Metadata-Version: 2.1
Name: medseglib
Version: 1.0.0
Summary: A package to segmentation tumor on CT and MRI
Home-page: UNKNOWN
Author: Ilya Zarudin
Author-email: zarilja@yandex.ru
License: UNKNOWN
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3.8
Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
Description-Content-Type: text/markdown
Requires-Dist: torch (>=2.0.1)
Requires-Dist: torchvision (>=0.15.2)
Requires-Dist: urllib3 (==1.26.6)

# MEDSEGLIB

This package designed for segmentation of oncology by CT and MRI images.

### Models
Currently, the models DeepLabV3(bb: resnet50), DeepLabV3(bb: resnet101) are used.
During the training, these models were used, as well as UNet and TransUnet, but DeepLabV3 showed the best results (dice and iou).

### Supported types of research
* Brain
* Liver

### How to use
```python
import numpy as np
from megseglib.sdk import Model

ct = np.zeros((1, 3, 512, 512), dtype=np.single)
model = Model(model_name="deeplab_v3_resnet101", model_type="liver")

result = model.forward(ct)
# or
result = model(ct)

print(type(result), result.shape) # <class 'numpy.ndarray'> (1, 1, 512, 512) - segmentation mask 
```

