kiselyovd commited on
Commit
9f7fb9f
·
verified ·
1 Parent(s): 946c9ff

Upload artifacts

Browse files
README.md ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ tags:
4
+ - semantic-segmentation
5
+ - medical-imaging
6
+ - brain-mri
7
+ - pytorch-lightning
8
+ - segformer
9
+ library_name: pytorch
10
+ datasets:
11
+ - mateuszbuda/lgg-mri-segmentation
12
+ pipeline_tag: image-segmentation
13
+ ---
14
+
15
+ # brain-mri-segmentation
16
+
17
+ Production-grade binary brain-tumor MRI segmentation (LGG / TCGA).
18
+
19
+ Binary semantic segmentation of brain-tumor regions (low-grade glioma) from FLAIR MRI slices. Main model: SegFormer-B2 fine-tuned on the Mateusz Buda LGG MRI dataset (TCGA, 110 patients, 3 929 paired slices) with a patient-level 80/10/10 split. Baseline: hand-rolled U-Net (4 levels, 32→256 ch, ~1.9 M params).
20
+
21
+ Output is a binary mask at 256 × 256 resolution (1 = tumor, 0 = background).
22
+
23
+ ## Metrics
24
+
25
+ | Metric | Value |
26
+ |---|---|
27
+ | Main model | SegFormer-B2 |
28
+ | Main Dice | 65.5% |
29
+ | Main IoU | 66.2% |
30
+ | Main Pixel accuracy | 99.73% |
31
+ | Baseline model | U-Net (small) |
32
+ | Baseline Dice | 51.9% |
33
+ | Baseline IoU | 57.7% |
34
+ | Baseline Pixel accuracy | 99.66% |
35
+ | Test size (slices) | 387 |
36
+
37
+ ## Usage
38
+
39
+ ```python
40
+ from huggingface_hub import snapshot_download
41
+
42
+ from brain_mri_segmentation.inference.predict import load_model, predict
43
+
44
+ ckpt_dir = snapshot_download("kiselyovd/brain-mri-segmentation")
45
+ model = load_model(f"{ckpt_dir}/best.ckpt")
46
+ result = predict(model, "path/to/slice.png")
47
+ # {"mask": [[0, 0, 1, ...], ...], "shape": [256, 256]}
48
+ ```
49
+
50
+ ## Intended use
51
+
52
+ Research and educational purposes only. **This is not a medical device.** Do not use for clinical decision-making. The model was trained on a single publicly available dataset and has not been validated against clinical ground truth, population diversity, acquisition-device variation, or downstream clinical workflows.
53
+
54
+ ## Source
55
+
56
+ https://github.com/kiselyovd/brain-mri-segmentation
baseline/checkpoints/best.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:067ed818857992d3a5cc481f9cdd403a13022ddf94c1b47621ce2a5d8e3bd3b2
3
+ size 23244634
best.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e318305b3f352691af0bf9fc0970b33224db541dd2f61c73213bf18d3d5ba3f2
3
+ size 328664225
checkpoints/best.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e318305b3f352691af0bf9fc0970b33224db541dd2f61c73213bf18d3d5ba3f2
3
+ size 328664225