Add coronary artery segmentation bundle (CoronSegmentator) #761

Foxconn-bgroup-aidd · 2025-07-16T09:19:56Z

proposed Model: CoronSegmentator

This PR adds a new coronary artery segmentation model named coronsegmentator under models/coronsegmentator/.

Coronsegmentator is an automated pipeline that performs dual-task segmentation on cardiac CT images, focusing on both whole-heart and coronary artery structures. It integrates MONAI’s Auto3DSeg for general cardiac segmentation and a custom nnU-Net model for detailed coronary artery segmentation. The pipeline further converts segmentation results (in STL format) into USD files for downstream 3D visualization and digital twin simulation using NVIDIA Omniverse.

for more information, see https://pre-commit.ci

yiheng-wang-nv · 2025-07-22T02:01:17Z

models/CoronSegmentator/.gitignore

@@ -0,0 +1,7 @@
+Input/


Hi @Foxconn-bgroup-aidd , can we remove this file?
A part of files are already ignored (set in https://github.com/Project-MONAI/model-zoo/blob/dev/.gitignore), and for customized folder like Input/ and Output/, I think we can remove them directly. What do you think?

There is no problem removing them. Thanks.

yiheng-wang-nv · 2025-07-22T02:05:57Z

models/CoronSegmentator/configs/inference.json

@@ -0,0 +1,17 @@
+{
+    "bundle_root": "./",


could you prepare a unit test script?
It's helpful to check if (1) large files are downloadable (2) inference pipeline works (3) dependencies mentioned in metadata are correct and enough to run the script.

You can refer to this example to prepare the script: https://github.com/Project-MONAI/model-zoo/blob/dev/ci/unit_tests/test_spleen_ct_segmentation.py

@yiheng-wang-nv, I prepare the script refer to above url. but when I try to run the script , I got the error: ModuleNotFoundError: Cannot locate class or function path: 'scripts.heart_digital_twin.CoroSegmentator_Pipeline'. It seems to can't access the custom class is defined in inference.json. I put the script in ci/unitest/ folder and run the script at model_zoo root directory. How can I solve the problem? Thanks for answer. Following is revised inference.json:

{ "imports": ["$import glob"], "bundle_root": "./", "ckpt_dir": "$@bundle_root + 'models'", "dataset_dir": "", "data": "$list(sorted(glob.glob(@dataset_dir + '/*.nii.gz')))", "output_ext": ".usd", "output_dir": "$@bundle_root + 'Output'", "device": "$torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')", "inferer": { "_target_": "scripts.heart_digital_twin.CoroSegmentator_Pipeline", "input_params": { "inputFile": "$@data[0]", "outputDir": "$@output_dir" } }, "dataset": {}, "evaluator": {}, "network_def": {}, "run": ["[email protected]()"] }

Hi @Foxconn-bgroup-aidd, please try create an init.py file in the scripts directory to make it as Python packages and also put the scripts folder just under you bundle instead of under the unittest folder. Thanks.

@KumoLiu I add the path of bundle into the sys.path and solve it, Thanks!

KumoLiu

Hi @Foxconn-bgroup-aidd, thanks for the contributions! Leave several comments inline.

KumoLiu · 2025-07-23T03:03:54Z

models/CoronSegmentator/configs/metadata.json

@@ -0,0 +1,88 @@
+{
+    "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",


Hi @Foxconn-bgroup-aidd, could you please utilize the latest schema available at the following link:
"https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20240725.json"

Based on the content, it appears that you are already using the most recent version, as it includes "required_packages_version." Thanks.

KumoLiu · 2025-07-23T03:05:00Z

models/CoronSegmentator/configs/metadata.json

+        "monai": " 1.4.0",
+        "nnunetv2": "2.6.0",
+        "nibabel": "5.3.2",
+        "numpy": "2.3.1",


Could you please double check the monai version you are using, the one mentioned above is "1.4.0"? Thanks!

Suggested change

"monai": " 1.4.0",

"nnunetv2": "2.6.0",

"nibabel": "5.3.2",

"numpy": "2.3.1",

"nnunetv2": "2.6.0",

"nibabel": "5.3.2",

KumoLiu · 2025-07-23T03:07:33Z

models/CoronSegmentator/configs/metadata.json

+                    "0": "Background",
+                    "1": "Coronary Artery"


It seems the "num_channels" is only one for the outputs.

KumoLiu · 2025-07-23T03:09:32Z

models/CoronSegmentator/configs/metadata.json

+    "optional_packages_version": {
+        "monai": "1.3.2",
+        "torch": "2.7.0+cu126",
+        "numpy": "2.3.1"
+    }


I guess you can remove this part, the "optional_packages_version" is no longer a required one in the latest schema. thanks.

The version of monai used in our project is “1.4.0”. “optional_packages_version" will be removed.

KumoLiu · 2025-07-23T03:10:38Z

models/CoronSegmentator/docs/README.md

+The ImageCAS dataset is publicly available at:
+https://github.com/XiaoweiXu/ImageCAS-A-Large-Scale-Dataset-and-Benchmark-for-Coronary-Artery-Segmentation-based-on-CT.git''


Would be better to mention the dataset link in the "Data" section, what do you think?

models/CoronSegmentator/requirements.txt

KumoLiu · 2025-07-23T03:15:24Z

models/CoronSegmentator/scripts/cardiac_segmentation/cardiac_seg.py

+            # try:
+            # Step 1: Convert input to NRRD
+            input_nrrd = self._convert_to_nrrd(self.input_path, input_tmp_dir)
+            # Step 2: Run segmentation
+            seg_nrrd = self._run_segmentation(input_nrrd, output_tmp_dir)
+            # Step 3: Process results
+            self._process_segmentation_results(seg_nrrd, output_tmp_dir)
+            # except Exception as e:
+            #     logger.error(f"Processing failed: {str(e)}")
+            #     raise


Could you please remove the comment code here or perhaps include the try-catch logic is also make sense.

KumoLiu · 2025-07-23T03:16:55Z

models/CoronSegmentator/scripts/usd_design/usd_mesh_creator.py

+                    base_center = Gf.Vec3d(
+                        (bbox_min[0] + bbox_max[0]) / 2, bbox_min[1], (bbox_min[2] + bbox_max[2]) / 2
+                    )
+                    # 计算平移量，使得基底中心移动到原点


Please check and remove all Chinese comments inline. Thanks.

KumoLiu · 2025-07-23T03:18:59Z

models/CoronSegmentator/configs/inference.json

+    "output_ext": ".usd",
+    "input_file": "Input/E_CT_0000.nii.gz",
+    "infer": {
+        "_target_": "scripts.heart_digital_twin.CoroSegmentator_Pipeline",


As @yiheng-wang-nv mentioned above, please include the unit test to ensure the bundle works as expected.
I didn't see this file under the scripts folder.

…and metadata.json

…/model-zoo into 760-CoronSegmentator

for more information, see https://pre-commit.ci

KumoLiu · 2025-07-25T14:11:02Z

ci/unit_tests/test_brain_image_synthesis_latent_diffusion_model.py

Did you mis-delete this test file?

We have fixed the issue of incorrectly deleting test files.

…/model-zoo into 760-CoronSegmentator

…n_model.py

for more information, see https://pre-commit.ci

Foxconn-bgroup-aidd and others added 2 commits July 16, 2025 17:15

Add coronary artery segmentation bundle (CoronSegmentator)

58b0998

[pre-commit.ci] auto fixes from pre-commit.com hooks

2e858aa

for more information, see https://pre-commit.ci

Nic-Ma requested review from yiheng-wang-nv and KumoLiu July 21, 2025 16:22

yiheng-wang-nv reviewed Jul 22, 2025

View reviewed changes

KumoLiu reviewed Jul 23, 2025

View reviewed changes

ericspod mentioned this pull request Jul 24, 2025

Add Coronary Artery Segmentator Model #760

Open

Foxconn-bgroup-aidd and others added 3 commits July 25, 2025 14:03

1. Add unit_test scripts 2. Add download.py 3. Update large_file.yml …

23ca47c

…and metadata.json

Merge branch '760-CoronSegmentator' of github.com:Foxconn-bgroup-aidd…

9d1626f

…/model-zoo into 760-CoronSegmentator

[pre-commit.ci] auto fixes from pre-commit.com hooks

c8b3ad0

for more information, see https://pre-commit.ci

KumoLiu reviewed Jul 25, 2025

View reviewed changes

Foxconn-bgroup-aidd and others added 4 commits July 28, 2025 08:55

Merge branch '760-CoronSegmentator' of github.com:Foxconn-bgroup-aidd…

009a68f

…/model-zoo into 760-CoronSegmentator

Restore deleted unit test: test_brain_image_synthesis_latent_diffusio…

8f3bc68

…n_model.py

Restore deleted test_brain_image_synthesis_latent_diffusion_model.py

0074aaf

[pre-commit.ci] auto fixes from pre-commit.com hooks

68cc018

for more information, see https://pre-commit.ci

		@@ -0,0 +1,88 @@
		{
		"schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",

		The ImageCAS dataset is publicly available at:
		https://github.com/XiaoweiXu/ImageCAS-A-Large-Scale-Dataset-and-Benchmark-for-Coronary-Artery-Segmentation-based-on-CT.git''

Add coronary artery segmentation bundle (CoronSegmentator) #761

Are you sure you want to change the base?

Add coronary artery segmentation bundle (CoronSegmentator) #761

Uh oh!

Conversation

Foxconn-bgroup-aidd commented Jul 16, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tylin7111095022 Jul 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

KumoLiu left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Foxconn-bgroup-aidd Jul 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

tylin7111095022 Jul 23, 2025 •

edited

Loading

Foxconn-bgroup-aidd Jul 23, 2025 •

edited

Loading