Abstract

Computer-Aided Decision is a vital component in modern smart medical care. A fast and accurate automated MRI brain tumor segmentation method is critical for clinical diagnosis and treatment of brain cancer. Convolutional Neural Network-based segmentation method has achieved impressive performance in medical image segmentation with powerful local representation capacities. Nevertheless, they have limitations in modeling global or long-range contextual interactions and spatial dependencies, which are critical for medical image segmentation. In this work, we presented an efficient and lightweight transformer-based Unet for automatic MRI brain tumor segmentation named 3D PSwinBTS, which utilize 3D Parallel Shifted Window-based Transformer module to extract long-range contextual information. Moreover, we utilize semantic supervision to introduce semantic priors in the encoder of 3D PSwinBTS for efficient semantic modeling. To demonstrate the superiority of our proposed method, we compared the performance of our proposed method with state-of-the-art methods on BraTS 2021 dataset, BraTS 2020 dataset and MSD brain tumor dataset. The results demonstrate that our 3D PSwinBTS achieved remarkable performance while computational complexity remains attractive.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call