Towards End-to-End Image-to-Tree for Vasculature Modeling

Manish Sharma, Matthew C. H. Lee, James Batten, Michiel Schaap, Ben Glocker

Keywords: Vasculature, tree extraction, retinal vessels, diabetic retinopathy, visualization

TL;DR: Extracting tree from medical image scans, and showing how they can be used/visualized.

Abstract: This work explores an end-to-end image-to-tree approach for extracting accurate representations of vessel structures which may be beneficial for diagnosis of stenosis (blockages) and modeling of blood flow. Current image segmentation approaches capture only an implicit representation, while this work utilizes a subscale U-Net to extract explicit tree representations from vascular scans. We obtain insights from these representations by associating tubular thickness with tree edges and visualizing the network of blood vessels from the Digital Retinal Vessel Extraction dataset (DRIVE).

Code Of Conduct: I have read and accept the code of conduct.