Apppublic

hung2903/synapse-unet-demo

sourceHugging Facemitupdated 8mo agoView on Hugging Face

0likes

App README

Synapse Multi-Organ Segmentation Demo

(Lightweight U-Net)

This model is a lightweight U-Net trained on the Synapse multi-organ CT dataset. It is designed to run on low-resource hardware (GTX 1650, 4GB VRAM) and CPU.

Model Details

—Architecture: U-Net (Filters: 32, 64, 128, 256)
—Input: 256x256 Axial CT Clices (Windowed [-125, 275], Normalized [0,1])
—Classes: 9 (Background + 8 organs)

Usage

Run with the provided app.py in the linked GitHub repository.

python

model = UNet(n_channels=1, n_classes=9)
model.load_state_dict(torch.load("best_model.pth"))

Dataset

Synapse Multi-Organ Segmentation Challenge. License: Review original dataset license.