2023 167 CSM

Multislice Multimodality Tumor Segmentation

By: Abdoulfatah Abdillahi, Brandon Bahr Maravilla, Yoonji Lee

Department: Computer Science 

Faculty Advisor: Dr. Ilmi Yoon 

In our project, we are using a U-Net model architecture to segment whole tumor by utilizing multiple slices from T1, T2, T1CE and Flair modalities. We are utilizing the BraTS (Brain Tumor Segmentation) dataset to determine which modality is the best results for the whole tumor segmentation. Based on the best modality, we will then do multi-class segmentation of the tumors. Our team has efficiently distributed the workload, with each member focusing on one modality to enhance the accuracy of whole tumor segmentation.