Deep Learning based Segmentation of Optical Coherence Tomographic Images of Human Saphenous Varicose Vein

This work addresses the need for automated segmentation in medical imaging for varicose vein analysis, but it appears incremental as it builds on existing U-Net architectures with modifications.

The researchers tackled the problem of segmenting optical coherence tomography images of human varicose veins by proposing a deep-learning model based on U-Net with atrous convolution and residual blocks, achieving an accuracy of 0.9932.