SA Unet Improved
arXiv:2308.15487v1h-index: 9
Originality Synthesis-oriented
AI Analysis
This addresses the problem of small dataset segmentation in medical imaging for doctors, but appears incremental as it combines existing methods.
The paper tackles retinal vessel segmentation for the DRIVE challenge by combining StyleGAN2 and SA-Unet, achieving improved segmentation to aid in diagnosing eye diseases.
Retinal vessels segmentation is well known problem in image processing on the medical field. Good segmentation may help doctors take better decisions while diagnose eyes disuses. This paper describes our work taking up the DRIVE challenge which include segmentation on retinal vessels. We invented a new method which combine using of StyleGAN2 and SA-Unet. Our innovation can help any small data set segmentation problem.