Hugging Face's logo

hanjang
/
Interactive-MEN-RT

Image Segmentation
PyTorch
medical-image-segmentation
interactive-segmentation
nnunet
meningioma
radiotherapy
Model card Files
xet
Community
Interactive-MEN-RT
File size: 2,399 Bytes 
b2fdf3c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
390fc8c
b2fdf3c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
 
---
library_name: pytorch
license: apache-2.0
tags:
- medical-image-segmentation
- interactive-segmentation
- nnunet
- meningioma
- radiotherapy
pipeline_tag: image-segmentation
---

# Interactive-MEN-RT

**Domain-Specialized Interactive Segmentation for Meningioma Radiotherapy Planning**

**Status**: Research prototype — Not for clinical use.

## Overview

Interactive 3D meningioma segmentation framework built on nnU-Net v2 and nnInteractive.

- **Performance**: 77.6% Dice, 64.8% IoU (BraTS 2025 Meningioma RT)
- **Interaction Modes**: Point, scribble, box, lasso
- **Input**: T1c MRI (contrast-enhanced)

## Files
```
nnUNetInteractionTrainer__nnUNetPlans__3d_fullres_scratch/
├── plans.json
├── dataset.json
└── fold_0/
    ├── checkpoint_best.pth (820 MB)
    └── checkpoint_final.pth (820 MB)
```

## Quick Start
```python
from huggingface_hub import snapshot_download
import os

# Download checkpoint
root = snapshot_download(
    "hanjang/Interactive-MEN-RT",
    allow_patterns=["nnUNetInteractionTrainer__nnUNetPlans__3d_fullres_scratch/**"]
)
CKPT = os.path.join(root, "nnUNetInteractionTrainer__nnUNetPlans__3d_fullres_scratch")

# Load and run inference
from Interactive_MEN_RT_predictor import InteractiveMENRTPredictor
import torch, numpy as np

predictor = InteractiveMENRTPredictor(
    device=torch.device("cuda" if torch.cuda.is_available() else "cpu")
)
predictor.initialize_from_trained_model_folder(
    model_training_output_dir=CKPT,
    use_fold=0,
    checkpoint_name="checkpoint_best.pth"
)

# Run on your volume (shape: 1, H, W, D)
predictor.reset_interactions()
predictor.set_image(volume)
predictor.set_target_buffer(np.zeros_like(volume[0], np.float32))
predictor._finish_preprocessing_and_initialize_interactions()
predictor._predict_without_interaction()
prediction = (predictor.target_buffer > 0.5).astype(np.uint8)
```

# Citation

```bibtex
@inproceedings{interactive-men-rt-2025,
  title={Domain-Specialized Interactive Segmentation Framework for Meningioma Radiotherapy Planning},
  author={Junhyeok Lee, Han Jang and Kyu Sung Choi},
  booktitle={MICCAI CLIP Workshop},
  year={2025}
}
```

# Links

GitHub: [snuh-rad-aicon/Interactive-MEN-RT](https://github.com/snuh-rad-aicon/Interactive-MEN-RT)

Contact: janghan001112@gmail.com

Developed at Seoul National University AICON Lab

Research only. Not for clinical use.

---