Gastro-Intestine Tract Image Segmentation using Semantic Segmentation and U-Net: Enhancing Cancer Diagnosis and Treatment

In this project, a deep learning model based on the U-Net architecture was created for image segmentation of MRI scans.The dataset comprised MRI scans collected over six months from patients undergoing MRI-guided radiotherapy.A segmentation model was developed and implemented to accurately differentiate various structures in the gastrointestinal (GI) tract.In other word, to develop a model that can automatically identify and differentiate the stomach, large intestines, and small intestines on MRI scans for cancer patients.The model's performance was evaluated using appropriate metrics and compared with existing methods.The potential of the model in enhancing cancer diagnosis and treatment through integration into medical imaging systems was explored.The project successfully implemented the UNet model that achieved satisfactory results using the Dice score and IOU score to evaluated the performance of the model, the Dice score reached to 0.8833 on the training and 0.8533 validation sets, respectively, and an IOU score of 0.8233on the training and 0.7933 on validation sets,.