Niftymic killed with Docker image

[yuezhang9842]

Dear all,

Thank you for this great package.
When I run the niftymic_reconstruct_volume, the program is always “Killed”.
I have set the memory to the maximum memory（16G + 4G swap）, but still not enough. When the memory usage of container reaches 16G, the program will be "killed". I wonder how can I solve this problem?

`# niftymic_reconstruct_volume
--filenames /www/cor/T2_bias.nii.gz /www/sag/T2_bias.nii.gz /www/tra/T2_bias.nii.gz
--filenames-masks /www/cor/T2_label.nii.gz /www/sag/T2_label.nii.gz /www/tra/T2_label.nii.gz
--output /www/srr.nii.gz \

---

*** Configuration: ***
--- alpha: 0.015
--- alpha_first: 0.2
--- bias_field_correction: 0
--- boundary_stacks:
10
10
0
--- config: None
--- dilation_radius: 3
--- extra_frame_target: 10
--- filenames:
/www/cor/T2_bias.nii.gz
/www/sag/T2_bias.nii.gz
/www/tra/T2_bias.nii.gz
--- filenames_masks:
/www/cor/T2_label.nii.gz
/www/sag/T2_label.nii.gz
/www/tra/T2_label.nii.gz
--- intensity_correction: 1
--- interleave: 3
--- isotropic_resolution: 1
--- iter_max: 10
--- iter_max_first: 5
--- iterations: 15
--- log_config: 1
--- metric: Correlation
--- metric_radius: 10
--- multiresolution: 0
--- outlier_rejection: 1
--- output: /www/srr.nii.gz
--- reconstruction_type: TK1L2
--- reference: None
--- reference_mask: None
--- s2v_hierarchical: False
--- sda: False
--- search_angle: 45
--- shrink_factors:
3
2
1
--- sigma: 1
--- slice_thicknesses: None
--- smoothing_sigmas:
1.5
1
0
--- subfolder_motion_correction: motion_correction
--- suffix_mask: _mask
--- target_stack: None
--- threshold: 0.8
--- threshold_first: 0.5
--- transforms_history: 0
--- two_step_cycles: 3
--- use_masks_srr: 0
--- v2v_method: RegAladin
--- v2v_robust: False
--- verbose: 0
--- viewer: itksnap
--- write_motion_correction: 1

NiftyMIC version: 0.9

---

--- File written to '/www/config_reconstruct_volume_20220604-094944.json'.

---

*** Read Data ***
--- 3 input stacks read for further processing

---

*** Data Preprocessing ***
--- 3 stacks were loaded for data preprocessing
--- Crop stacks to their masks

---

*** Volume-to-Volume Registration ***
--- Volume-to-Volume Registration -- Stack 1/3
220604-09:50:09,211 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0
220604-09:50:09,343 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0
--- Volume-to-Volume Registration -- Stack 2/3
220604-09:50:47,548 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0
220604-09:50:47,655 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0
--- Volume-to-Volume Registration -- Stack 3/3
220604-09:51:32,424 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0
220604-09:51:32,529 nipype.interface WARNING:
version_from_command member of CommandLine was Deprecated in nipype-1.0.0 and deleted in 1.1.0

---

*** Intensity Correction ***
--- Stack 1 (T2_bias): Reference image. Skipped.
--- Stack 2 (T2_bias): Intensity Correction ... done (c1 = 0.908708)
--- Stack 3 (T2_bias): Intensity Correction ... done (c1 = 0.949986)

---

*** Reconstruction Space Generation ***
--- Isotropic reconstruction space with 1 mm resolution is created
--- Isotropic reconstruction space is centered around joint stack masks.

---

*** First Estimate of HR Volume ***

*** SDA Approximation Image ***
--- Chosen SDA approach: Shepard-YVV
--- Smoothing parameter sigma = [1. 1. 1.]
--- Stack 1/3
--- Stack 2/3
--- Stack 3/3
--- Required computational time: 0:00:19.953441

*** SDA Approximation Image Mask ***
--- Chosen SDA approach: Shepard-YVV
--- Smoothing parameter sigma = [1. 1. 1.]
--- Stack 1/3
--- Stack 2/3
--- Stack 3/3
--- Required computational time: 0:00:04.960164

---

*** Two-step S2V-Registration and SRR Reconstruction ***

---

*** Slice-to-Volume Registration ***
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 1/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 2/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 3/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 4/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 5/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 6/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 7/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 8/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 9/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 10/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 11/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 12/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 13/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 14/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 15/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 16/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 17/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 18/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 1/3 (T2_bias) -- Slice 19/19
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 1/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 2/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 3/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 4/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 5/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 6/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 7/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 8/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 9/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 10/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 11/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 12/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 13/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 14/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 15/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 16/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 2/3 (T2_bias) -- Slice 17/17
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 3/3 (T2_bias) -- Slice 1/21
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 3/3 (T2_bias) -- Slice 2/21
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 3/3 (T2_bias) -- Slice 3/21
--- Cycle 1/3: Slice-to-Volume Registration -- Stack 3/3 (T2_bias) -- Slice 4/21
Killed
`

[renbem]

Hi, if it's killed because of not enough memory, there's not much I can do, unfortunately. Besides having more RAM available, you could create tighter masks or lower the resolution, e.g. set --isotropic-resolution to 1.5 or 2 mm. Both will reduce the memory requirements. The latter of course will reduce image quality but could be a good first step to check that you can get successful recontructions

[renbem]

One thing though, Docker may have less memory available than what you have on your system. Perhaps worth checking as well.

[yuezhang9842]

Hello. The “killed” problem has been solved, it seems that the ARM64 system used by Apple M1 is not suitable for docker. So I switched to Windows.
However I also wonder if I need to download the images of ITK_Niftymic and Simplereg to use Niftymic?

[yuezhang9842]

Meanwhile, I‘m having troubles with “Transformation to Template Space”. It seems that the registration was not successful.
Here is my code:

niftymic_register_image \
--fixed ./NiftyMIC/data/templates/STA33.nii.gz \
--fixed-mask ./NiftyMIC/data/templates/STA33_mask.nii.gz \
--moving /www1/results/results-0.01-t/srr.nii.gz \
--moving-mask /www1/results/results-0.01-t/srr_mask.nii.gz \
--dir-input-mc /www1/results/results-0.01-t/motion_correction \
--output /www1/results/results-0.01-t/registration-transform.txt
niftymic_reconstruct_volume_from_slices \
--filenames /www1/cor/T2_bias.nii.gz /www1/sag/T2_bias.nii.gz /www1/tra/T2_bias.nii.gz \
--filenames-masks /www1/cor/T2_label.nii.gz /www1/sag/T2_label.nii.gz /www1/tra/T2_label.nii.gz \
--output /www1/results/srr-0.01-t-mc.nii.gz \
--reconstruction-space ./NiftyMIC/data/templates/STA33.nii.gz \
--reconstruction-type TK1L2 \
--alpha 0.01
***************************************************************************************************
*** Configuration: ***
--- config: None
--- dir_input_mc: /www1/results/results-0.01-t/motion_correction
--- fixed: ./NiftyMIC/data/templates/STA33.nii.gz
--- fixed_mask: ./NiftyMIC/data/templates/STA33_mask.nii.gz
--- init_pca: False
--- initial_transform: None
--- log_config: 1
--- method: RegAladin
--- moving: /www1/results/results-0.01-t/srr.nii.gz
--- moving_mask: /www1/results/results-0.01-t/srr_mask.nii.gz
--- output: /www1/results/results-0.01-t/registration-transform.txt
--- verbose: 0

NiftyMIC version: 0.9
***************************************************************************************************

--- File written to '/www1/results/results-0.01-t/config_register_image_20220619-153337.json'.


***************************************************************************************************
*** Read Data ***


***************************************************************************************************
*** Registration ***
--- Run Registration (RegAladin) ... done
--- Registration transformation written to '/www1/results/results-0.01-t/registration-transform.txt'


***************************************************************************************************
*** Update Motion-Correction Transformations ***
--- 0 transformations written to '/www1/results/results-0.01-t/motion_correction'


***************************************************************************************************
*** Summary ***
register_image | Computational Time: 0:00:15.605149

And here is the results of SRR, But I don‘t know what is the problem causing such strange results. Hope to get your help！！！

[wangyunai]

I met similar problem after SSRI. Did you solve this issue ? Thanks.