Hugging Face's logo

Lab-Rasool
/
CLN-Segmenter-MSD-fold0

Image Segmentation
English
nnunetv2
nnunet
medical-imaging
segmentation
3d-segmentation
ct
lung
lung-cancer
tumor-segmentation
Model card Files
xet
 Community
maazshahbaz commited on
Commit
6d81569
·
verified ·
1 Parent(s): 4b7052f

Initial upload: nnU-Net MSD Task06 fold 0 (Pseudo Dice 0.8155)

Browse files
Files changed (8) hide show
  1. .gitattributes +1 -0
  2. README.md +195 -0
  3. checkpoint_best.pth +3 -0
  4. dataset.json +15 -0
  5. dataset_fingerprint.json +648 -0
  6. nnUNetPlans.json +532 -0
  7. progress.png +3 -0
  8. splits_final.json +347 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ progress.png filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-sa-4.0
3
+ tags:
4
+ - nnunet
5
+ - nnunetv2
6
+ - medical-imaging
7
+ - segmentation
8
+ - 3d-segmentation
9
+ - ct
10
+ - lung
11
+ - lung-cancer
12
+ - tumor-segmentation
13
+ library_name: nnunetv2
14
+ pipeline_tag: image-segmentation
15
+ datasets:
16
+ - MSD-Task06-Lung
17
+ language:
18
+ - en
19
+ ---
20
+
21
+ # CLN-Segmenter — MSD Task06 Lung Tumor Segmentation (fold 0)
22
+
23
+ A 3D U-Net (nnU-Net v2 `3d_fullres`) trained on the **Medical Segmentation Decathlon Task06: Lung Tumor** dataset, fold 0 of 5-fold cross-validation. Released as part of the CLN-Segmenter project at the Rasool Lab, Moffitt Cancer Center.
24
+
25
+ This is a single-fold pretrain checkpoint, intended as a starting point for downstream lung-lesion segmentation work — not a clinical-grade tool.
26
+
27
+ ## Quick stats
28
+
29
+ | | |
30
+ |--|--|
31
+ | **Architecture** | nnU-Net v2 `3d_fullres` (PlainConvUNet, 6 stages, features `[32, 64, 128, 256, 320, 320]`) |
32
+ | **Training data** | MSD Task06 Lung — 63 cases (50 train / 13 val for fold 0) |
33
+ | **Loss** | Dice + Cross-Entropy (nnU-Net default), `batch_dice=True` |
34
+ | **Schedule** | 1000 epochs, polynomial LR decay 0.01 → 0, batch size 2, patch `[80, 192, 160]` |
35
+ | **Hardware** | 1× NVIDIA H100 80GB, ~6h wall-time |
36
+ | **Best EMA Pseudo Dice** | **0.8155** (epoch ~755) |
37
+ | **Expected real test Dice** | ~0.82–0.84 via sliding-window inference |
38
+ | **Comparison** | At the top of published nnU-Net Task06 baselines (0.69–0.78) |
39
+
40
+ ## Files in this repo
41
+
42
+ | File | Role |
43
+ |------|------|
44
+ | `checkpoint_best.pth` | Model weights — saved at the EMA Pseudo Dice peak (~epoch 755), *before* the late-epoch overfitting plateau |
45
+ | `nnUNetPlans.json` | Architecture spec + preprocessing plans. **Required** for inference. |
46
+ | `dataset.json` | Channel names, label names, file ending (nnU-Net v2 schema). **Required** for inference. |
47
+ | `dataset_fingerprint.json` | HU intensity stats from training data |
48
+ | `splits_final.json` | Train/val case ID splits for fold 0 (reproducibility) |
49
+ | `progress.png` | Training curves: loss, Pseudo Dice, epoch duration, learning rate |
50
+
51
+ ## Training data and provenance
52
+
53
+ This model was trained **only on the publicly available MSD Task06 Lung dataset** (Antonelli et al. 2022, *Nature Communications*, CC-BY-SA 4.0). It contains expert pixel-level lung tumor annotations from 63 diagnostic CT scans.
54
+
55
+ **No patient-identifiable or institutional data was used.** This checkpoint contains no information derived from any non-public source.
56
+
57
+ ## Intended use
58
+
59
+ - **Pretrained starting point** for finetuning on related lung-lesion segmentation tasks (smaller datasets, domain shift, etc.)
60
+ - **Reference baseline** for published Task06 numbers
61
+ - **Input to ensembling** with other folds (when 5-fold runs are available)
62
+
63
+ ## How NOT to use it
64
+
65
+ - ❌ Not validated for clinical diagnosis or treatment decisions
66
+ - ❌ Not validated on low-dose screening CT (LDCT) — see Limitations
67
+ - ❌ Single fold, not an ensemble — paper-grade results require all 5 folds
68
+ - ❌ Not validated outside the MSD Task06 case distribution
69
+
70
+ ## How to use
71
+
72
+ ### 1. Download the checkpoint and metadata
73
+
74
+ ```python
75
+ from huggingface_hub import snapshot_download
76
+ local_dir = snapshot_download(repo_id="Lab-Rasool/CLN-Segmenter-MSD-fold0")
77
+ print("Files at:", local_dir)
78
+ ```
79
+
80
+ ### 2. Set up an nnU-Net inference directory
81
+
82
+ nnU-Net expects a specific directory structure for results:
83
+
84
+ ```
85
+ nnUNet_results/
86
+ └── Dataset502_MSDLung/
87
+ └── nnUNetTrainer__nnUNetPlans__3d_fullres/
88
+ ├── dataset.json
89
+ ├── plans.json (rename from nnUNetPlans.json)
90
+ ├── dataset_fingerprint.json
91
+ └── fold_0/
92
+ ├── checkpoint_best.pth
93
+ └── splits_final.json
94
+ ```
95
+
96
+ You can build this with:
97
+
98
+ ```bash
99
+ DST=/path/to/nnUNet_results/Dataset502_MSDLung/nnUNetTrainer__nnUNetPlans__3d_fullres
100
+ mkdir -p $DST/fold_0
101
+ cp $local_dir/dataset.json $DST/dataset.json
102
+ cp $local_dir/nnUNetPlans.json $DST/plans.json
103
+ cp $local_dir/dataset_fingerprint.json $DST/dataset_fingerprint.json
104
+ cp $local_dir/checkpoint_best.pth $DST/fold_0/checkpoint_best.pth
105
+ cp $local_dir/splits_final.json $DST/fold_0/splits_final.json
106
+ ```
107
+
108
+ ### 3. Run inference with nnU-Net
109
+
110
+ ```bash
111
+ export nnUNet_results=/path/to/nnUNet_results
112
+ nnUNetv2_predict \
113
+ -i /path/to/your/input_images \
114
+ -o /path/to/output_predictions \
115
+ -d 502 \
116
+ -c 3d_fullres \
117
+ -tr nnUNetTrainer \
118
+ -p nnUNetPlans \
119
+ -f 0 \
120
+ -chk checkpoint_best.pth
121
+ ```
122
+
123
+ Input images should be CT volumes named with the nnU-Net channel suffix: `<case_id>_0000.nii.gz`.
124
+
125
+ ## Training procedure
126
+
127
+ - **Framework**: nnU-Net v2.7.0 (default trainer)
128
+ - **Preprocessing**: CT-specific normalization (HU clipping at the 0.5/99.5 percentiles of foreground voxels, then per-case z-score), resampling to target spacing `[1.245, 0.785, 0.785]` mm
129
+ - **Augmentation**: nnU-Net's default 3D augmentation pipeline (rotation, scaling, gamma, mirroring, gaussian noise/blur, low-resolution simulation)
130
+ - **Optimization**: SGD + Nesterov momentum (β=0.99), polynomial LR decay (initial LR 0.01)
131
+ - **Iterations**: fixed 250 per epoch (nnU-Net default; independent of dataset size)
132
+ - **Best-checkpoint mechanism**: nnU-Net automatically tracks EMA of validation Pseudo Dice and saves `checkpoint_best.pth` at the peak
133
+
134
+ ## Evaluation
135
+
136
+ | Metric | Value |
137
+ |--------|-------|
138
+ | Best EMA Pseudo Dice (fold 0 validation) | **0.8155** |
139
+ | Pseudo Dice raw (jagged) range | 0.50–0.85 |
140
+ | Final-epoch train loss | -0.85 |
141
+ | Final-epoch val loss | -0.75 |
142
+ | Train/val gap | ~0.10 (mild late-stage overfitting; `checkpoint_best` predates this) |
143
+
144
+ The training plot (`progress.png`) shows a smooth Pseudo Dice climb from 0 → 0.7 in the first ~50 epochs and a slow refinement to 0.81 by epoch ~750. After that, train loss continues to drop while val loss plateaus — this is the overfitting signature, and nnU-Net's best-checkpoint mechanism preserves the pre-overfit weights.
145
+
146
+ Note that **Pseudo Dice is voxel-pooled across validation patches**, not per-case averaged. Real test-time Dice (per-case, full-volume sliding-window inference) typically lands 0.5–3% higher than Pseudo Dice — so the 0.8155 number translates to roughly **0.82–0.84 real test Dice**, which we expect to confirm via `nnUNetv2_predict` on the 13 fold-0 validation cases.
147
+
148
+ ## Limitations
149
+
150
+ - **Single fold of 5-fold CV** — not an ensemble. Published-grade numbers require all 5 folds either averaged or ensembled at inference.
151
+ - **Trained on diagnostic CT only** — performance on low-dose screening CT (LDCT) is unknown and likely lower without finetuning.
152
+ - **Small training set** — 50 cases. The model showed mild late-stage overfitting consistent with this scale; the best-checkpoint is from before that point but generalization is bounded by data size.
153
+ - **MSD Task06 case distribution** — annotations focus on primary lung tumors (median volume ~5.2 cm³). Performance on small nodules (e.g. <5mm) or non-tumor lung lesions is not characterized.
154
+ - **No clinical validation** — this is a research artifact, not a medical device.
155
+
156
+ ## License
157
+
158
+ **CC-BY-SA 4.0**, inherited from the share-alike clause of the MSD Task06 source dataset license.
159
+
160
+ ## Citation
161
+
162
+ If you use this model, please cite:
163
+
164
+ ```bibtex
165
+ @article{isensee2021nnunet,
166
+ title = {nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation},
167
+ author = {Isensee, Fabian and Jaeger, Paul F and Kohl, Simon A A and Petersen, Jens and Maier-Hein, Klaus H},
168
+ journal = {Nature Methods},
169
+ volume = {18},
170
+ number = {2},
171
+ pages = {203--211},
172
+ year = {2021}
173
+ }
174
+
175
+ @article{antonelli2022medical,
176
+ title = {The Medical Segmentation Decathlon},
177
+ author = {Antonelli, Michela and Reinke, Annika and Bakas, Spyridon and others},
178
+ journal = {Nature Communications},
179
+ volume = {13},
180
+ number = {1},
181
+ pages = {4128},
182
+ year = {2022}
183
+ }
184
+ ```
185
+
186
+ ## Project context
187
+
188
+ Part of **CLN-Segmenter** at the Rasool Lab, Moffitt Cancer Center: a two-stage approach for lung lesion segmentation that pretrains on public datasets (this is one component) and finetunes on internal data with domain-specific loss formulations.
189
+
190
+ - **Code**: https://github.com/lab-rasool/CLN-Segmenter
191
+ - **Lab**: https://huggingface.co/Lab-Rasool
192
+
193
+ Other models in this series:
194
+ - `Lab-Rasool/CLN-Segmenter-NLSTseg-fold0` — single-dataset NLSTseg POC (LDCT, 605 expert cases)
195
+ - `Lab-Rasool/CLN-Segmenter-Dataset500-fold0` — unified MSD + NLSTseg pretrain (planned)
checkpoint_best.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c08ec796fa2b786fe99ddfad4fe6314e98fdd8820c59527be3c4105f7955d87
3
+ size 246731810
dataset.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "channel_names": {
3
+ "0": "CT"
4
+ },
5
+ "labels": {
6
+ "background": 0,
7
+ "cancer": 1
8
+ },
9
+ "numTraining": 63,
10
+ "file_ending": ".nii.gz",
11
+ "name": "Dataset502_MSDLung",
12
+ "description": "MSD Task06 lung tumor segmentation \u2014 POC dataset",
13
+ "reference": "Antonelli et al. 2022, Medical Segmentation Decathlon (Task06_Lung)",
14
+ "licence": "CC-BY-SA 4.0"
15
+ }
dataset_fingerprint.json ADDED
@@ -0,0 +1,648 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "foreground_intensity_properties_per_channel": {
3
+ "0": {
4
+ "max": 2671.0,
5
+ "mean": -273.4598083496094,
6
+ "median": -162.0,
7
+ "min": -1024.0,
8
+ "percentile_00_5": -1024.0,
9
+ "percentile_99_5": 311.0,
10
+ "std": 346.9495849609375
11
+ }
12
+ },
13
+ "median_relative_size_after_cropping": 1.0,
14
+ "shapes_after_crop": [
15
+ [
16
+ 304,
17
+ 512,
18
+ 512
19
+ ],
20
+ [
21
+ 288,
22
+ 512,
23
+ 512
24
+ ],
25
+ [
26
+ 276,
27
+ 512,
28
+ 512
29
+ ],
30
+ [
31
+ 329,
32
+ 512,
33
+ 512
34
+ ],
35
+ [
36
+ 567,
37
+ 512,
38
+ 512
39
+ ],
40
+ [
41
+ 301,
42
+ 512,
43
+ 512
44
+ ],
45
+ [
46
+ 241,
47
+ 512,
48
+ 512
49
+ ],
50
+ [
51
+ 589,
52
+ 512,
53
+ 512
54
+ ],
55
+ [
56
+ 277,
57
+ 512,
58
+ 512
59
+ ],
60
+ [
61
+ 228,
62
+ 512,
63
+ 512
64
+ ],
65
+ [
66
+ 273,
67
+ 512,
68
+ 512
69
+ ],
70
+ [
71
+ 314,
72
+ 512,
73
+ 512
74
+ ],
75
+ [
76
+ 483,
77
+ 512,
78
+ 512
79
+ ],
80
+ [
81
+ 531,
82
+ 512,
83
+ 512
84
+ ],
85
+ [
86
+ 376,
87
+ 512,
88
+ 512
89
+ ],
90
+ [
91
+ 227,
92
+ 512,
93
+ 512
94
+ ],
95
+ [
96
+ 264,
97
+ 512,
98
+ 512
99
+ ],
100
+ [
101
+ 251,
102
+ 512,
103
+ 512
104
+ ],
105
+ [
106
+ 255,
107
+ 512,
108
+ 512
109
+ ],
110
+ [
111
+ 304,
112
+ 512,
113
+ 512
114
+ ],
115
+ [
116
+ 260,
117
+ 512,
118
+ 512
119
+ ],
120
+ [
121
+ 296,
122
+ 512,
123
+ 512
124
+ ],
125
+ [
126
+ 271,
127
+ 512,
128
+ 512
129
+ ],
130
+ [
131
+ 126,
132
+ 512,
133
+ 512
134
+ ],
135
+ [
136
+ 125,
137
+ 512,
138
+ 512
139
+ ],
140
+ [
141
+ 240,
142
+ 512,
143
+ 512
144
+ ],
145
+ [
146
+ 125,
147
+ 512,
148
+ 512
149
+ ],
150
+ [
151
+ 129,
152
+ 512,
153
+ 512
154
+ ],
155
+ [
156
+ 282,
157
+ 512,
158
+ 512
159
+ ],
160
+ [
161
+ 252,
162
+ 512,
163
+ 512
164
+ ],
165
+ [
166
+ 226,
167
+ 512,
168
+ 512
169
+ ],
170
+ [
171
+ 636,
172
+ 512,
173
+ 512
174
+ ],
175
+ [
176
+ 322,
177
+ 512,
178
+ 512
179
+ ],
180
+ [
181
+ 291,
182
+ 512,
183
+ 512
184
+ ],
185
+ [
186
+ 155,
187
+ 512,
188
+ 512
189
+ ],
190
+ [
191
+ 252,
192
+ 512,
193
+ 512
194
+ ],
195
+ [
196
+ 241,
197
+ 512,
198
+ 512
199
+ ],
200
+ [
201
+ 242,
202
+ 512,
203
+ 512
204
+ ],
205
+ [
206
+ 112,
207
+ 512,
208
+ 512
209
+ ],
210
+ [
211
+ 241,
212
+ 512,
213
+ 512
214
+ ],
215
+ [
216
+ 218,
217
+ 512,
218
+ 512
219
+ ],
220
+ [
221
+ 211,
222
+ 512,
223
+ 512
224
+ ],
225
+ [
226
+ 243,
227
+ 512,
228
+ 512
229
+ ],
230
+ [
231
+ 248,
232
+ 512,
233
+ 512
234
+ ],
235
+ [
236
+ 128,
237
+ 512,
238
+ 512
239
+ ],
240
+ [
241
+ 241,
242
+ 512,
243
+ 512
244
+ ],
245
+ [
246
+ 242,
247
+ 512,
248
+ 512
249
+ ],
250
+ [
251
+ 266,
252
+ 512,
253
+ 512
254
+ ],
255
+ [
256
+ 289,
257
+ 512,
258
+ 512
259
+ ],
260
+ [
261
+ 493,
262
+ 512,
263
+ 512
264
+ ],
265
+ [
266
+ 369,
267
+ 512,
268
+ 512
269
+ ],
270
+ [
271
+ 423,
272
+ 512,
273
+ 512
274
+ ],
275
+ [
276
+ 115,
277
+ 512,
278
+ 512
279
+ ],
280
+ [
281
+ 251,
282
+ 512,
283
+ 512
284
+ ],
285
+ [
286
+ 450,
287
+ 512,
288
+ 512
289
+ ],
290
+ [
291
+ 249,
292
+ 512,
293
+ 512
294
+ ],
295
+ [
296
+ 256,
297
+ 512,
298
+ 512
299
+ ],
300
+ [
301
+ 241,
302
+ 512,
303
+ 512
304
+ ],
305
+ [
306
+ 267,
307
+ 512,
308
+ 512
309
+ ],
310
+ [
311
+ 245,
312
+ 512,
313
+ 512
314
+ ],
315
+ [
316
+ 228,
317
+ 512,
318
+ 512
319
+ ],
320
+ [
321
+ 231,
322
+ 512,
323
+ 512
324
+ ],
325
+ [
326
+ 251,
327
+ 512,
328
+ 512
329
+ ]
330
+ ],
331
+ "spacings": [
332
+ [
333
+ 1.0,
334
+ 0.693359375,
335
+ 0.693359375
336
+ ],
337
+ [
338
+ 1.2456597089767456,
339
+ 0.9375,
340
+ 0.9375
341
+ ],
342
+ [
343
+ 1.2454710006713867,
344
+ 0.828125,
345
+ 0.828125
346
+ ],
347
+ [
348
+ 1.0,
349
+ 0.654296875,
350
+ 0.654296875
351
+ ],
352
+ [
353
+ 0.625,
354
+ 0.9765620231628418,
355
+ 0.9765620231628418
356
+ ],
357
+ [
358
+ 1.25,
359
+ 0.8203120231628418,
360
+ 0.8203120231628418
361
+ ],
362
+ [
363
+ 1.25,
364
+ 0.5976560115814209,
365
+ 0.5976560115814209
366
+ ],
367
+ [
368
+ 0.625,
369
+ 0.7421879768371582,
370
+ 0.7421879768371582
371
+ ],
372
+ [
373
+ 1.25,
374
+ 0.8203120231628418,
375
+ 0.8203120231628418
376
+ ],
377
+ [
378
+ 1.2445175647735596,
379
+ 0.7773439884185791,
380
+ 0.7773439884185791
381
+ ],
382
+ [
383
+ 1.25,
384
+ 0.859375,
385
+ 0.859375
386
+ ],
387
+ [
388
+ 1.0,
389
+ 0.677734375,
390
+ 0.677734375
391
+ ],
392
+ [
393
+ 0.625,
394
+ 0.8417969942092896,
395
+ 0.8417969942092896
396
+ ],
397
+ [
398
+ 0.625,
399
+ 0.8203120231628418,
400
+ 0.8203120231628418
401
+ ],
402
+ [
403
+ 0.9973404407501221,
404
+ 0.8007810115814209,
405
+ 0.8007810115814209
406
+ ],
407
+ [
408
+ 1.2444933652877808,
409
+ 0.78125,
410
+ 0.78125
411
+ ],
412
+ [
413
+ 1.2452651262283325,
414
+ 0.859375,
415
+ 0.859375
416
+ ],
417
+ [
418
+ 1.2450199127197266,
419
+ 0.9101560115814209,
420
+ 0.9101560115814209
421
+ ],
422
+ [
423
+ 1.2450979948043823,
424
+ 0.8457030057907104,
425
+ 0.8457030057907104
426
+ ],
427
+ [
428
+ 1.245888113975525,
429
+ 0.703125,
430
+ 0.703125
431
+ ],
432
+ [
433
+ 1.245192289352417,
434
+ 0.8203120231628418,
435
+ 0.8203120231628418
436
+ ],
437
+ [
438
+ 1.2457770109176636,
439
+ 0.8984379768371582,
440
+ 0.8984379768371582
441
+ ],
442
+ [
443
+ 1.2453874349594116,
444
+ 0.78125,
445
+ 0.78125
446
+ ],
447
+ [
448
+ 2.5,
449
+ 0.78125,
450
+ 0.78125
451
+ ],
452
+ [
453
+ 2.5,
454
+ 0.78125,
455
+ 0.78125
456
+ ],
457
+ [
458
+ 1.2447916269302368,
459
+ 0.8203120231628418,
460
+ 0.8203120231628418
461
+ ],
462
+ [
463
+ 2.5,
464
+ 0.7324219942092896,
465
+ 0.7324219942092896
466
+ ],
467
+ [
468
+ 2.5,
469
+ 0.7695310115814209,
470
+ 0.7695310115814209
471
+ ],
472
+ [
473
+ 1.2455673217773438,
474
+ 0.859375,
475
+ 0.859375
476
+ ],
477
+ [
478
+ 1.245039701461792,
479
+ 0.859375,
480
+ 0.859375
481
+ ],
482
+ [
483
+ 1.2444690465927124,
484
+ 0.78125,
485
+ 0.78125
486
+ ],
487
+ [
488
+ 1.0,
489
+ 0.78515625,
490
+ 0.78515625
491
+ ],
492
+ [
493
+ 1.0,
494
+ 0.814453125,
495
+ 0.814453125
496
+ ],
497
+ [
498
+ 1.25,
499
+ 0.703125,
500
+ 0.703125
501
+ ],
502
+ [
503
+ 2.5,
504
+ 0.78125,
505
+ 0.78125
506
+ ],
507
+ [
508
+ 1.245039701461792,
509
+ 0.9375,
510
+ 0.9375
511
+ ],
512
+ [
513
+ 1.244813323020935,
514
+ 0.7578120231628418,
515
+ 0.7578120231628418
516
+ ],
517
+ [
518
+ 1.2448346614837646,
519
+ 0.8203120231628418,
520
+ 0.8203120231628418
521
+ ],
522
+ [
523
+ 2.5,
524
+ 0.78125,
525
+ 0.78125
526
+ ],
527
+ [
528
+ 1.244813323020935,
529
+ 0.859375,
530
+ 0.859375
531
+ ],
532
+ [
533
+ 1.2442660331726074,
534
+ 0.8203120231628418,
535
+ 0.8203120231628418
536
+ ],
537
+ [
538
+ 1.2440757751464844,
539
+ 0.859375,
540
+ 0.859375
541
+ ],
542
+ [
543
+ 1.2448559999465942,
544
+ 0.8203120231628418,
545
+ 0.8203120231628418
546
+ ],
547
+ [
548
+ 1.2449597120285034,
549
+ 0.703125,
550
+ 0.703125
551
+ ],
552
+ [
553
+ 2.5,
554
+ 0.7265620231628418,
555
+ 0.7265620231628418
556
+ ],
557
+ [
558
+ 1.244813323020935,
559
+ 0.8867189884185791,
560
+ 0.8867189884185791
561
+ ],
562
+ [
563
+ 1.2448346614837646,
564
+ 0.7871090173721313,
565
+ 0.7871090173721313
566
+ ],
567
+ [
568
+ 1.2453007698059082,
569
+ 0.7617189884185791,
570
+ 0.7617189884185791
571
+ ],
572
+ [
573
+ 1.245674729347229,
574
+ 0.703125,
575
+ 0.703125
576
+ ],
577
+ [
578
+ 0.625,
579
+ 0.7890620231628418,
580
+ 0.7890620231628418
581
+ ],
582
+ [
583
+ 1.0,
584
+ 0.625,
585
+ 0.625
586
+ ],
587
+ [
588
+ 1.0,
589
+ 0.798828125,
590
+ 0.798828125
591
+ ],
592
+ [
593
+ 2.4782607555389404,
594
+ 0.6035159826278687,
595
+ 0.6035159826278687
596
+ ],
597
+ [
598
+ 1.2450199127197266,
599
+ 0.9296879768371582,
600
+ 0.9296879768371582
601
+ ],
602
+ [
603
+ 0.625,
604
+ 0.6171879768371582,
605
+ 0.6171879768371582
606
+ ],
607
+ [
608
+ 1.244979977607727,
609
+ 0.8808590173721313,
610
+ 0.8808590173721313
611
+ ],
612
+ [
613
+ 1.2451171875,
614
+ 0.78125,
615
+ 0.78125
616
+ ],
617
+ [
618
+ 1.244813323020935,
619
+ 0.78125,
620
+ 0.78125
621
+ ],
622
+ [
623
+ 1.2453182935714722,
624
+ 0.78125,
625
+ 0.78125
626
+ ],
627
+ [
628
+ 1.2448979616165161,
629
+ 0.9492189884185791,
630
+ 0.9492189884185791
631
+ ],
632
+ [
633
+ 1.2445175647735596,
634
+ 0.78125,
635
+ 0.78125
636
+ ],
637
+ [
638
+ 1.2445887327194214,
639
+ 0.7714840173721313,
640
+ 0.7714840173721313
641
+ ],
642
+ [
643
+ 1.2450199127197266,
644
+ 0.7832030057907104,
645
+ 0.7832030057907104
646
+ ]
647
+ ]
648
+ }
nnUNetPlans.json ADDED
@@ -0,0 +1,532 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset_name": "Dataset502_MSDLung",
3
+ "plans_name": "nnUNetPlans",
4
+ "original_median_spacing_after_transp": [
5
+ 1.244979977607727,
6
+ 0.78515625,
7
+ 0.78515625
8
+ ],
9
+ "original_median_shape_after_transp": [
10
+ 252,
11
+ 512,
12
+ 512
13
+ ],
14
+ "image_reader_writer": "SimpleITKIO",
15
+ "transpose_forward": [
16
+ 0,
17
+ 1,
18
+ 2
19
+ ],
20
+ "transpose_backward": [
21
+ 0,
22
+ 1,
23
+ 2
24
+ ],
25
+ "configurations": {
26
+ "2d": {
27
+ "data_identifier": "nnUNetPlans_2d",
28
+ "preprocessor_name": "DefaultPreprocessor",
29
+ "batch_size": 12,
30
+ "patch_size": [
31
+ 512,
32
+ 512
33
+ ],
34
+ "median_image_size_in_voxels": [
35
+ 512.0,
36
+ 512.0
37
+ ],
38
+ "spacing": [
39
+ 0.78515625,
40
+ 0.78515625
41
+ ],
42
+ "normalization_schemes": [
43
+ "CTNormalization"
44
+ ],
45
+ "use_mask_for_norm": [
46
+ false
47
+ ],
48
+ "resampling_fn_data": "resample_data_or_seg_to_shape",
49
+ "resampling_fn_seg": "resample_data_or_seg_to_shape",
50
+ "resampling_fn_data_kwargs": {
51
+ "is_seg": false,
52
+ "order": 3,
53
+ "order_z": 0,
54
+ "force_separate_z": null
55
+ },
56
+ "resampling_fn_seg_kwargs": {
57
+ "is_seg": true,
58
+ "order": 1,
59
+ "order_z": 0,
60
+ "force_separate_z": null
61
+ },
62
+ "resampling_fn_probabilities": "resample_data_or_seg_to_shape",
63
+ "resampling_fn_probabilities_kwargs": {
64
+ "is_seg": false,
65
+ "order": 1,
66
+ "order_z": 0,
67
+ "force_separate_z": null
68
+ },
69
+ "architecture": {
70
+ "network_class_name": "dynamic_network_architectures.architectures.unet.PlainConvUNet",
71
+ "arch_kwargs": {
72
+ "n_stages": 8,
73
+ "features_per_stage": [
74
+ 32,
75
+ 64,
76
+ 128,
77
+ 256,
78
+ 512,
79
+ 512,
80
+ 512,
81
+ 512
82
+ ],
83
+ "conv_op": "torch.nn.modules.conv.Conv2d",
84
+ "kernel_sizes": [
85
+ [
86
+ 3,
87
+ 3
88
+ ],
89
+ [
90
+ 3,
91
+ 3
92
+ ],
93
+ [
94
+ 3,
95
+ 3
96
+ ],
97
+ [
98
+ 3,
99
+ 3
100
+ ],
101
+ [
102
+ 3,
103
+ 3
104
+ ],
105
+ [
106
+ 3,
107
+ 3
108
+ ],
109
+ [
110
+ 3,
111
+ 3
112
+ ],
113
+ [
114
+ 3,
115
+ 3
116
+ ]
117
+ ],
118
+ "strides": [
119
+ [
120
+ 1,
121
+ 1
122
+ ],
123
+ [
124
+ 2,
125
+ 2
126
+ ],
127
+ [
128
+ 2,
129
+ 2
130
+ ],
131
+ [
132
+ 2,
133
+ 2
134
+ ],
135
+ [
136
+ 2,
137
+ 2
138
+ ],
139
+ [
140
+ 2,
141
+ 2
142
+ ],
143
+ [
144
+ 2,
145
+ 2
146
+ ],
147
+ [
148
+ 2,
149
+ 2
150
+ ]
151
+ ],
152
+ "n_conv_per_stage": [
153
+ 2,
154
+ 2,
155
+ 2,
156
+ 2,
157
+ 2,
158
+ 2,
159
+ 2,
160
+ 2
161
+ ],
162
+ "n_conv_per_stage_decoder": [
163
+ 2,
164
+ 2,
165
+ 2,
166
+ 2,
167
+ 2,
168
+ 2,
169
+ 2
170
+ ],
171
+ "conv_bias": true,
172
+ "norm_op": "torch.nn.modules.instancenorm.InstanceNorm2d",
173
+ "norm_op_kwargs": {
174
+ "eps": 1e-05,
175
+ "affine": true
176
+ },
177
+ "dropout_op": null,
178
+ "dropout_op_kwargs": null,
179
+ "nonlin": "torch.nn.LeakyReLU",
180
+ "nonlin_kwargs": {
181
+ "inplace": true
182
+ }
183
+ },
184
+ "_kw_requires_import": [
185
+ "conv_op",
186
+ "norm_op",
187
+ "dropout_op",
188
+ "nonlin"
189
+ ]
190
+ },
191
+ "batch_dice": true
192
+ },
193
+ "3d_lowres": {
194
+ "data_identifier": "nnUNetPlans_3d_lowres",
195
+ "preprocessor_name": "DefaultPreprocessor",
196
+ "batch_size": 2,
197
+ "patch_size": [
198
+ 80,
199
+ 192,
200
+ 160
201
+ ],
202
+ "median_image_size_in_voxels": [
203
+ 132,
204
+ 267,
205
+ 267
206
+ ],
207
+ "spacing": [
208
+ 2.3855103790575867,
209
+ 1.5044405671133478,
210
+ 1.5044405671133478
211
+ ],
212
+ "normalization_schemes": [
213
+ "CTNormalization"
214
+ ],
215
+ "use_mask_for_norm": [
216
+ false
217
+ ],
218
+ "resampling_fn_data": "resample_data_or_seg_to_shape",
219
+ "resampling_fn_seg": "resample_data_or_seg_to_shape",
220
+ "resampling_fn_data_kwargs": {
221
+ "is_seg": false,
222
+ "order": 3,
223
+ "order_z": 0,
224
+ "force_separate_z": null
225
+ },
226
+ "resampling_fn_seg_kwargs": {
227
+ "is_seg": true,
228
+ "order": 1,
229
+ "order_z": 0,
230
+ "force_separate_z": null
231
+ },
232
+ "resampling_fn_probabilities": "resample_data_or_seg_to_shape",
233
+ "resampling_fn_probabilities_kwargs": {
234
+ "is_seg": false,
235
+ "order": 1,
236
+ "order_z": 0,
237
+ "force_separate_z": null
238
+ },
239
+ "architecture": {
240
+ "network_class_name": "dynamic_network_architectures.architectures.unet.PlainConvUNet",
241
+ "arch_kwargs": {
242
+ "n_stages": 6,
243
+ "features_per_stage": [
244
+ 32,
245
+ 64,
246
+ 128,
247
+ 256,
248
+ 320,
249
+ 320
250
+ ],
251
+ "conv_op": "torch.nn.modules.conv.Conv3d",
252
+ "kernel_sizes": [
253
+ [
254
+ 3,
255
+ 3,
256
+ 3
257
+ ],
258
+ [
259
+ 3,
260
+ 3,
261
+ 3
262
+ ],
263
+ [
264
+ 3,
265
+ 3,
266
+ 3
267
+ ],
268
+ [
269
+ 3,
270
+ 3,
271
+ 3
272
+ ],
273
+ [
274
+ 3,
275
+ 3,
276
+ 3
277
+ ],
278
+ [
279
+ 3,
280
+ 3,
281
+ 3
282
+ ]
283
+ ],
284
+ "strides": [
285
+ [
286
+ 1,
287
+ 1,
288
+ 1
289
+ ],
290
+ [
291
+ 2,
292
+ 2,
293
+ 2
294
+ ],
295
+ [
296
+ 2,
297
+ 2,
298
+ 2
299
+ ],
300
+ [
301
+ 2,
302
+ 2,
303
+ 2
304
+ ],
305
+ [
306
+ 2,
307
+ 2,
308
+ 2
309
+ ],
310
+ [
311
+ 1,
312
+ 2,
313
+ 2
314
+ ]
315
+ ],
316
+ "n_conv_per_stage": [
317
+ 2,
318
+ 2,
319
+ 2,
320
+ 2,
321
+ 2,
322
+ 2
323
+ ],
324
+ "n_conv_per_stage_decoder": [
325
+ 2,
326
+ 2,
327
+ 2,
328
+ 2,
329
+ 2
330
+ ],
331
+ "conv_bias": true,
332
+ "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d",
333
+ "norm_op_kwargs": {
334
+ "eps": 1e-05,
335
+ "affine": true
336
+ },
337
+ "dropout_op": null,
338
+ "dropout_op_kwargs": null,
339
+ "nonlin": "torch.nn.LeakyReLU",
340
+ "nonlin_kwargs": {
341
+ "inplace": true
342
+ }
343
+ },
344
+ "_kw_requires_import": [
345
+ "conv_op",
346
+ "norm_op",
347
+ "dropout_op",
348
+ "nonlin"
349
+ ]
350
+ },
351
+ "batch_dice": false,
352
+ "next_stage": "3d_cascade_fullres"
353
+ },
354
+ "3d_fullres": {
355
+ "data_identifier": "nnUNetPlans_3d_fullres",
356
+ "preprocessor_name": "DefaultPreprocessor",
357
+ "batch_size": 2,
358
+ "patch_size": [
359
+ 80,
360
+ 192,
361
+ 160
362
+ ],
363
+ "median_image_size_in_voxels": [
364
+ 252.0,
365
+ 512.0,
366
+ 512.0
367
+ ],
368
+ "spacing": [
369
+ 1.244979977607727,
370
+ 0.78515625,
371
+ 0.78515625
372
+ ],
373
+ "normalization_schemes": [
374
+ "CTNormalization"
375
+ ],
376
+ "use_mask_for_norm": [
377
+ false
378
+ ],
379
+ "resampling_fn_data": "resample_data_or_seg_to_shape",
380
+ "resampling_fn_seg": "resample_data_or_seg_to_shape",
381
+ "resampling_fn_data_kwargs": {
382
+ "is_seg": false,
383
+ "order": 3,
384
+ "order_z": 0,
385
+ "force_separate_z": null
386
+ },
387
+ "resampling_fn_seg_kwargs": {
388
+ "is_seg": true,
389
+ "order": 1,
390
+ "order_z": 0,
391
+ "force_separate_z": null
392
+ },
393
+ "resampling_fn_probabilities": "resample_data_or_seg_to_shape",
394
+ "resampling_fn_probabilities_kwargs": {
395
+ "is_seg": false,
396
+ "order": 1,
397
+ "order_z": 0,
398
+ "force_separate_z": null
399
+ },
400
+ "architecture": {
401
+ "network_class_name": "dynamic_network_architectures.architectures.unet.PlainConvUNet",
402
+ "arch_kwargs": {
403
+ "n_stages": 6,
404
+ "features_per_stage": [
405
+ 32,
406
+ 64,
407
+ 128,
408
+ 256,
409
+ 320,
410
+ 320
411
+ ],
412
+ "conv_op": "torch.nn.modules.conv.Conv3d",
413
+ "kernel_sizes": [
414
+ [
415
+ 3,
416
+ 3,
417
+ 3
418
+ ],
419
+ [
420
+ 3,
421
+ 3,
422
+ 3
423
+ ],
424
+ [
425
+ 3,
426
+ 3,
427
+ 3
428
+ ],
429
+ [
430
+ 3,
431
+ 3,
432
+ 3
433
+ ],
434
+ [
435
+ 3,
436
+ 3,
437
+ 3
438
+ ],
439
+ [
440
+ 3,
441
+ 3,
442
+ 3
443
+ ]
444
+ ],
445
+ "strides": [
446
+ [
447
+ 1,
448
+ 1,
449
+ 1
450
+ ],
451
+ [
452
+ 2,
453
+ 2,
454
+ 2
455
+ ],
456
+ [
457
+ 2,
458
+ 2,
459
+ 2
460
+ ],
461
+ [
462
+ 2,
463
+ 2,
464
+ 2
465
+ ],
466
+ [
467
+ 2,
468
+ 2,
469
+ 2
470
+ ],
471
+ [
472
+ 1,
473
+ 2,
474
+ 2
475
+ ]
476
+ ],
477
+ "n_conv_per_stage": [
478
+ 2,
479
+ 2,
480
+ 2,
481
+ 2,
482
+ 2,
483
+ 2
484
+ ],
485
+ "n_conv_per_stage_decoder": [
486
+ 2,
487
+ 2,
488
+ 2,
489
+ 2,
490
+ 2
491
+ ],
492
+ "conv_bias": true,
493
+ "norm_op": "torch.nn.modules.instancenorm.InstanceNorm3d",
494
+ "norm_op_kwargs": {
495
+ "eps": 1e-05,
496
+ "affine": true
497
+ },
498
+ "dropout_op": null,
499
+ "dropout_op_kwargs": null,
500
+ "nonlin": "torch.nn.LeakyReLU",
501
+ "nonlin_kwargs": {
502
+ "inplace": true
503
+ }
504
+ },
505
+ "_kw_requires_import": [
506
+ "conv_op",
507
+ "norm_op",
508
+ "dropout_op",
509
+ "nonlin"
510
+ ]
511
+ },
512
+ "batch_dice": true
513
+ },
514
+ "3d_cascade_fullres": {
515
+ "inherits_from": "3d_fullres",
516
+ "previous_stage": "3d_lowres"
517
+ }
518
+ },
519
+ "experiment_planner_used": "ExperimentPlanner",
520
+ "label_manager": "LabelManager",
521
+ "foreground_intensity_properties_per_channel": {
522
+ "0": {
523
+ "max": 2671.0,
524
+ "mean": -273.4598083496094,
525
+ "median": -162.0,
526
+ "min": -1024.0,
527
+ "percentile_00_5": -1024.0,
528
+ "percentile_99_5": 311.0,
529
+ "std": 346.9495849609375
530
+ }
531
+ }
532
+ }
progress.png ADDED

Git LFS Details

  • SHA256: 91bc59896e03ba2f12bc322afdf6f46e82ed4dd8221de7d9b7daff6329ffadf0
  • Pointer size: 132 Bytes
  • Size of remote file: 1.02 MB
splits_final.json ADDED
@@ -0,0 +1,347 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "train": [
4
+ "lung_001",
5
+ "lung_003",
6
+ "lung_004",
7
+ "lung_005",
8
+ "lung_009",
9
+ "lung_014",
10
+ "lung_015",
11
+ "lung_016",
12
+ "lung_018",
13
+ "lung_020",
14
+ "lung_022",
15
+ "lung_023",
16
+ "lung_025",
17
+ "lung_026",
18
+ "lung_027",
19
+ "lung_028",
20
+ "lung_029",
21
+ "lung_031",
22
+ "lung_036",
23
+ "lung_037",
24
+ "lung_038",
25
+ "lung_043",
26
+ "lung_044",
27
+ "lung_045",
28
+ "lung_047",
29
+ "lung_049",
30
+ "lung_051",
31
+ "lung_053",
32
+ "lung_054",
33
+ "lung_055",
34
+ "lung_057",
35
+ "lung_058",
36
+ "lung_061",
37
+ "lung_062",
38
+ "lung_064",
39
+ "lung_069",
40
+ "lung_071",
41
+ "lung_073",
42
+ "lung_074",
43
+ "lung_075",
44
+ "lung_078",
45
+ "lung_080",
46
+ "lung_081",
47
+ "lung_083",
48
+ "lung_084",
49
+ "lung_086",
50
+ "lung_092",
51
+ "lung_093",
52
+ "lung_095",
53
+ "lung_096"
54
+ ],
55
+ "val": [
56
+ "lung_006",
57
+ "lung_010",
58
+ "lung_033",
59
+ "lung_034",
60
+ "lung_041",
61
+ "lung_042",
62
+ "lung_046",
63
+ "lung_048",
64
+ "lung_059",
65
+ "lung_065",
66
+ "lung_066",
67
+ "lung_070",
68
+ "lung_079"
69
+ ]
70
+ },
71
+ {
72
+ "train": [
73
+ "lung_001",
74
+ "lung_003",
75
+ "lung_005",
76
+ "lung_006",
77
+ "lung_009",
78
+ "lung_010",
79
+ "lung_014",
80
+ "lung_016",
81
+ "lung_018",
82
+ "lung_020",
83
+ "lung_023",
84
+ "lung_025",
85
+ "lung_026",
86
+ "lung_027",
87
+ "lung_028",
88
+ "lung_029",
89
+ "lung_033",
90
+ "lung_034",
91
+ "lung_037",
92
+ "lung_041",
93
+ "lung_042",
94
+ "lung_043",
95
+ "lung_044",
96
+ "lung_045",
97
+ "lung_046",
98
+ "lung_047",
99
+ "lung_048",
100
+ "lung_049",
101
+ "lung_051",
102
+ "lung_054",
103
+ "lung_055",
104
+ "lung_057",
105
+ "lung_058",
106
+ "lung_059",
107
+ "lung_061",
108
+ "lung_065",
109
+ "lung_066",
110
+ "lung_070",
111
+ "lung_073",
112
+ "lung_074",
113
+ "lung_078",
114
+ "lung_079",
115
+ "lung_080",
116
+ "lung_083",
117
+ "lung_084",
118
+ "lung_086",
119
+ "lung_092",
120
+ "lung_093",
121
+ "lung_095",
122
+ "lung_096"
123
+ ],
124
+ "val": [
125
+ "lung_004",
126
+ "lung_015",
127
+ "lung_022",
128
+ "lung_031",
129
+ "lung_036",
130
+ "lung_038",
131
+ "lung_053",
132
+ "lung_062",
133
+ "lung_064",
134
+ "lung_069",
135
+ "lung_071",
136
+ "lung_075",
137
+ "lung_081"
138
+ ]
139
+ },
140
+ {
141
+ "train": [
142
+ "lung_003",
143
+ "lung_004",
144
+ "lung_006",
145
+ "lung_010",
146
+ "lung_014",
147
+ "lung_015",
148
+ "lung_016",
149
+ "lung_018",
150
+ "lung_020",
151
+ "lung_022",
152
+ "lung_023",
153
+ "lung_025",
154
+ "lung_027",
155
+ "lung_028",
156
+ "lung_029",
157
+ "lung_031",
158
+ "lung_033",
159
+ "lung_034",
160
+ "lung_036",
161
+ "lung_038",
162
+ "lung_041",
163
+ "lung_042",
164
+ "lung_043",
165
+ "lung_045",
166
+ "lung_046",
167
+ "lung_048",
168
+ "lung_051",
169
+ "lung_053",
170
+ "lung_054",
171
+ "lung_055",
172
+ "lung_057",
173
+ "lung_058",
174
+ "lung_059",
175
+ "lung_061",
176
+ "lung_062",
177
+ "lung_064",
178
+ "lung_065",
179
+ "lung_066",
180
+ "lung_069",
181
+ "lung_070",
182
+ "lung_071",
183
+ "lung_073",
184
+ "lung_075",
185
+ "lung_079",
186
+ "lung_081",
187
+ "lung_084",
188
+ "lung_092",
189
+ "lung_093",
190
+ "lung_095",
191
+ "lung_096"
192
+ ],
193
+ "val": [
194
+ "lung_001",
195
+ "lung_005",
196
+ "lung_009",
197
+ "lung_026",
198
+ "lung_037",
199
+ "lung_044",
200
+ "lung_047",
201
+ "lung_049",
202
+ "lung_074",
203
+ "lung_078",
204
+ "lung_080",
205
+ "lung_083",
206
+ "lung_086"
207
+ ]
208
+ },
209
+ {
210
+ "train": [
211
+ "lung_001",
212
+ "lung_003",
213
+ "lung_004",
214
+ "lung_005",
215
+ "lung_006",
216
+ "lung_009",
217
+ "lung_010",
218
+ "lung_015",
219
+ "lung_022",
220
+ "lung_025",
221
+ "lung_026",
222
+ "lung_031",
223
+ "lung_033",
224
+ "lung_034",
225
+ "lung_036",
226
+ "lung_037",
227
+ "lung_038",
228
+ "lung_041",
229
+ "lung_042",
230
+ "lung_044",
231
+ "lung_045",
232
+ "lung_046",
233
+ "lung_047",
234
+ "lung_048",
235
+ "lung_049",
236
+ "lung_051",
237
+ "lung_053",
238
+ "lung_054",
239
+ "lung_055",
240
+ "lung_059",
241
+ "lung_061",
242
+ "lung_062",
243
+ "lung_064",
244
+ "lung_065",
245
+ "lung_066",
246
+ "lung_069",
247
+ "lung_070",
248
+ "lung_071",
249
+ "lung_073",
250
+ "lung_074",
251
+ "lung_075",
252
+ "lung_078",
253
+ "lung_079",
254
+ "lung_080",
255
+ "lung_081",
256
+ "lung_083",
257
+ "lung_086",
258
+ "lung_092",
259
+ "lung_093",
260
+ "lung_095",
261
+ "lung_096"
262
+ ],
263
+ "val": [
264
+ "lung_014",
265
+ "lung_016",
266
+ "lung_018",
267
+ "lung_020",
268
+ "lung_023",
269
+ "lung_027",
270
+ "lung_028",
271
+ "lung_029",
272
+ "lung_043",
273
+ "lung_057",
274
+ "lung_058",
275
+ "lung_084"
276
+ ]
277
+ },
278
+ {
279
+ "train": [
280
+ "lung_001",
281
+ "lung_004",
282
+ "lung_005",
283
+ "lung_006",
284
+ "lung_009",
285
+ "lung_010",
286
+ "lung_014",
287
+ "lung_015",
288
+ "lung_016",
289
+ "lung_018",
290
+ "lung_020",
291
+ "lung_022",
292
+ "lung_023",
293
+ "lung_026",
294
+ "lung_027",
295
+ "lung_028",
296
+ "lung_029",
297
+ "lung_031",
298
+ "lung_033",
299
+ "lung_034",
300
+ "lung_036",
301
+ "lung_037",
302
+ "lung_038",
303
+ "lung_041",
304
+ "lung_042",
305
+ "lung_043",
306
+ "lung_044",
307
+ "lung_046",
308
+ "lung_047",
309
+ "lung_048",
310
+ "lung_049",
311
+ "lung_053",
312
+ "lung_057",
313
+ "lung_058",
314
+ "lung_059",
315
+ "lung_062",
316
+ "lung_064",
317
+ "lung_065",
318
+ "lung_066",
319
+ "lung_069",
320
+ "lung_070",
321
+ "lung_071",
322
+ "lung_074",
323
+ "lung_075",
324
+ "lung_078",
325
+ "lung_079",
326
+ "lung_080",
327
+ "lung_081",
328
+ "lung_083",
329
+ "lung_084",
330
+ "lung_086"
331
+ ],
332
+ "val": [
333
+ "lung_003",
334
+ "lung_025",
335
+ "lung_045",
336
+ "lung_051",
337
+ "lung_054",
338
+ "lung_055",
339
+ "lung_061",
340
+ "lung_073",
341
+ "lung_092",
342
+ "lung_093",
343
+ "lung_095",
344
+ "lung_096"
345
+ ]
346
+ }
347
+ ]