+PZ5999qQ5bECciJWVlQWLxYLDhw9rth8+fBi5ubmdjp83bx7mzp0rv3u9XhQUFMgLp4FYk8kEk8kEl8sFm80Gs9kMl8slb2RKSgo8Ho98iJFIRBKL/qd2wuEwrFarJJHVapXH8ocDAFarFcnJybIvdFxraytCoRBcLpemHwA0BLNYLJJcFosFoVBIs53OL4SQfRRCwGq1SsJTKg/dE3rR7Ha7vNakpCRNHziJTSYTgsEgQqGQbIN+Ew6HkZSUhFAopDnHiRAXr9Bms6GkpATvvfee3BaJRPDee+9hwoQJnY632+1ITU3VfOg3ADSkImKFQiFJrkAggEAggPT0dADtby7dQC4FzGazZhsnEv9LUsBsNneyNYQQiEQiCIfD8qFTP0wmkySl1WqF0+lEKBSC1+uF1+uF3++XkhCAPF7/AgkhpGR0Op2SfK2trfB6vWhtbUVra6vsB/WJSMmlNJGKJHFbW1uXz0z/Mp0McVOFc+fOxbRp0zBu3DiMHz8eixYtQnNzM2bMmBF1G0QkunH0xrvdbrS2tsLhcMg3MhgMIjU1VUorUmNcanXVLpGM3noCnY/fcK5G6OETiTgJSRIA7WqOpA/tt1qtaG1thcVigclk6qS6hBBSTZJ65RKQfsulG7VNqo+Op+ugY0wmE1pbW6X6JSlFBI0WcSPWD3/4Qxw9ehQPPvggamtrMXbsWKxdu7aTQX8y6NWDxWKBw+GQdg23Pcgm42ShB043G4AU+1xycfJxG4w/9FAopJEeZPsRGcLhcKdz2Ww2JCUlITk5WT5oIiuXcpzkpOYikYg8D10ff3EInPCkPuk6uDSm/gUCAZhMJiQlJcFkMklbLiGIBQC33347br/99l7/nut6kjL0lofDYaSkpGgkh94O0Usg2kdvcTgc7mSHcZVLD9Zut0tpQqqEHh7fRwYylyZcmtHv+XciOP9usViQnJwsHRBqn5OcjHuLxSKlDxGezsuvnatps9ksJaLdbofdbtd45dHgtMluoLfJZrMBaH+ryXANBAJISkqC1WrV2B16gnFbjUsIAF1KDr6fJAKRkQjJH7jJZNJIG/rQfiIjV3m8fQK9PElJSXA4HPLl4SrWYrGgra0Nzc3N0jOle9Ha2qqxs/R2HJ27ra0NwWBQer49iRkmNLH4A6YbRcYsqSAAUi3SVCmuwuj3esLQzSU1wtUGd8PNZrOUOqRKSA2R7aI/l81mk3YRPy9JFNpH3iC1TxKLzkMkpWPonACkOiMytbW1SWna1taGlpYW2SciLXmkdL0kkUlK9gQJEcfqDvxto4eTnJyMUCgEu90ub1QwGERaWhoAraTgv+OGPDfM9dJMH54A0KUNRe3rDXv6LZcyFGogAnXnNJAKI4nD7TS/369R1USi5ORkWCwWqQJNJpOMt/F7CEBjF9I5ieycyNEgoYlFD5QkCD14IhK94SSxrFarNOT10ol+r2+LqyC9CuW2EycQgdtt3HMldUkeGKlnCktQ/6iv3PgmopFES0lJkaqOQgZ2u11KKJ/PB5fLpXmRSN21trZKlUpeotPp1EhTsvkSxiuMFfiD4HYUvanciNY/aO5RctuKSx26qdwT5GqWtvHAJJ2T2mltbZWk5pKR1FVrayv8fj9sNpskKPcO+Uwc/tBJFROZSAqSsU7X0NzcLKWd1WqVgVDqb1tbWydbkGw1u92OQCAwsIjF3XOgXZ3ZbDZpxJL04qEGLnXo5uulDEkVOp6rRq7+eKiCAq566UJqmXuFdB6gQ+rxMAW3Z4hYdrtdfqe2udojwoZCIbS0tMgXRk9UPkRF3jMPkFKb/PculwvNzc0au+xkSGjjHeggF9Aef3I6nZqgH98HQBMcpN8D0JCEf+d2GI8NAZAPiewrClySiolEIvD7/QgGg9LGCYVCaGtrk0Y6GeNJSUmw2WxSanHpyp0RUmMUdacwCjfkqd88BEMvXEtLCwKBAOx2Ozwej8beArQ2F52f7LKeIKElFr1R3G2nh52UlIRIJIJgMAir1Srdcr206iqmpbfd9KQj6WW1WmGz2aQL39bWpom8Ux8pGk7nodgUSQwAkozcuyPycIKQ1AkGg5K0TqcTDocDbrcbVqsVXq9X9oUHaele0eB7MBiU10Hn1TssnKQ9QUJLLP5g6MaFQiH5JpPEsNlschCXe1sEnjmg/5/aBtApk4EeGNlC+vgX2TUkIXlQlPZT+zxaToPfFJUnFUXnIFVParCxsRF+v1+SmNrkJCEVnJKSIvtEZgKXzNx208f09ENfJ0JCSyx+0fyNdDgc0i0PhUJwu92aoQ4ujagdHtAkkKHODV1SCzTGR1LTZrOhpaUFoVAIDodDqjrufZGtxSPvPKBqMpk0wzJ8WIdeIj4EQy8MhS2ampo0EpnCDGTQc4eE+qxXm1xKc1tR7yGfDAlNLH7BZrMZDoejk3pra2uTEozbEgQy1DmpeHiAHgLFmiiyD3R4VADk+Qk0gMvVHnmH9OFjkn6/X4ZGyKPl4QFqU6/6qS2eB8avKxwOyz5S1gfdB64m6ff6e0toaWnpZGOeCAlNLEAbe6J4jcPh0ESxXS4XgA5XHehQc/Tg6K2mlBpSNfSweOki7v7Tgyb7iKSfw+GQdg4ATRCU8sVIggnRnvJL6T0tLS1wOBzSLqTr5GqcVCpXkTxcwr1Vs9mMtrY2jQPDsyC4Tam3sSwWS4/HCYEEt7G4203xHBoyoZvJ7RH6Dfck9W47EY7bRSTxuP3Fj6F29GOCXN2QoU+ZDC6XC06nU3qCGRkZyMzMlATmthk9YL09SefmyXlAR6YCABnbM5lM0minD0lfvk0fBqE+UAZstEhoYvG3mC4+EolIz5CGIrhHCGiNZkBLCk44IgTFxPix/IHTg+S2EgBp5PO8rZSUFDlmye06CpVQCED/oEnacuNcP0hO0ornWnEHg142s9kMt9stQxA8+EsvD5kR9KLyWFo0OC2IxcfAgA5VRWqAj311BS71eDSb7BdSPdS2PojIDWqemgxAQ0r6zge9+YCzEAJ2u13OPtJLJlJlZL9xyctjdvyekNSxWq2S1ElJSfD7/WhtbZUxNvod/eXXyckXLRLaxuJxLMrnJtuDhlL0Yl5/k3jAkR405TDpVQXdeIp0c2OWVBpPSaaHYrPZZFCUbB0+8AtAI71I8vJBbbpekkJcsphMJrS0tGgCqfylIw+Z7Ef6S+ejtCIe8iBSc0nfE3IltMQCOlQTxanIBQcgjXAiCI/ZANp0W27Uk23Ex/XoIfMHTZKM1BMfNKaPPoOUe2RAR6iD2tJLKj6eyccoSRLRQDENRAcCAXkOvWHe3NwsRwDo43A4JLk4+am/dN84waJBwhOL3lqa7sQNdYopAdoHScY1PVBqh8/M4UFBbufwcUmyy+gBUmoyV31EZCJ+U1OTtHO4J8bJpx+T5O3oHQ/yIgFoho641xsKhZCcnIzk5GR5PeFwWOaoUdyP95syU/lwVU+Q8KqQbrw+2Y7SUex2u2amij5jgRu3BK5OeOyIsjIByIwA6gN/CEBHHIlLIZPJBL/fD6/XK6Uoj6bTuUkKtbW1SbuRVDuNJnD7DOiwN4m0+twpSqFOTk6WJgLvL5f0gUBAGu+kDXhYJxoktMTibzzZMVz8B4PBLufS6cFDC/RQuO1Fv9MPafC3mfpCAUgu5bghTNmbXq9XE64ge4Z/p2EXUk3UJk12JYLQw6cXh1Qiz6Zobm5GW1sbrFarZiSCZ6TykQSSdF0lBEaDhJZY/A2i6DPN0uUGalfjf/qAINkwJOGoTX5D9UTj0obbajyGpPfoAoEAgsEg6uvrpZQl4nOvj9QydzacTqdU95SlYDKZZNYsRfJJRRJhKJEvGAzKa7BarZJ89J2ulSQUqVUu5aNFQkssAo9VAdrIOE9t4d4SV19EFp7bzUMU3ODmNg+/0TzQylNz+IOx2WwyIh8IBHDkyBE0NDRIg5vapqwF2sYDr6SSXS4XHA4HQqEQmpub5RgjDVrzPjkcDk1w0263S5Ly6WP6zIvk5GQ5GtHT6HtCSyygc6Ict0mAjvQTfeSdEw3QpovwsTluwNOD4kMr1DZJGB4L4sfRsBE3kskT8/v9cLvdMszg8/nkA6f+kITkU+5JutJsagDyBaOwBtlmFPWnvlG/KXwBdJ5Ox+1G/ajDyZDQxOLxFbr5ZBSTPcGLY3AbhsfA6Dv9NikpCYFAoFMeExFDP9BNkXNugNN0K5Opo0AHgew+UlvNzc1obm7WEIfsI5J0PO2HJAw30ulYIiS9UDReyp0bsvF4NisfdKYALP2Gq/pokdDE4jEdesMoVEDqi4cOgM6zbfg2HiQkFcDn6XFPEegYWKY2AGjsK/4/gWxBGoSm85Ha41Vt6IXw+/0y4Q/Qeog8ak+ZDDQeSak8ZA5QnMrr9WpeMB4r444Ot626yw7pDglNLC7SuSQhVUDfebiBJBB303l7JBl4arPeW9ITmrvu9ADJcOfShc5LsTUuZfUD4LSdcrn4hAaSZjwGplflXOIFg0EZ8CWHhmY3kxfJMy3o2ugFpWugUYJoEHPjfcH/v4Qt/4wYMULuDwQCmDNnDjIzM+F2u3H99dd3KmcULbhRC0BmEgAdqoGHEYCOnHfaxoOXQIfXyD0oAicaN9K5SuISR2+/8bY4SYnQtJ3UWHJyMtxut0yhIaO/paVFEyqgsENqaqrMbG1paUFTU5Pst8/nkwY4SSE+R6ClpUXTZwIFXXsirQCDJNaoUaPw7rvvdpyEBevuvvtuvPXWW1i1ahXS0tJw++23Y8qUKfj44497fB7K/eaSQZ+loP+fD8vwAV69gQ9As5/A1QaXHnR+cvn1DgJ5nHrHQN837gXyfRZLe7ETioiTquUqms7BpY7NZpPhCC6BiOj0lzxTGgAn0lKbXJpFA0OIZbVauyyg1tjYiBdeeAGvvPIKLr/8cgDAiy++iHPOOQebN2/GhRde2KPz0FvMA5tcLXGPjkskHouih8dtCfqf2x98riCPlAPQHMuJxW0/sq34wC7PqvD7/ZpxRW67ke1HqSs8IEzXw4OiZC86HA6ZtkNSmQKuRPKkpCQZ2iB1R6pUT6S4x7F27tyJ/Px8DB06FFOnTsW+ffsAAFu2bEFbWxvKysrksSNGjEBhYeEJi9oGg0FZnIw+QMdbT+qM2yhtbW3SsOW2UFfjXkQU/YRS/fgi/07gRi+pJyIFH17iUpVLF+ordwSAjtgZN9iJZBQ6cDgc8juFK3hKsn7Yhkf66V7ReCFPCLRarXLmD0lK3qdoEHNilZaWYunSpVi7di2efvppVFdX46KLLoLP50NtbS1sNhs8Ho/mNzk5Oaitre22zYULFyItLU1+aLkQcpe5ZKGbSekh+jmEXRmofBSfos1Ax5gh7eO/AbSqUk86fi59Og+B9nF7j37L7T7eF+588Mka9DLwiSQtLS1SQhGBidBEVKomw/P6eSCWyEuD1dEi5qpw8uTJ8v/Ro0ejtLQURUVFePXVV3u95G13NUhJEvCMAx6w1E8l16es6Me+yBCmImj6cUA+6YJyn3j1PN4OPSx64FwV8/9JbdIgMYGPHOjTYDjZm5ubNdPDhBCSXCkpKTKPno8CcDVM3ibdI6CD2JFIR1kkkrzRwvAhHY/Hg7POOgu7du1Cbm4uWltb0dDQoDmmu6K2hO5qkJLEolQXAkmsQCCApqYmKRG6GkTWG9TchuKhBB7ncjqdcoKGPvzAQxY8nkYGMw9e0gMOh8NSvXFSUl/o3NQfIhgfvuHXRIPWVImPAqwUeKX+kl1G0o9m4hDh+Tgixd2iheHEampqwu7du5GXl4eSkhIkJSVpitp+/fXX2LdvX5dFbU8GbrMAHeEHUovknvOhF/0gNA+YklHMpQIAjUTkExMoqs2He6hf9OEPjKBvn4aiTKaO9GreV5KWdC7eBmWjkmQiKc1rOOgrNtN18nNRf2mYSW8y0G+iRcxV4b333otrrrkGRUVFOHjwIObPnw+LxYKKigqkpaXhpptuwty5c5GRkYHU1FTccccdmDBhQo89QkCbqUCSgECk42sG8gdKN46HBGhyAU8JJlLyCD6vz8Dn5ZGkIvXGc73a2tqkVCKVxYdQyIDnfeUSTh8H05MT6JzBStu4YwF0pORQOjWvUEO2KWkBPpjeE68w5sTav38/KioqUFdXh+zsbEycOBGbN29GdnY2AODxxx+H2WzG9ddfj2AwiPLy8k4LckcLLokAbWyIpElTU5M0WLua2Ml/qycQtzXIped2lr4fvB1uD3EVZrfb4ff7NcX+ubfK7RxS1fpALb0ERFAyyglJSUmS+CS5CKQSue1EWRc0xkkSj8jNzYFoEXNirVix4oT7HQ4HlixZgiVLlpzyufiD4RKIRLgQAvX19RppoE/644FSbljrA4L8RvMYDz+O/vLhHGqfT98iG4pUJHlevEwjB0/J4cNDKSkpcDqdUirzHC0iCdlIJIHoXmRkZMja8kRUh8MBp9MJn8+nCb7q72s0SOixQv4W641i2ub1etHS0gKXy9UpGg50nuypJwpXjSRRulLB3GbTS7vs7GxkZGQgEomgrq4OXq9X497zYC61qw/kUt/03qfD4UBGRgaSkpJw8OBBSU7KaPB4PEhLS4Pdbkdra6u0OZ1Op7Sn+CQPPvRF94mHbKJFQhOLSyz+YOnGkpF95MgRZGZmdjKYiYicPIC2jJFeGnU1DEPfuT2UnJyMcePGobGxEfX19XC73VJFHTt2TA5UE1G5A0LGsz4fi/pGpATaQy9utxsejweNjY3SBAAAt9uNnJwcjbQmm6qhoUGOHVJwmHLpKX9L/xLGVRX2NfSRZKBj6MFkal8fZt++fSguLpaDuXRDSQLRSgxc1HPRzyUhf+B8zJAfZzKZkJubi7POOgtutxvHjh3D1q1bcejQITQ2Nso+kkrjZOX5YXwEgAdjeWQ/EAjA6/UiJSUFbrcbPp+vU2FbfX8pfNDc3CyHcmjyBp2XpoVxydUTYiV0anJXep/nEBEOHTqEQ4cOaWJU9HDImCevh6tG/rD5wgDc8OdBRB4P83g8OH78OLZu3YrW1lZceuml8Hg88Pv90kkAoIn0c++PJBOdh64X6IjCC9GeWkNJe1T7ivrAEwzpGnlYha6Z7hUPkXAPU3/+aJDQxCLQm8/BI8xerxd79uyBz+eTwxT6DAJ+MwGtVCAbiI/fkcfFI/I80c7j8aChoQEHDx7Evn37cOzYMY13Rm1wYnHVQ5KC53PxoCsfIiLjm8b4uG2mV+VkOlD2Ar+HHo8HDodDYxIQekIqIMGJxSPTXNzzB0IPe8+ePaipqQEAGbHm5OIPjZcfojZ5CIHHm7iUIuKlpKTIgrButxuDBw+Gz+fDsWPHNBKPHlZXhWapXT6ux8lGH55J6na7NUVHeEiDro9+x4nFz52SkqKx6aif3LmI6tn04Dn2O+hH7wFt/jt9N5lMaGhowOeff46GhgZpQ/AxPn5zI5GItMP0kpA/JP1YIhnHo0ePxhlnnIEzzzwTY8eOhdVqRVVVFZqbmwFoZ1TzNgBtFiv1h8e5+GgDJxdNyEhPT5f9aGpq0nh+FLk/fvw4jh8/DgAyzEDq3mq1yu9627EnxEpo453bD9yTA6BRE4RDhw7h888/x7hx45CamirtDD7xgae98KEcemPpodOx3KjPzMzEueeei7Fjx8Jut+Po0aPYsWMH9u7di6NHj2psFvoNkYzn19OLwoOsZHvxF4Gi/SaTScaeqDxRY2OjNO6pvhelJtO5+KxuADLuRQSk3HwCl+InQ0ITSx9z4TEcnv4BtD+ItrY2fPnll7Db7SgpKZFvNv0G6Mg5Irc/EonIdN+srCx4PB75MGm4xGazYfDgwSgqKkJ6ejqCwSC2bNmCjRs34ujRo1L60bk4aCiFjxrwxD6uHull4TnzPEBMQzM8p5+kGcW1KBhrMplk+IP2UQSfXgAiFjcVon42p/Bc4w7926QPmOqHbYRoTzOpqqrCzp07AXTkmBORuJqjB8TtEYvFgubmZtjtdmRmZiI9PV2u60zZmNu3b0dlZSXq6+tlbVHu6emHR/j4ItBRVJerPO4hkldJQ0TkLHBpxjNEeWCVSErhBb6iKlfvJLG4qTGgIu/cxgE6ezL0JtN3IdorvnzxxRfIzc1FZmamXOSIS0CuBomkLS0tGDx4MNra2nD48GG57ozJZMLx48fh8/kwfPhw5OXlIScnB4cPH9aoUb3tRw+MB1n5cTxswq+RJAmpSR5kpVACjW2aTCa43W7NILReaurPw0cueKrygFGFeo+Q21z04W8r2Q5CCBw9ehTV1dVIS0uT0oZ7XTzqzm0iyiVLS0uTqSfUfmNjI3bt2oWxY8eivLwcQ4cOxe7du3H06FEcP35crtDF6yToo/zUFl/BnlQ1V6UUw6L/Sfrw+Yc0sExE5UM31DagjV+RhNMHbfWxwZMhoYmlf5t5FJ6P4fE4EE8h8fl80gAm0E3mXhflSlGmABn9VGOBYmNutxvJycmora2Fz+eD2WzGBRdcAIfDgUOHDmHTpk04cuSIPD+PN/GHSX0mInEbinuJZG9R+MRsNsucLD7sQ2qdB0U54bhdSsfRbCA6Vj9OeTIkNLF4qizX/1wtdvXwAEg7RR+R5jYHt81MJpPMIaeHzXOySIV5PB6MHj0aHo8HNTU1qKmpQXZ2NvLz83H++eejsrISTU1NGilALwVPveEE4y8JH5bhmaQknQ4dOiSzZrsaAtLXQKV9lF0BdMy95ImAPUVCG+88ak5utJ5g3OOi35jN7SW2Bw8eLB8ilyAUVqC3mEe7KZWX2xvUdjAYxP79+/HRRx9h69atyM/PR0ZGBhoaGrB79254PB6cf/750uimNvQJfrSNHnxXQyvUZxrwpjmHR44c0cyq5qA2+SQRMvTpvrjdbo2JwYPIA2askDwYIhAf/uAk4TeRbmR+fj4GDRoEAJ3cbD5rhYxhHgmnNvnbzw3xQCCAPXv2YO/evSgoKIDT6URaWhpyc3M1y9vRhA9KY+EjCfRgeayMZ2LQ72jShMlkwuHDh3H8+HHNC8GdAj6Rls9u5uqfDH9OTH7N0SKhVSGP/QDauXj8IXFPx2KxwO12Y8SIEZrFI+lvV8FWklDcZacHwR0DoOMhWSwW7Nq1C0OGDMHw4cPhcrlQU1ODTz/9VJPRSlKCouMANKEG6jNP86HzOBwOZGVlwWQyoa6uDnv37pUp0FyicrVKhKOMDgqa0gsXCoWQmpoKk6ljBg+vpBMtEppYPLOABzm5oaoX4RaLBYMHD0Z+fr4MchIB+RAK3WjuENBfbutw8pIUozaamprw4Ycf4rzzzkNTUxM+++wzeL1eabvQBFMiIpVO4uBOB69QSEHburo67NmzR2bK0n6K6HNJzu8T1dHiDhCP7qempsoAMbcDo0VCEwvQZnmS3WI2mzX5RFyNOBwODBkyRK63wwlJ7QEdK8PzWJn+OK5y9PEz2n/o0CEcPnxYEpcPrdBfXutd7+bz2Bf3VgOBAPbv3w8hhCZrgg9HkaTjao3bhzwATKACI3zKPUnAARPHAjpHi+nDMyDpRpnNZuTk5KCgoAAWi0XWNOdB1K6i9URKaodsK7Jx6Diudmm4B4BM1yGSUGlHt9stq7zwIKmewFyVUV9pphCPL7W1tUnjm6tmTjIAckUKkq4844I8XZo4wiVZT5DwxOKiGuhwqSn2xAlit9tRUFAgC+Dqc5L0mQzcLuGqkH5L3hUnBK8CSL+lQCZ5mgBk3pNevVDVF6oZSsfwaVx8iRIiHV8YgJNUfxylIJPzQFKTXhB+P2kskcfSokVCe4VAx4OmUANJIIrz8DfN6XRK24reeAKPsuvb5tKCE4yrKQpRANq6XeTeky3IywiRCtPXBqVoPrXD1b1+uIk7JjRcRZKMVCeXSjSZgvpCqlJflIS32xsktMTi0fFAICDrl/OQAVdjVFQE6LxavZ4ktGil/uFxY51+y2sn8LADD6ry6s366oNEWOov98JI0nBiculCIGO8tbUVTU1NMmDK1TYnHb8mUulEWHIu2traZHiEB5ejQUJLLB6/ou9kdPJBWnLtc3JyZGESvTEOaMnGA5BcvRAR9FIJ0I5FUl/M5vYp7vR78uaIWDy0QN85YXjqDEmXrjxdOh8fbqIVviiZj7Id6IWhaDt/qYQQMi+LXhhe6SZaJDyxSGKRa0wLaJPEohuSlJQk598B2oozenICWpIAnRd44uenLE0KWlIRWbK59BH9rgx+oGPZFn02hp68/C+PjhOofZ5PT4tg8tRlHvSlc5FTwZMg+czoaJHQxKJyQ0D7cAovVcQHTmkIJy0tTSPVAK064SQCoDGSAXRSXVya6fcTyNgmogPavHxOaB4z46qSe7XcGdGnRrvdbqnm6BqDwaBcBJ0Gy6kmA8UBeTIfvaBcdXeVu3Uy9JhYGzZswDXXXCON4NWrV2v2CyHw4IMPIi8vD06nE2VlZTKpjlBfX4+pU6ciNTUVHo8HN910kyzE2hMUFhZKj0tfa4BHtoGOynWANtjJoZ9EwCUP0OG6k9HLycBDAvw7j9Drs1rpuO7sNiIYV318cgWVd0pLS0N6ejoyMjKQlpYm1R6119LSIucd8heOJDtHKNS+0gVJOHqmXP1Hgx4Tq7m5GWPGjOm29sIjjzyCJ554As888wwqKyvhcrlQXl4uazUBwNSpU7Ft2zasW7cOb775JjZs2IBZs2b1tCvIzMyUhi6XQjwKTv/b7Xb5FtIxgLaiMd9HNg2pOa5y9GEMHrHmg9p0rF4CAR0emT7S3pVq4/uAdu82MzMTGRkZ8Hg8chCaxg0zMjKQlZUlc91DofaVK7iK5ddNKpocCyIv9Z9sNEMj75MnT9ZU7eMQQmDRokX49a9/jeuuuw4A8NJLLyEnJwerV6/GjTfeiK+++gpr167FJ598gnHjxgEAFi9ejCuvvBJ/+MMfkJ+fH3VfKFZFXhyfukWpJEQYMppDoZC0hfiN5Q+UHji3hQicHHrVBXRUeukq7sONdJIYBH0ZIx4xJ+lDRTv0lf94YJiMbrfbDQAyMZGOo/oNfMiHB5G5GiQDn6R/T9bTiamNVV1djdraWk3x2rS0NJSWlsritZs2bYLH45GkAoCysjKYzWZUVlZ22W53xW15mUM+cg90LNxED5KrGzoe0C73oR+WsVqtcLlcUt2S1OGeGldl9KbzmBCPguvddf04nr49il2RJHK5XHLgnPbxaV30chFpqEa8zWaD0+mUoQiK+elVM0XxqX5pJNKekkP2laGq8ESgArU5OTma7bx4bW1trUxXIVBZne4K3HZX3Pbw4cOayZ50U2gbnzvIbRQuaUgNcPLx4+k7H94AtAuUUwEzvp+XZuTjiURu/nt+PKkuqqHlcrng8XjgcrlkKIJH21taWjTzBnmJR7O5YxECl8slPVgeFOXkpu801EXmC0n5nqQmJ4RXOG/ePDQ2NsoPzWg+ePCgxpvh0794HIZuGlcbQIfNQm+5njhckvB0FRo24nMCeaCVzgVoPUDqC5c4dH5SUX6/H01NTTIblb8cpOrDYW25cUpH5vlpfDjK7/fj6NGjsggIEY8b5WSTUpyPXlKaZELnihYxJRYVqNUvYcKL1+bm5uLIkSOa/aFQCPX19d0WuO2uuC0vA8ndb3rzyT7g0kqvjriHRg4AJ5ReXXCjnI7lXiLFfEgl8sAm/y2pUz5+RwuGJycnyyVhiKTknXJ1R+fnMTf+QnD7kZOCbE56KbjTw50ckoA8thctYkqs4uJi5ObmaorXer1eVFZWyuK1EyZMQENDA7Zs2SKPef/99xGJRFBaWtqj83H3nG4+PUDyZrha03tmdOPpfx5QJSLxuBWXMLwtUl/6CD0nO8WyqH0e6OQ2THZ2NrKzszUjBJSNQAPUVLude8L666R+U/YrXy2VBulpvUN6KegYum/6BMOeDOn02CtsamrCrl275Pfq6mp89tlnyMjIQGFhIe666y789re/xfDhw1FcXIwHHngA+fn5+O53vwsAOOecczBp0iTMnDkTzzzzDNra2nD77bfjxhtv7JFHCGjXcqZhDMrGJFVIbx/PZCBCcfXFvUIe5OSZE0D7mByllnBycJXHMyAArcfHpYPFYpFGssPhQHp6OtLT0zt5oeTp0gC7HiQZ+RgiZZJyW5GulU8bo+AorTTm9/s1Rjpdp17qnQw9Jtann36Kyy67TH6nwv7Tpk3D0qVL8fOf/xzNzc2YNWsWGhoaMHHiRKxdu1azqsGyZctw++2344orroDZ3F7o9oknnuhpVwBosxGoniZJCS7WudHLHxy9xXzSAEmerlRLd8Y9DzlQe3QcTc/i9g9XmzR9ny+vywlIROF2ER+cppXpeXkjXiSX/hL59H3nEp6IxY17uh/6gPKJ0GNiXXrppScUiSaTCQ8//DAefvjhbo/JyMjAK6+80tNTdwK/UHpQfLyQKteRmNcP3tLv9BkJ3H7i0otUG52Lt8HDBFyKclKRzaKf+ZKeno6UlBQNgfgwEKla/fmAjmlb3LakvpK047/t7nooA4OGgfh+slf71coURoJLAKDD9qHJlhQKIBuDbjAPDHKJw99kMsrpoelDFHSc3uvjUpEnAuoHvckQd7lcMphJxjRVP+Zr3/DFmUiN8rQeUs90LPWR7DKypchEINVGtha3A+l+UtlJkmA9kVgJTSwCj+uYzR312C0Wi5xWRSQh8MFqaoO767QN6Egi5BKPG7M8VgVAI7m4gQ1AY8xTAJJnXPBMAiI3qSU+iE3DLEQyPqREUi4cDsPn80nJo8/sIBvL4XBIp4BeFLLFuM1oqI3VX0EPgR4OEYskll69EfTpIPooPDfwAXRaDoT+cgOeG/x8EJm2URYnrajFbTl9FicFK2mf0+mU56QXyG63y+1kX5lM7XUmKMhJSZD8BSLVTPeIx7d4mnVbW1uP1tEBElxicVsB6PCGuJ1AtTkByOL6ZMRyqaMX9dwg54PKXA1yW0dvLBPR+W/I/iNVTSWI+PF8yWF+XfSXq1Ie3iBVSb8NBoOaYmypqalyTR2LxSJJGQ63r97BY2308vF7wiV5NEhoYgHaqDhFocnOAtpztmi4hMjGI+P6tnghDR6rolgTd7u5VOJk4FIMgFRvlBdF7fJ1dbj3SfEkfTCWB2JJGpOK5J5gc3Mzjh8/Lu0uLgnpWiiAGw6HNRF5HhckNcw92miR0KowKytLlqLmxjaNxHOvkEfA6cHQA+Q3jEsJTlhulOsnP9DD42Nw1AZN2uCeG9BR8I3OT/+TYU2EoRpXZA8BHdPlec4ZbWttbcXx48c10tlkMsmxR3rxyDygF5ATjRc5oWExIM4rrPYlLrnkEgwdOlRjONPFU0otv/m0jRNJ7+l0RzSeScCzKoAOicJjWfQQm5ub5UPVG7/c6yR7hk975zYhtxV5nIyHLpqamtDQ0CDnKlLfTKb2SR20nRZSIMnF1TqX1Hq7dMAY78XFxcjMzEQwGERNTY18AzMyMmREOykpCS6XS5YOopvFxwgJXMVRJgGPcZEaAbTZCPqhIfKwSIrqicxDBDzZjx4gn9BBx5L04KEMeglaW1vR3Nyskd7k0ZFkJslHxKUwAqXFkJdK56FcNlKH1PdokdDEslgsGDRoEM4991xZqpGKXZDharVaZUSbJBY32olkXUkuTj4+tGE2m2WdBXqIpLp4nhYRgn6jJyplNFAcyWazyXOQlOUrzvOYGTkUZLc1NTVpptqTt0fL2FH/eMEQCt7ycUJ9bj63OXuChCYWGe75+flIT0+XxKJM0UAgAIfDIQ14XlCfSABoB3B5gh5B7y0CHSqX2qCHzVN1OLHoYfJwBE10SE5OllFv6gvPTaf/qT0KovLZzzyQy+NeZJiTaiOJS14kv2bu0PAXgrJIupLy3eG0IBYnBQ/0NTc3y8xLIha3H8htp4dNJOHRbX28Sj8kQrYHqUWKqNMgcFeSkCQdkf/48eNIT0/vVJKRG9lAR9zK5/PJcpDkzfHIO0/6o/95OITsTn4N3K7jjgqfTtcTJDSxgPabT7NKKAWFhiboTaM8cfLQ9CtbUTv0QHk5ILrR+omkXN1xQnKSU8YAJ6M+g5ViWgCk5KJ+kHQSoj1zg2ZV86EZHkDlQ1MkpcxmswxuUts88k7Xx2NhpGb5tXGpGw0Sklj0ICg57ptvvoHP5wPQMX3J7XZLwpEb39LSgqamJlm8tat5gwT9UA9PW6GHwyUGDQaTy88RCASk7aJ/OPQbmm5Fdg+1Sf3nA8NcglFQ1WQywel0akhCK1AQ0WigmQK1fEY2ZYaQU0H3gEs23oeTwSSiOaqfYc+ePRg2bFi8uzFgUVNTgzPOOOOExySkxMrIyAAA7Nu3Txb5UOgMr9eLgoIC1NTUyHTuU4EQAj6fL6qEzIQkFtkgaWlpMblhpzv4PIFTRbQvckJH3hX6LxSxFAxBQhLLbrdj/vz5Pc4RGmiI531KSK9Qof8jISWWQv+HIpaCIVDEUjAEilgKhiAhibVkyRK5bElpaSn+9a9/xbtLfYb+VKrzREg4Yq1cuRJz587F/PnzsXXrVowZMwbl5eWdKticruhPpTpPCJFgGD9+vJgzZ478Hg6HRX5+vli4cGEcexUfABCvvfaa/B6JRERubq549NFH5baGhgZht9vF8uXLhRBCbN++XQAQn3zyiTxmzZo1wmQyiQMHDsSsbwklsVpbW7FlyxZNKUqz2YyysjJZinIgw6hSnb1BQhHr2LFjCIfDJyxFOZBhVKnO3iChiKWQOEgoYmVlZcFisZywFOVAhlGlOnuDhCKWzWZDSUmJphRlJBLBe++9J0tRDmT0danOEyJmbkAfYcWKFcJut4ulS5eK7du3i1mzZgmPxyNqa2vj3bU+gc/nE1VVVaKqqkoAEI899pioqqoSe/fuFUII8fvf/154PB7x+uuviy+++EJcd911ori4WPj9ftnGpEmTxHnnnScqKyvFxo0bxfDhw0VFRUVM+5lwxBJCiMWLF4vCwkJhs9nE+PHjxebNm+PdpT7D+vXrBYBOn2nTpgkh2kMODzzwgMjJyRF2u11cccUV4uuvv9a0UVdXJyoqKoTb7RapqalixowZwufzxbSfKm1GwRAklI2lkDhQxFIwBIpYCoZAEUvBEChiKRgCRSwFQ6CIpWAIFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxFAyBIpaCIVDEMhhDhgzB9OnT5fcPPvgAJpMJH3zwQdz6pIe+j7HAaU+spUuXatbtczgcOOuss3D77bd3mibVn/H2229jwYIF8e5G1EjIcty9wcMPP4zi4mIEAgFs3LgRTz/9NN5++218+eWXcnWwvsDFF18Mv98vF0iKFm+//TaWLFmSMOQaMMSaPHmyrFdw8803IzMzE4899hhef/11VFRUdDq+ubkZLpcr5v0wm81wOBwxb7e/4bRXhd3h8ssvB9BeSGP69Olwu93YvXs3rrzySqSkpGDq1KkA2ifELlq0CKNGjYLD4UBOTg5mz56N48ePa9oTQuC3v/0tzjjjDCQnJ+Oyyy7Dtm3bOp23OxursrISV155JdLT0+FyuTB69Gj86U9/AgBMnz5dli3iap0Q6z7GAgNGYumxe/duAEBmZiaA9mnm5eXlmDhxIv7whz9I9Th79mwsXboUM2bMwM9+9jNUV1fjySefRFVVFT7++GO5PvSDDz6I3/72t7jyyitx5ZVXYuvWrfjOd77TacGmrrBu3TpcffXVyMvLw5133onc3Fx89dVXePPNN3HnnXdi9uzZOHjwINatW4eXX3650+/7oo89RkxnKfZDvPjiiwKAePfdd8XRo0dFTU2NWLFihcjMzBROp1Ps379fTJs2TQAQ999/v+a3H330kQAgli1bptm+du1azfYjR44Im80mrrrqKhGJRORxv/zlLzWTSYXomHC6fv16IYQQoVBIFBcXi6KiInH8+HHNeXhbc+bMEV09LiP6GAsMGFVYVlaG7OxsFBQU4MYbb4Tb7cZrr72GwYMHy2NuvfVWzW9WrVqFtLQ0fPvb38axY8fkp6SkBG63G+vXrwcAvPvuu2htbcUdd9yhUVF33XXXSftVVVWF6upq3HXXXfB4PJp9vK3u0Bd97A0GjCpcsmQJzjrrLFitVuTk5ODss8/WrH5qtVo7LZW2c+dONDY2dqonRaCqLXv37gUADB8+XLM/Ozsb6enpJ+wXqeRzzz23ZxfUh33sDQYMscaPH6+pYqeH3W7vtMxuJBLBoEGDsGzZsi5/k52dHdM+9gb9tY8Dhli9wbBhw/Duu+/i//2//wen09ntcUVFRQDapcfQoUPl9qNHj3byzLo6BwB8+eWXmhKPenSnFvuij73BgLGxeoMf/OAHCIfD+M1vftNpXygUQkNDA4B2+y0pKQmLFy/WLGu7aNGik57j/PPPR3FxMRYtWiTbI/C2KKamP6Yv+tgbKIl1AlxyySWYPXs2Fi5ciM8++wzf+c53kJSUhJ07d2LVqlX405/+hO9///vIzs7Gvffei4ULF+Lqq6/GlVdeiaqqKqxZswZZWVknPIfZbMbTTz+Na665BmPHjsWMGTOQl5eHHTt2YNu2bXjnnXcAACUlJQCAn/3sZygvL4fFYsGNN97YJ33sFWLqY/ZDULiBl5/WY9q0acLlcnW7/7nnnhMlJSXC6XSKlJQU8a1vfUv8/Oc/FwcPHpTHhMNh8dBDD4m8vDzhdDrFpZdeKr788ktRVFR0wnADYePGjeLb3/62SElJES6XS4wePVosXrxY7g+FQuKOO+4Q2dnZwmQydQo9xLKPsYCqj6VgCJSNpWAIFLEUDIEiloIhiCuxBvIqXqc74kasgb6K1+mOuHmFpaWluOCCC/Dkk08CaB+aKCgowB133IH7778/Hl1SiCHiEiClVbzmzZsnt/VkFa9IJIKDBw8iJSUlqgwAhdhACAGfz4f8/PxO46p6xIVYJ1rFa8eOHZ2ODwaDCAaD8vuBAwcwcuRIw/up0DVqamo6ZYLokRBe4cKFC5GWliY/ilTxRUpKykmPiQuxerqK17x589DY2Cg/NTU1fdVVhS4QjfkRF2L1dBUvu92O1NRUzUehnyOmI489wKms4tXY2NjlQkXq0zefxsbGkz6juGY39HYVL0Ws/k+shMxu8Hq9SEtLi3c3BiwaGxtPao4khFeokHhQxFIwBIpYCoZAEUvBEChiKRgCRSwFQ6CIpWAIFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxFAyBIpaCIVDEUjAEilgKhkARS8EQKGIpGAJFLAVDoIilYAgUsRQMgSKWgiFQxFIwBDEn1oIFCzTLy5pMJowYMULuDwQCmDNnDjIzM+F2u3H99dcn1GryCtHBEIk1atQoHDp0SH42btwo991999144403sGrVKnz44Yc4ePAgpkyZYkQ3FOKJU6sX0xnz588XY8aM6XJfQ0ODSEpKEqtWrZLbvvrqKwFAbNq0KepzqGoz/b/ajCESa+fOncjPz8fQoUMxdepU7Nu3DwCwZcsWtLW1adblGzFiBAoLC09Y1DYYDMLr9Wo+Cv0bMSdWaWkpli5dirVr1+Lpp59GdXU1LrroIvh8PtTW1sJms3Va+zgnJwe1tbXdtqmvQVpQUBDrbivEGDGvmjx58mT5/+jRo1FaWoqioiK8+uqrJ1wB9ESYN28e5s6dK797vV5Frn4Ow8MNHo8HZ511Fnbt2oXc3Fy0trZ2WiW0u6K2BFWDNPFgOLGampqwe/du5OXloaSkBElJSZqitl9//TX27dvXZVFbhQRG1K5YlLjnnnvEBx98IKqrq8XHH38sysrKRFZWljhy5IgQQohbbrlFFBYWivfff198+umnYsKECWLChAk9OofyCvu/VxhzG2v//v2oqKhAXV0dsrOzMXHiRGzevBnZ2dkAgMcffxxmsxnXX389gsEgysvL8dRTT8W6G6c1KioqZA3W48ePY+XKlXHuURfolViKMwayxJo5c6bw+XzyXni9XvH3v/9dlJSU9CuJpcYK+wijR4/G2LFj4Xa7T6kdfRspKSn43ve+h3fffRcul+tUuxkzxGWRpoGCOXPmyBXKHnzwQTgcDlx++eVYv359zM+VmpqKG2+8ES+88ELM2+4VjFRZRiERVOF9990nAoGApt/Lli0THo/nlNpdsmRJt/dl06ZNwmq1KlV4umLQoEG4/PLLYbfbNdv37dvXKYYXS5SWluIXv/iFYe33BIpYBuCyyy7DpEmT+vy8JpMJkyZNQmFhYZ+fWw9FrBjDZDJ1aaDv2LEDTzzxhOHnnzhxIvLz8w0/z8mgiBVjTJkyBc8991yn7e+88w4OHTp0Sm3n5eVh2LBhp9RGn+GULek4oD8a7yNHjhTPPvusqK+v1/T1ww8/FDNnzhQ2m+2Uz3HttddGdX8uvPDCuBvvKtzQA/zxj3/sNtvV6XRq1riur6/H5MmTsXv3btTV1fVVF/sNFLF6gKysLAwZMiSqY9944w1s27YNzc3Nxnaqn0IRyyBMmzYN4XAYN998M0TiLQl5ylDGu4GYPn06jh49iltvvRVnnnlmvLvTp1DE6gHWrFmDpqamqI83m83IzMzEU089hfXr12PixIkG9q5/QanCHmDFihXYv38/HA5Ht8c89dRTKC4uhtWqvbVnnHEGXn75Zdxwww349NNPDevjSy+9hK+++sqw9qNGb9z9eKM/hhvo43A4xKRJk0RtbW2XfW9oaBB2u73H7VosFnH//fef9L5Mnz7d8Gvs96vY9xb9mVj0ueqqq0RDQ0Onvvv9/l4RKy0tTYRCoW7vSTgcFjNnzuyTa1PEivOnvLxctLS0yH4HAgExY8YMYTKZetxWamqq8Hq9Xd6PYDAobrnlFmE2mxWxTgWJQiwAYvLkyWL16tVi9erV4qc//ekptfXd7363073w+Xzi5ptv7tNrUpH3foA1a9ZgzZo1MWkrEokAaC+8cuTIEQBAQ0MDli9fHpP2YwmTEIkXvfN6vXIywUCCy+VCXl4e9u7di7a2trj1o7Gx8aRzO5XESiA0Nzdj165d8e5GVFABUgVDoIilYAgUsRQMgSKWgiFQxFIwBD0m1oYNG3DNNdcgPz8fJpMJq1ev1uwXQuDBBx9EXl4enE4nysrKsHPnTs0x9fX1mDp1KlJTU+HxeHDTTTf1KGtAof+jx8Rqbm7GmDFjsGTJki73P/LII3jiiSfwzDPPoLKyEi6XC+Xl5QgEAvKYqVOnYtu2bVi3bh3efPNNbNiwAbNmzer9VSj0P5zK0AoA8dprr8nvkUhE5ObmikcffVRuo9H85cuXCyGE2L59uwAgPvnkE3nMmjVrhMlkEgcOHIjqvIk0pHM6fvp8JnR1dTVqa2s1xWvT0tJQWloqi9du2rQJHo8H48aNk8eUlZXBbDajsrKyy3ZVcdvEQ0yJRQVq+WwV+k77amtrMWjQIM1+q9WKjIyMbgvcquK2iYeE8ArnzZuHxsZG+ampqYl3lxROgpgSiwrU6pcw4cVrc3Nz5cg8IRQKob6+vtsCt6q4beIhpsQqLi5Gbm6upnit1+tFZWWlLF47YcIENDQ0YMuWLfKY999/H5FIBKWlpbHsjkI8EaUDKOHz+URVVZWoqqoSAMRjjz0mqqqqxN69e4UQQvz+978XHo9HvP766+KLL74Q1113nSguLhZ+v1+2MWnSJHHeeeeJyspKsXHjRjF8+HBRUVERdR+UV9j/vcIeE2v9+vVdnmzatGlCiPaQwwMPPCBycnKE3W4XV1xxhfj66681bdTV1YmKigrhdrtFamqqmDFjhqau5smgiNX/iaUS/RR6jGgS/RLCK1RIPChiKRgCRSwFQ6CIpWAIFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBKSWAk4bn5aIZr7n5DEGogrPfQn+Hy+kx6TkGWMMjIyALSv/6fSZ7qH1+tFQUEBampqYpLOLYSAz+eLanWxhCSW2dwuaNPS0lT+exSI5TyBaF/khFSFCv0filgKhiAhiWW32zF//vxOay4raBHP+5SQOe8K/R8JKbEU+j8UsRQMgSKWgiFQxFIwBAlJrCVLlmDIkCFwOBwoLS3Fv/71r3h3qc+QKKU6E45YK1euxNy5czF//nxs3boVY8aMQXl5eacKNqcrEqZUZ9QFE/oJxo8fL+bMmSO/h8NhkZ+fLxYuXBjHXsUHQHxKdUaDhJJYra2t2LJli6YUpdlsRllZmSxFOZBhVKnO3iChiHXs2DGEw+ETlqIcyDCqVGdvkFDEUkgcJBSxsrKyYLFYTliKciDDqFKdvUFCEctms6GkpERTijISieC9996TpSgHMvpVqc6YuQF9hBUrVgi73S6WLl0qtm/fLmbNmiU8Ho+ora2Nd9f6BP2hVGc0SDhiCSHE4sWLRWFhobDZbGL8+PFi8+bN8e5Sn6E/lOqMBiptRsEQJJSNpZA4UMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxFAyBIpaCIVDEUjAEilgKhkARS8EQKGIpGAJFLAVDoIilYAgUsRQMgSKWgiFQxFIwBIpYDCaTKarPBx98EO+uavDPf/4TCxYsQENDQ7y7IpGQ5biNwssvv6z5/tJLL2HdunWdtp9zzjl92a2T4p///CceeughTJ8+HR6PJ97dAaCIpcGPf/xjzffNmzdj3bp1nbb3BkIIBAIBOJ3OU24rEaBUYQ/x4osv4vLLL8egQYNgt9sxcuRIPP30052OGzJkCK6++mq88847GDduHJxOJ5599lkAwN69e3HttdfC5XJh0KBBuPvuu/HOO+90qWYrKysxadIkpKWlITk5GZdccgk+/vhjuX/BggW47777ALRPWCV1/c033xh2D6KBklg9xNNPP41Ro0bh2muvhdVqxRtvvIHbbrsNkUgEc+bM0Rz79ddfo6KiArNnz8bMmTNx9tlno7m5GZdffjkOHTqEO++8E7m5uXjllVewfv36Tud6//33MXnyZJSUlGD+/Pkwm82S2B999BHGjx+PKVOm4D//+Q+WL1+Oxx9/HFlZWQCA7OzsPrkf3SKmsxRPM8yZM0fob1FLS0un48rLy8XQoUM124qKigQAsXbtWs32P/7xjwKAWL16tdzm9/vFiBEjBACxfv16IUT7xNPhw4eL8vJyEYlENOcvLi4W3/72t+W2Rx99VAAQ1dXVvb3UmEOpwh6C20iNjY04duwYLrnkEuzZsweNjY2aY4uLi1FeXq7ZtnbtWgwePBjXXnut3OZwODBz5kzNcZ999hl27tyJH/3oR6irq8OxY8dw7NgxNDc344orrsCGDRsQiUQMuMLYQKnCHuLjjz/G/PnzsWnTJrS0tGj2NTY2ahYxKi4u7vT7vXv3YtiwYTCZTJrtZ555puY71Q2dNm1at31pbGxEenp6j6+hL6CI1QPs3r0bV1xxBUaMGIHHHnsMBQUFsNlsePvtt/H44493kiCn4gFSW48++ijGjh3b5TFut7vX7RsNRawe4I033kAwGMQ//vEPFBYWyu1dGd7doaioCNu3b4cQQiO1du3apTlu2LBhANqXhOOlH7uCXvr1BygbqwewWCwAtEvXNjY24sUXX4y6jfLychw4cAD/+Mc/5LZAIIA///nPmuNKSkowbNgw/OEPf+iyVPbRo0fl/y6XCwBU5D1R8Z3vfAc2mw3XXHMNZs+ejaamJvz5z3/GoEGDcOjQoajamD17Np588klUVFTgzjvvRF5eHpYtWwaHwwGgQ/qYzWY8//zzmDx5MkaNGoUZM2Zg8ODBOHDgANavX4/U1FS88cYbANpJCAC/+tWvcOONNyIpKQnXXHONJFxcEG+3tD+jq3DDP/7xDzF69GjhcDjEkCFDxP/8z/+Iv/zlL53c/aKiInHVVVd12e6ePXvEVVddJZxOp8jOzhb33HOP+Nvf/iYAdKr1VVVVJaZMmSIyMzOF3W4XRUVF4gc/+IF47733NMf95je/EYMHDxZms7lfhB5Ufax+gkWLFuHuu+/G/v37MXjw4Hh355ShiBUH+P1+jccYCARw3nnnIRwO4z//+U8cexY7KBsrDpgyZQoKCwsxduxYNDY24q9//St27NiBZcuWxbtrMYMiVhxQXl6O559/HsuWLUM4HMbIkSOxYsUK/PCHP4x312KHeBp4Tz75pCgqKhJ2u12MHz9eVFZWxrM7CjFE3OJYA30Vr9MdcTPeS0tLccEFF+DJJ58E0D6EUVBQgDvuuAP3339/PLqkEEPExcaiVbzmzZsnt/VkFa9IJIKDBw8iJSWlXw5nnK4QQsDn8yE/Px9m84mVXVyIdaJVvHbs2NHp+GAwiGAwKL8fOHAAI0eONLyfCl2jpqYGZ5xxxgmPSYixwoULFyItLU1+FKnii5SUlJMeExdi9XQVr3nz5qGxsVF+ampq+qqrCl0gGvMjLsTq6Spedrsdqampmo9CP0e84hynsopXY2NjlwsVqU/ffBobG0/6jOIaIO3tKl6KWP2fWAk5CO31ejW55Qp9i8bGxpOaIwnhFSokHhSxFAyBIpaCIVDEUjAEilgKhkARS8EQKGIpGAJFLAVDoIilYAgUsRQMgSKWgiFQxFIwBIpYCoZAEUvBEChiKRgCRSwFQ6CIpWAIFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxFAxBzIm1YMGCTqu+jxgxQu4PBAKYM2cOMjMz4Xa7cf3113cqZ6SQ+DBEYo0aNQqHDh2Sn40bN8p9d999N9544w2sWrUKH374IQ4ePIgpU6YY0Q2FeOLU6sV0xvz588WYMWO63NfQ0CCSkpLEqlWr5LavvvpKABCbNm2K+hyq2kz/rzZjiMTauXMn8vPzMXToUEydOhX79u0DAGzZsgVtbW2a9fdGjBiBwsLCExa1DQaD8Hq9mo9C/0bMiVVaWoqlS5di7dq1ePrpp1FdXY2LLroIPp8PtbW1sNls8Hg8mt/k5OSgtra22zb1NUgLCgpi3W2FGCPmVZMnT54s/x89ejRKS0tRVFSEV199tddL2c6bNw9z586V371eryJXP4fh4QaPx4OzzjoLu3btQm5uLlpbWzutBNpdUVuCqkGaeDCcWE1NTdi9ezfy8vJQUlKCpKQkTVHbr7/+Gvv27euyqK1CAiNqVyxK3HPPPeKDDz4Q1dXV4uOPPxZlZWUiKytLHDlyRAghxC233CIKCwvF+++/Lz799FMxYcIEMWHChB6dQ3mF/d8rjLmNtX//flRUVKCurg7Z2dmYOHEiNm/ejOzsbADA448/DrPZjOuvvx7BYBDl5eV46qmnYt2N0xoVFRWyBuvx48excuXKOPeoC/RKLMUZA11i7dy5U94Lr9cr/v73v4uSkpJ+JbHUWGEfYfTo0Rg7dizcbvcptVNQUAC73S6/p6Sk4Hvf+x7efffd+K5ar4MiloGYM2cO7r//fpx99tn47ne/i0WLFuGCCy44pTbvv//+LkMtqampuPHGG0+p7ZjCSJVlFBJBFd53330iEAgIIYT497//LZ5//nkxdOhQ4fF4TqndJUuWdHtfNm3aJKxWq1KFpysGDRqEyy+/XKqsc889F//93/+NGTNmdIrhxRKlpaX4xS9+YVj7PYEilgG47LLLMGnSJM02k8kEq9XY5SFNJhMsFouh54gWilgxhslk6tJA37FjB5544gnDz3/rrbdi9OjRhp/nZFDEijGmTJmC5557rtP2d955B4cOHTqltvPy8jBs2LATHpObm4vk5ORTOk9MEDOLug/RH433kSNHimeffVbU19dr+vrhhx+KmTNnCpvNdsrnuPbaa6O6PxdeeGHcjfe4rAmdqMjMzMQ777yDzMzMTvucTqdmjev6+npMnjwZu3fvRl1dXV92s19AEasHeOyxx1BSUhLVsffeey/+9a9/xfT8e/fuxVdffYVzzjknpu0aAWVjGYTbbrsNixcvjmr95Gjx+eefY/369TFrz0goYhmEcePG4bbbbsPRo0dx66234swzz4x3l/oUilg9wJo1a9DU1BT18WazGZmZmXjqqaewfv16TJw40cDeteOf//wnDh48aPh5TgZlY/UAK1aswP79++FwOLo95qmnnkJxcXGnYOgZZ5yBl19+GTfccAM+/fRTQ/onhMDbb78tJ6/EFb1x9+ON/hhuoI/D4RCTJk0StbW1Xfa9oaFB2O32XrWdmZkpXnvttW7vy+bNm/vNWKEilkGfq666SjQ0NHTqu9/v7zWxThTHCofDYubMmX1ybWoQOo5466238MMf/hB+v19uCwaDuO2229Da2hrTc7W2tmLOnDl44YUXYtruKeGUREeckAgSiz6TJ08Wq1evFqtXrxY//elPT6mtriSWz+cTN998c59eUzQSyySEED2jYvzh9XplzvdAwpAhQzTzNgGgoaEBy5cv79N+NDY2nnQKniKWQo8RDbGUjaVgCBSxFAyBIpaCIVDEUjAEilgKhkARS8EQ9JhYGzZswDXXXIP8/HyYTCasXr1as18IgQcffBB5eXlwOp0oKyvDzp07NcfU19dj6tSpSE1NhcfjwU033dSjrAGF/o8eE6u5uRljxozBkiVLutz/yCOP4IknnsAzzzyDyspKuFwulJeXIxAIyGOmTp2Kbdu2Yd26dXjzzTexYcMGzJo1q/dXodD/cCpDKwA0o+2RSETk5uaKRx99VG6j0fzly5cLIYTYvn27ACA++eQTecyaNWuEyWQSBw4ciOq8iTSkczp++nwQurq6GrW1tZritWlpaSgtLZXFazdt2gSPx4Nx48bJY8rKymA2m1FZWdllu6q4beIhpsSiArV8tgp9p321tbUYNGiQZr/VakVGRka3BW5VcdvEQ0J4hfPmzUNjY6P81NTUxLtLCidBTIlFBWr1S5jw4rW5ubk4cuSIZn8oFEJ9fX23BW5VcdvEQ0yJVVxcjNzcXE3xWq/Xi8rKSlm8dsKECWhoaMCWLVvkMe+//z4ikQhKS0tj2R2FeCJKB1DC5/OJqqoqUVVVJQCIxx57TFRVVYm9e/cKIYT4/e9/Lzwej3j99dfFF198Ia677jpRXFws/H6/bGPSpEnivPPOE5WVlWLjxo1i+PDhoqKiIuo+KK+w/3uFPSbW+vXruzzZtGnThBDtIYcHHnhA5OTkCLvdLq644grx9ddfa9qoq6sTFRUVwu12i9TUVDFjxgzh8/mi7oMiVv8nlkr0U+gxVKKfQtygiKVgCBSxFAyBIpaCIVDEUjAEilgKhkARS8EQKGIpGAJFLAVDoIilYAgUsRQMgSKWgiFISGIl4Lj5aYVo7n9CEmsgrvTQn+Dz+U56TEJWTc7IyAAA7Nu3T6XPnABerxcFBQWoqamJSTq3EAI+nw/5+fknPTYhiWU2twvatLQ0lf8eBWI5TyDaFzkhVaFC/4ciloIhSEhi2e12zJ8/X665rNA14nmfEjLnXaH/IyEllkL/hyKWgiFQxFIwBIpYCoYgIYm1ZMkSDBkyBA6HA6WlpTFfe7k/I1FKdSYcsVauXIm5c+di/vz52Lp1K8aMGYPy8vJOFWxOVyRMqc6oCyb0E4wfP17MmTNHfg+HwyI/P18sXLgwjr2KD4D4lOqMBgklsVpbW7FlyxZNKUqz2YyysjJZinIgw6hSnb1BQhHr2LFjCIfDJyxFOZBhVKnO3iChiKWQOEgoYmVlZcFisZywFOVAhlGlOnuDhCKWzWZDSUmJphRlJBLBe++9J0tRDmT0q1KdMXMD+ggrVqwQdrtdLF26VGzfvl3MmjVLeDweUVtbG++u9Qn6Q6nOaJBwxBJCiMWLF4vCwkJhs9nE+PHjxebNm+PdpT5DfyjVGQ1U2oyCIUgoG0shcaCIpWAIFLEUDIEiloIhUMRSMASKWAqGQBFLwRAoYikYAkUsBUOgiKVgCBSxFAyBIpaCIfj/ABl2DusDCdgKAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Train IoU: 0.9526, Best Train Dice: 0.9663 Best Train Acc: 0.9990\n"
     ]
    }
   ],
   "source": [
    "# Using the best model to evaluate on the train set\n",
    "\n",
    "model = smp.UnetPlusPlus(encoder_name='resnet34', encoder_depth=5, encoder_weights=\"imagenet\", decoder_use_batchnorm=True, decoder_channels=(256, 128, 64, 32, 16), decoder_attention_type=None, in_channels=1, classes=1, activation=None, aux_params=None)\n",
    "\n",
    "# 设置设备\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "model = model.to(device)\n",
    "\n",
    "test_path = \"./dataset/Spineweb_dataset15/train/\"\n",
    "\n",
    "transform = transforms.Compose([\n",
    "    # transforms.ToPILImage(),  # 转换为 PIL 图像对象\n",
    "    transforms.Resize((128, 128)),  # 调整大小为 128x128\n",
    "    transforms.ToTensor()  # 转换为张量\n",
    "])\n",
    "\n",
    "# 加载已保存的模型参数\n",
    "checkpoint_path = \"./model/UNet++_checkpoint.pth\"  # 模型文件路径\n",
    "checkpoint = torch.load(checkpoint_path)\n",
    "model.load_state_dict(checkpoint)\n",
    "\n",
    "# 设置模型为推理模式\n",
    "model.eval()\n",
    "\n",
    "batch_size = 64\n",
    "\n",
    "confidence = 0.9\n",
    "\n",
    "testset = SpineWeb15(data_path=test_path, transform1=transform, transform2=transform)\n",
    "test_loader = DataLoader(testset, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "# data_iter = iter(test_loader)\n",
    "# images, labels = next(data_iter)\n",
    "# print(\"加载的样本数量:\", images.shape[0])\n",
    "\n",
    "total_iou_test = 0.0\n",
    "total_dice_test = 0.0\n",
    "total_acc_test = 0.0\n",
    "total_correct_pixels_test = 0\n",
    "total_pixels_test = 0\n",
    "\n",
    "# test set results\n",
    "with torch.no_grad():\n",
    "    for inputs, targets in test_loader:\n",
    "        inputs = inputs.to(device)\n",
    "        targets = targets.to(device)\n",
    "        targets = targets * 255\n",
    "        targets[targets <= poster_threshold] = 0\n",
    "        targets[targets > poster_threshold] = 1\n",
    "        outputs = model(inputs)\n",
    "        # squeeze the dimension that pytorch add for channels\n",
    "        outputs = outputs.squeeze(1)     # tensor size ([16,128,128])\n",
    "        targets = targets.squeeze(1)\n",
    "        outputs = torch.sigmoid(outputs)\n",
    "        # process the result, bigger than 0.5 is reliable\n",
    "        predicted = (outputs > confidence).float()     # tensor size ([16,128,128])\n",
    "        intersection = torch.logical_and(predicted, targets).sum((1, 2))\n",
    "        union = torch.logical_or(predicted, targets).sum((1, 2))\n",
    "        iou = (intersection / (union + 1e-7))  # 平均每个样本的IoU\n",
    "        dice = (2 * intersection / (predicted.sum((1, 2)) + targets.sum((1, 2))+1e-7))  # 平均每个样本的Dice Coefficient\n",
    "\n",
    "\n",
    "        batch_correct_pixels,batch_total_pixels = calculate_pixel_accuracy(predicted, targets)\n",
    "        total_iou_test += torch.sum(iou).item()\n",
    "        total_dice_test += torch.sum(dice).item()\n",
    "        total_correct_pixels_test += batch_correct_pixels\n",
    "        total_pixels_test += batch_total_pixels\n",
    "        \n",
    "\n",
    "inputs_npimage = inputs.squeeze(1)[0].detach().cpu().numpy()\n",
    "predicted_npimage = predicted[0].detach().cpu().numpy()\n",
    "targets_npimage = targets[0].detach().cpu().numpy()\n",
    "fig, axes = plt.subplots(3, 1)\n",
    "axes[0].imshow(inputs_npimage, cmap='gray')\n",
    "axes[0].set_title('Original')\n",
    "axes[1].imshow(predicted_npimage, cmap='gray')\n",
    "axes[1].set_title('Predicted')\n",
    "axes[2].imshow(targets_npimage, cmap='gray')\n",
    "axes[2].set_title('Target')\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "\n",
    "\n",
    "test_iou = total_iou_test / len(testset)\n",
    "test_dice = total_dice_test / len(testset)\n",
    "test_acc = total_correct_pixels_test / total_pixels_test\n",
    "\n",
    "# output the result, the result may vary a little due to the resize function in transform\n",
    "print(f\"Best Train IoU: {test_iou:.4f}, Best Train Dice: {test_dice:.4f} Best Train Acc: {test_acc:.4f}\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "dataEngineering",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.4"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}