This project applies deep learning techniques to segment liver tumors from CT scan images using the U-Net architecture, a widely-used model in medical image segmentation. The approach focuses on accurately identifying and outlining liver tumors, aiding medical professionals in diagnosis and treatment planning.
Perform semantic segmentation on liver CT scans.
Use U-Net, a convolutional network designed for biomedical image segmentation.
Enhance the model with attention mechanisms to improve segmentation precision.
Evaluate performance using the LiTS17 dataset (Liver Tumor Segmentation Challenge 2017).
Python
TensorFlow / Keras
OpenCV
NumPy, Pandas
Matplotlib, Seaborn
Scikit-image
Dataset: LiTS17 - Liver Tumor Segmentation Challenge
Data format: CT scan images (.nii files) and corresponding ground truth masks.