Upload Model

.gitattributes

@@ -32,3 +32,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0/progress.png filter=lfs diff=lfs merge=lfs -text
+nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1/progress.png filter=lfs diff=lfs merge=lfs -text
+nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_2/progress.png filter=lfs diff=lfs merge=lfs -text
+nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_3/progress.png filter=lfs diff=lfs merge=lfs -text
+nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_4/progress.png filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,13 @@
+FROM nvidia/cuda:11.6.2-base-ubuntu20.04
+
+RUN apt update
+RUN apt upgrade -y
+RUN apt install -y python3 python3-pip
+RUN pip install torch torchvision torchaudio nnunet matplotlib
+
+ENV RESULTS_FOLDER="/root/results"
+
+COPY inference.sh inference.sh
+COPY nnUNet_trained_models /root/results
+
+RUN chmod +x inference.sh

README.md

@@ -1,3 +1,156 @@
 ---
 license: mit
 ---
+
+# Ascites Segmentation with nnUNet
+
+## Method 1: Run Inference using `nnunet_predict.py`
+
+1. Install the latest version of [nnUNet](https://github.com/MIC-DKFZ/nnUNet#installation) and [PyTorch](https://pytorch.org/get-started/locally/).
+
+```shell
+user@machine:~/ascites_segmentation$ pip install torch torchvision torchaudio nnunet matplotlib
+```
+
+2. Run inference with command:
+
+```shell
+user@machine:~/ascites_segmentation$ python nnunet_predict.py -i file_list.txt -t TMP_DIR -o OUTPUT_FOLDER -m /path/to/nnunet/model_weights
+```
+
+```shell
+usage: tmp.py [-h] [-i INPUT_LIST] -t TMP_FOLDER -o OUTPUT_FOLDER -m MODEL [-v]
+
+Inference using nnU-Net predict_from_folder Python API
+
+optional arguments:
+  -h, --help            show this help message and exit
+  -i INPUT_LIST, --input_list INPUT_LIST
+                        Input image file_list.txt
+  -t TMP_FOLDER, --tmp_folder TMP_FOLDER
+                        Temporary folder
+  -o OUTPUT_FOLDER, --output_folder OUTPUT_FOLDER
+                        Output Segmentation folder
+  -m MODEL, --model MODEL
+                        Trained Model
+  -v, --verbose         Verbose Output
+```
+
+
+
+N.B. 
+- `model_weights` folder should contain `fold0`, `fold1`, etc...
+- WARNING: the program will try to create file links first, but will fallback to filecopy if fails
+
+
+## Method 2: Run Inference using `nnUNet_predict` from shell
+
+1. Install the latest version of [nnUNet](https://github.com/MIC-DKFZ/nnUNet#installation) and [PyTorch](https://pytorch.org/get-started/locally/).
+
+```shell
+user@machine:~/ascites_segmentation$ pip install torch torchvision torchaudio nnunet matplotlib
+```
+
+2. Place checkpoints in directory tree:
+
+```shell
+user@machine:~/ascites_segmentation$ tree .
+.
+├── nnUNet_preprocessed
+├── nnUNet_raw_data_base
+└── nnUNet_trained_models
+    └── nnUNet
+        └── 3d_fullres
+            └── Task505_TCGA-OV
+                └── nnUNetTrainerV2__nnUNetPlansv2.1
+                    ├── fold_0
+                    │   ├── debug.json
+                    │   ├── model_final_checkpoint.model
+                    │   ├── model_final_checkpoint.model.pkl
+                    │   └── progress.png
+                    ├── fold_1
+                    │   ├── debug.json
+                    │   ├── model_final_checkpoint.model
+                    │   ├── model_final_checkpoint.model.pkl
+                    │   └── progress.png
+                    ├── fold_2
+                    │   ├── model_final_checkpoint.model
+                    │   ├── model_final_checkpoint.model.pkl
+                    │   └── progress.png
+                    ├── fold_3
+                    │   ├── model_final_checkpoint.model
+                    │   ├── model_final_checkpoint.model.pkl
+                    │   └── progress.png
+                    ├── fold_4
+                    │   ├── model_final_checkpoint.model
+                    │   ├── model_final_checkpoint.model.pkl
+                    │   └── progress.png
+                    └── plans.pkl
+```
+
+3. Setup environment variables so that nnU-Net knows where to find trained models: 
+
+```shell
+user@machine:~/ascites_segmentation$ export nnUNet_raw_data_base="/absolute/path/to/nnUNet_raw_data_base"
+user@machine:~/ascites_segmentation$ export nnUNet_preprocessed="/absolute/path/to/nnUNet_preprocessed"
+user@machine:~/ascites_segmentation$ export RESULTS_FOLDER="/absolute/path/to/nnUNet_trained_models"
+```
+
+4. Run inference with command:
+
+```shell
+user@machine:~/ascites_segmentation$ nnUNet_predict -i INPUT_FOLDER -o OUTPUT_FOLDER -t 505 -m 3d_fullres -f N --save_npz 
+```
+
+where:
+- `-i`: input folder of `.nii.gz` scans to predict. NB, filename needs to end with `_0000.nii.gz` to tell nnU-Net only one kind of modality
+- `-o`: output folder to store predicted segmentations, automatically created if not exist
+- `-t 505`: (do not change) Ascites pretrained model name
+- `-m 3d_fullres` (do not change) Ascites pretrained model name
+- `N`: Ascites pretrained model fold, can be `[0, 1, 2, 3, 4]`
+- `--save_npz`: save softmax scores, required for ensembling multiple folds 
+
+### Optional [Additional] Inference Steps
+
+a. use `nnUNet_find_best_configuration` to automatically get the inference commands needed to run the trained model on data.
+
+b. ensemble predictions using `nnUNet_ensemble` by running:
+
+```shell
+user@machine:~/ascites_segmentation$ nnUNet_ensemble -f FOLDER1 FOLDER2 ... -o OUTPUT_FOLDER -pp POSTPROCESSING_FILE
+```
+
+where `FOLDER1` and `FOLDER2` are predicted outputs by nnUNet (requires `--save_npz` when running `nnUNet_predict`).
+
+## Method 3: Docker Inference
+
+Requires `nvidia-docker` to be installed on the system ([Installation Guide](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html)). This `nnunet_docker` predicts ascites with all 5 trained folds and ensembles output to a single prediction.
+
+1. Build the `nnunet_docker` image from `Dockerfile`:
+
+```shell
+user@machine:~/ascites_segmentation$ sudo docker build -t nnunet_docker .
+```
+
+2. Run docker image on test volumes:
+
+```shell
+user@machine:~/ascites_segmentation$ sudo docker run \
+--gpus 0 \
+--volume /absolute/path/to/INPUT_FOLDER:/tmp/INPUT_FOLDER \
+--volume /absolute/path/to/OUTPUT_FOLDER:/tmp/OUTPUT_FOLDER \
+nnunet_docker /bin/sh inference.sh
+```
+
+
+
+- `--gpus` parameter:
+  - `0, 1, 2, ..., n` for integer number of GPUs 
+  - `all` for all available GPUs on the system
+  - `'"device=2,3"'` for specific GPU with ID
+
+- `--volume` parameter
+  - `/absolute/path/to/INPUT_FOLDER` and `/absolute/path/to/OUTPUT_FOLDER` folders on the host system needs to be specified
+  - `INPUT_FOLDER` contains all `.nii.gz` volumes to be predicted
+  - predicted results will be written to `OUTPUT_FOLDER`
+

inference.sh

@@ -0,0 +1,9 @@
+#! /bin/bash
+
+nnUNet_predict -i /tmp/INPUT_FOLDER/ -o output_fold_0 -t 505 -m 3d_fullres -f 0 --save_npz
+nnUNet_predict -i /tmp/INPUT_FOLDER/ -o output_fold_1 -t 505 -m 3d_fullres -f 1 --save_npz
+nnUNet_predict -i /tmp/INPUT_FOLDER/ -o output_fold_2 -t 505 -m 3d_fullres -f 2 --save_npz
+nnUNet_predict -i /tmp/INPUT_FOLDER/ -o output_fold_3 -t 505 -m 3d_fullres -f 3 --save_npz
+nnUNet_predict -i /tmp/INPUT_FOLDER/ -o output_fold_4 -t 505 -m 3d_fullres -f 4 --save_npz
+
+nnUNet_ensemble -f output_fold_0 output_fold_1 output_fold_2 output_fold_3 output_fold_4 -o /tmp/OUTPUT_FOLDER/

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0/debug.json

@@ -0,0 +1,87 @@
+{
+    "all_tr_losses": "[]",
+    "all_val_eval_metrics": "[]",
+    "all_val_losses": "[]",
+    "all_val_losses_tr_mode": "[]",
+    "also_val_in_tr_mode": "False",
+    "amp_grad_scaler": "None",
+    "base_num_features": "32",
+    "basic_generator_patch_size": "[ 32 263 263]",
+    "batch_dice": "True",
+    "batch_size": "2",
+    "best_MA_tr_loss_for_patience": "None",
+    "best_epoch_based_on_MA_tr_loss": "None",
+    "best_val_eval_criterion_MA": "None",
+    "classes": "[1]",
+    "conv_per_stage": "2",
+    "data_aug_params": "{'selected_data_channels': None, 'selected_seg_channels': [0], 'do_elastic': False, 'elastic_deform_alpha': (0.0, 200.0), 'elastic_deform_sigma': (9.0, 13.0), 'p_eldef': 0.2, 'do_scaling': True, 'scale_range': (0.7, 1.4), 'independent_scale_factor_for_each_axis': False, 'p_independent_scale_per_axis': 1, 'p_scale': 0.2, 'do_rotation': True, 'rotation_x': (-3.141592653589793, 3.141592653589793), 'rotation_y': (-0.5235987755982988, 0.5235987755982988), 'rotation_z': (-0.5235987755982988, 0.5235987755982988), 'rotation_p_per_axis': 1, 'p_rot': 0.2, 'random_crop': False, 'random_crop_dist_to_border': None, 'do_gamma': True, 'gamma_retain_stats': True, 'gamma_range': (0.7, 1.5), 'p_gamma': 0.3, 'do_mirror': True, 'mirror_axes': (0, 1, 2), 'dummy_2D': True, 'mask_was_used_for_normalization': OrderedDict([(0, False)]), 'border_mode_data': 'constant', 'all_segmentation_labels': None, 'move_last_seg_chanel_to_data': False, 'cascade_do_cascade_augmentations': False, 'cascade_random_binary_transform_p': 0.4, 'cascade_random_binary_transform_p_per_label': 1, 'cascade_random_binary_transform_size': (1, 8), 'cascade_remove_conn_comp_p': 0.2, 'cascade_remove_conn_comp_max_size_percent_threshold': 0.15, 'cascade_remove_conn_comp_fill_with_other_class_p': 0.0, 'do_additive_brightness': False, 'additive_brightness_p_per_sample': 0.15, 'additive_brightness_p_per_channel': 0.5, 'additive_brightness_mu': 0.0, 'additive_brightness_sigma': 0.1, 'num_threads': 12, 'num_cached_per_thread': 2, 'patch_size_for_spatialtransform': array([ 32, 224, 224])}",
+    "dataset_directory": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV",
+    "deep_supervision_scales": "[[1, 1, 1], [1.0, 0.5, 0.5], [1.0, 0.25, 0.25], [0.5, 0.125, 0.125], [0.25, 0.0625, 0.0625]]",
+    "deterministic": "False",
+    "dl_tr": "<nnunet.training.dataloading.dataset_loading.DataLoader3D object at 0x2aab60629a90>",
+    "dl_val": "<nnunet.training.dataloading.dataset_loading.DataLoader3D object at 0x2aab60840a60>",
+    "do_dummy_2D_aug": "True",
+    "ds_loss_weights": "[0.53333333 0.26666667 0.13333333 0.06666667 0.        ]",
+    "epoch": "0",
+    "experiment_name": "nnUNetTrainerV2",
+    "fold": "0",
+    "folder_with_preprocessed_data": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetData_plans_v2.1_stage1",
+    "fp16": "True",
+    "gt_niftis_folder": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/gt_segmentations",
+    "inference_pad_border_mode": "constant",
+    "inference_pad_kwargs": "{'constant_values': 0}",
+    "init_args": "('/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetPlansv2.1_plans_3D.pkl', 0, '/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1', '/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV', True, 1, True, False, True)",
+    "initial_lr": "0.01",
+    "log_file": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0/training_log_2023_3_6_15_19_10.txt",
+    "lr_scheduler": "None",
+    "lr_scheduler_eps": "0.001",
+    "lr_scheduler_patience": "30",
+    "lr_threshold": "1e-06",
+    "max_num_epochs": "1000",
+    "min_region_size_per_class": "None",
+    "min_size_per_class": "None",
+    "net_conv_kernel_sizes": "[[1, 3, 3], [1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]]",
+    "net_num_pool_op_kernel_sizes": "[[1, 2, 2], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2]]",
+    "net_pool_per_axis": "[3, 5, 5]",
+    "normalization_schemes": "OrderedDict([(0, 'CT')])",
+    "num_batches_per_epoch": "250",
+    "num_classes": "2",
+    "num_input_channels": "1",
+    "num_val_batches_per_epoch": "50",
+    "online_eval_fn": "[]",
+    "online_eval_foreground_dc": "[]",
+    "online_eval_fp": "[]",
+    "online_eval_tp": "[]",
+    "only_keep_largest_connected_component": "None",
+    "optimizer": "SGD (\nParameter Group 0\n    dampening: 0\n    foreach: None\n    lr: 0.01\n    maximize: False\n    momentum: 0.99\n    nesterov: True\n    weight_decay: 3e-05\n)",
+    "output_folder": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0",
+    "output_folder_base": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1",
+    "oversample_foreground_percent": "0.33",
+    "pad_all_sides": "None",
+    "patch_size": "[ 32 224 224]",
+    "patience": "50",
+    "pin_memory": "True",
+    "plans_file": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetPlansv2.1_plans_3D.pkl",
+    "regions_class_order": "None",
+    "save_best_checkpoint": "True",
+    "save_every": "50",
+    "save_final_checkpoint": "True",
+    "save_intermediate_checkpoints": "True",
+    "save_latest_only": "True",
+    "stage": "1",
+    "threeD": "True",
+    "tr_gen": "<batchgenerators.dataloading.multi_threaded_augmenter.MultiThreadedAugmenter object at 0x2aab60860e50>",
+    "train_loss_MA": "None",
+    "train_loss_MA_alpha": "0.93",
+    "train_loss_MA_eps": "0.0005",
+    "transpose_backward": "[0, 1, 2]",
+    "transpose_forward": "[0, 1, 2]",
+    "unpack_data": "True",
+    "use_mask_for_norm": "OrderedDict([(0, False)])",
+    "use_progress_bar": "False",
+    "val_eval_criterion_MA": "None",
+    "val_eval_criterion_alpha": "0.9",
+    "val_gen": "<batchgenerators.dataloading.multi_threaded_augmenter.MultiThreadedAugmenter object at 0x2aab60860070>",
+    "was_initialized": "True",
+    "weight_decay": "3e-05"
+}

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0/model_final_checkpoint.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56644bf1f0b2f32355a157a7382f328fb21baae7c50a03764b27a0c4422146cd
+size 248004539

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_0/model_final_checkpoint.model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a628a985ecf0ea47a291512e0f73153fd3c74ece2bc1ad568d5506a6afb29206
+size 124440

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1/debug.json

@@ -0,0 +1,87 @@
+{
+    "all_tr_losses": "[]",
+    "all_val_eval_metrics": "[]",
+    "all_val_losses": "[]",
+    "all_val_losses_tr_mode": "[]",
+    "also_val_in_tr_mode": "False",
+    "amp_grad_scaler": "None",
+    "base_num_features": "32",
+    "basic_generator_patch_size": "[ 32 263 263]",
+    "batch_dice": "True",
+    "batch_size": "2",
+    "best_MA_tr_loss_for_patience": "None",
+    "best_epoch_based_on_MA_tr_loss": "None",
+    "best_val_eval_criterion_MA": "None",
+    "classes": "[1]",
+    "conv_per_stage": "2",
+    "data_aug_params": "{'selected_data_channels': None, 'selected_seg_channels': [0], 'do_elastic': False, 'elastic_deform_alpha': (0.0, 200.0), 'elastic_deform_sigma': (9.0, 13.0), 'p_eldef': 0.2, 'do_scaling': True, 'scale_range': (0.7, 1.4), 'independent_scale_factor_for_each_axis': False, 'p_independent_scale_per_axis': 1, 'p_scale': 0.2, 'do_rotation': True, 'rotation_x': (-3.141592653589793, 3.141592653589793), 'rotation_y': (-0.5235987755982988, 0.5235987755982988), 'rotation_z': (-0.5235987755982988, 0.5235987755982988), 'rotation_p_per_axis': 1, 'p_rot': 0.2, 'random_crop': False, 'random_crop_dist_to_border': None, 'do_gamma': True, 'gamma_retain_stats': True, 'gamma_range': (0.7, 1.5), 'p_gamma': 0.3, 'do_mirror': True, 'mirror_axes': (0, 1, 2), 'dummy_2D': True, 'mask_was_used_for_normalization': OrderedDict([(0, False)]), 'border_mode_data': 'constant', 'all_segmentation_labels': None, 'move_last_seg_chanel_to_data': False, 'cascade_do_cascade_augmentations': False, 'cascade_random_binary_transform_p': 0.4, 'cascade_random_binary_transform_p_per_label': 1, 'cascade_random_binary_transform_size': (1, 8), 'cascade_remove_conn_comp_p': 0.2, 'cascade_remove_conn_comp_max_size_percent_threshold': 0.15, 'cascade_remove_conn_comp_fill_with_other_class_p': 0.0, 'do_additive_brightness': False, 'additive_brightness_p_per_sample': 0.15, 'additive_brightness_p_per_channel': 0.5, 'additive_brightness_mu': 0.0, 'additive_brightness_sigma': 0.1, 'num_threads': 12, 'num_cached_per_thread': 2, 'patch_size_for_spatialtransform': array([ 32, 224, 224])}",
+    "dataset_directory": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV",
+    "deep_supervision_scales": "[[1, 1, 1], [1.0, 0.5, 0.5], [1.0, 0.25, 0.25], [0.5, 0.125, 0.125], [0.25, 0.0625, 0.0625]]",
+    "deterministic": "False",
+    "dl_tr": "<nnunet.training.dataloading.dataset_loading.DataLoader3D object at 0x2aab60629a90>",
+    "dl_val": "<nnunet.training.dataloading.dataset_loading.DataLoader3D object at 0x2aab60840a60>",
+    "do_dummy_2D_aug": "True",
+    "ds_loss_weights": "[0.53333333 0.26666667 0.13333333 0.06666667 0.        ]",
+    "epoch": "0",
+    "experiment_name": "nnUNetTrainerV2",
+    "fold": "1",
+    "folder_with_preprocessed_data": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetData_plans_v2.1_stage1",
+    "fp16": "True",
+    "gt_niftis_folder": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/gt_segmentations",
+    "inference_pad_border_mode": "constant",
+    "inference_pad_kwargs": "{'constant_values': 0}",
+    "init_args": "('/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetPlansv2.1_plans_3D.pkl', 1, '/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1', '/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV', True, 1, True, False, True)",
+    "initial_lr": "0.01",
+    "log_file": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1/training_log_2023_3_6_15_23_41.txt",
+    "lr_scheduler": "None",
+    "lr_scheduler_eps": "0.001",
+    "lr_scheduler_patience": "30",
+    "lr_threshold": "1e-06",
+    "max_num_epochs": "1000",
+    "min_region_size_per_class": "None",
+    "min_size_per_class": "None",
+    "net_conv_kernel_sizes": "[[1, 3, 3], [1, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3], [3, 3, 3]]",
+    "net_num_pool_op_kernel_sizes": "[[1, 2, 2], [1, 2, 2], [2, 2, 2], [2, 2, 2], [2, 2, 2]]",
+    "net_pool_per_axis": "[3, 5, 5]",
+    "normalization_schemes": "OrderedDict([(0, 'CT')])",
+    "num_batches_per_epoch": "250",
+    "num_classes": "2",
+    "num_input_channels": "1",
+    "num_val_batches_per_epoch": "50",
+    "online_eval_fn": "[]",
+    "online_eval_foreground_dc": "[]",
+    "online_eval_fp": "[]",
+    "online_eval_tp": "[]",
+    "only_keep_largest_connected_component": "None",
+    "optimizer": "SGD (\nParameter Group 0\n    dampening: 0\n    foreach: None\n    lr: 0.01\n    maximize: False\n    momentum: 0.99\n    nesterov: True\n    weight_decay: 3e-05\n)",
+    "output_folder": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1",
+    "output_folder_base": "/data/houbb/nnunet/nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1",
+    "oversample_foreground_percent": "0.33",
+    "pad_all_sides": "None",
+    "patch_size": "[ 32 224 224]",
+    "patience": "50",
+    "pin_memory": "True",
+    "plans_file": "/data/houbb/nnunet/nnUNet_preprocessed/Task505_TCGA-OV/nnUNetPlansv2.1_plans_3D.pkl",
+    "regions_class_order": "None",
+    "save_best_checkpoint": "True",
+    "save_every": "50",
+    "save_final_checkpoint": "True",
+    "save_intermediate_checkpoints": "True",
+    "save_latest_only": "True",
+    "stage": "1",
+    "threeD": "True",
+    "tr_gen": "<batchgenerators.dataloading.multi_threaded_augmenter.MultiThreadedAugmenter object at 0x2aab60860c10>",
+    "train_loss_MA": "None",
+    "train_loss_MA_alpha": "0.93",
+    "train_loss_MA_eps": "0.0005",
+    "transpose_backward": "[0, 1, 2]",
+    "transpose_forward": "[0, 1, 2]",
+    "unpack_data": "True",
+    "use_mask_for_norm": "OrderedDict([(0, False)])",
+    "use_progress_bar": "False",
+    "val_eval_criterion_MA": "None",
+    "val_eval_criterion_alpha": "0.9",
+    "val_gen": "<batchgenerators.dataloading.multi_threaded_augmenter.MultiThreadedAugmenter object at 0x2aab60860370>",
+    "was_initialized": "True",
+    "weight_decay": "3e-05"
+}

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1/model_final_checkpoint.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:960d42b0ce9a6dc9cb9f2bd1466b77b309dbc2c88d4603d9142d279a400b4816
+size 248004347

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_1/model_final_checkpoint.model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3877369906a58ea40e8ba93bc61b55557c413d5ba0c1c3e5d348b955cbd15a59
+size 124440

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_2/model_final_checkpoint.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc631cca47f1312fb67e05cc72734ff1efb6403945eaecbb8b98983661bba303
+size 248004667

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_2/model_final_checkpoint.model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1225873cf4566e40fb775e58084c42a1c7fc826d29c77e117dbf71909b884104
+size 124440

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_3/model_final_checkpoint.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eeb73292862ef9e00e3efe0a8790a51cc9ec48c979deeceac16f5bfe83982b6
+size 248004731

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_3/model_final_checkpoint.model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37825c0116208da93671ddd1fd7dd6e0aa2195f1d63594c0f0d92a7ca554fa94
+size 124440

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_4/model_final_checkpoint.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2516e80b12354bda0aafed366307ef707f7e203488d8ec18a1367f2f67d7dc5
+size 248004475

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/fold_4/model_final_checkpoint.model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf068fd718b62f38878729a1a8667d73dcd1abe8ec1d6dc4de3fc5d499d9b701
+size 124440

nnUNet_trained_models/nnUNet/3d_fullres/Task505_TCGA-OV/nnUNetTrainerV2__nnUNetPlansv2.1/plans.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e337a85325a1b318c215a551e395a96ba9fa7e89fce8ad4a3a112e71a479e2e
+size 123970

nnunet_predict.py

@@ -0,0 +1,52 @@
+import argparse
+import os
+import shutil
+import random
+# import time
+
+from nnunet.inference.predict import predict_from_folder
+
+
+if __name__ == '__main__':
+
+    parser = argparse.ArgumentParser(description='Inference using nnU-Net predict_from_folder Python API')
+    parser.add_argument('-i', '--input_list', help='Input image file_list.txt')
+    parser.add_argument('-t', '--tmp_folder', help='Temporary folder', required=True)
+    parser.add_argument('-o', '--output_folder', help='Output Segmentation folder', required=True)
+    parser.add_argument('-m', '--model', help='Trained Model', required=True)
+    parser.add_argument('-v', '--verbose', help='Verbose Output', action='store_true', default=False)
+    args = vars(parser.parse_args())
+
+    # Append 8bit random hex string to ensure tmp_folder is unique
+    args['tmp_folder'] += f'_{str(hex(random.getrandbits(8)))}'
+
+    # Create temp directory
+    os.mkdir(args['tmp_folder'])
+
+    # Read input filelist
+    with open(args['input_list']) as f:
+        image_list = f.read().splitlines()
+
+    # Make destination file paths to tmp_folder
+    image_list_link = [os.path.join(args['tmp_folder'], os.path.basename(x).replace('.nii.gz', '_0000.nii.gz'))
+                       for x in image_list]
+
+    # Create hard link or copy
+    for src, dst in zip(image_list, image_list_link):
+        try:
+            os.link(src, dst)
+        except:
+            shutil.copyfile(src, dst)
+
+    # Run nnU-Net predict on tmp_folder
+    # start = time.time()
+    predict_from_folder(args['model'], args['tmp_folder'], args['output_folder'], folds=None, save_npz=False,
+                        num_threads_preprocessing=6, num_threads_nifti_save=2,
+                        lowres_segmentations=None, part_id=0, num_parts=1, tta=False,
+                        overwrite_existing=False, mode="fastest", overwrite_all_in_gpu=None,
+                        mixed_precision=True, step_size=0.5, checkpoint_name="model_final_checkpoint")
+    # end = time.time()
+    # print(f"pred time: {end - start}")
+
+    # Cleanup and delete tmp_folder
+    shutil.rmtree(args['tmp_folder'])