Authors:

Jingyuan Liang Leone V. Luzzatto

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Classifying complex biological shapes is fundamental to numerous applications, from diagnosing diseases to interpreting cellular functions. Blood vessels are often characterized by highly fractal shapes, exhibiting intricate branching patterns. Research has shown that aging alters blood vessel structure, leading to features such as vessel dilation, tortuosity, microaneurysms, and reduced cerebral blood flow [1]. These structural differences suggest that blood vessels in aging individuals appear distinct from those in younger individuals. However, existing shape comparison methods often struggle to efficiently capture the features of these structures and, therefore, are unable to detect biologically relevant differences between shapes, such as the structural differences caused by aging and diseases. Inspired by the pioneering work of Levitz and Tchoubar on disordered porous solids [2], this project establishes a shape quantification method that utilizes chord lines, which are straight-line segments that connect two points within a shape such that no other point in between belongs to this shape. This computational method includes three stages: image segmentation, chord-length measurement, and distribution comparison. The method is tested on ten mouse retina images belonging to two distinct groups (controls and experiments). The testing revealed measurable differences between the statistical signatures of two groups. It showcases that by transforming complex shapes into statistical distributions, we are able to characterize them geometrically.

Source:

Northwestern University

Topics:

No topics listed

Co-authors:

Jingyuan Liang Leone V. Luzzatto