Circle Representation for Medical Object Detection

TL;DR

This paper introduces CircleNet, a novel anchor-free detection framework using circle representation, optimized for biomedical objects like glomeruli, offering improved detection accuracy and rotation invariance over traditional box-based methods.

Contribution

The paper proposes a simple circle representation and a new detection framework, CircleNet, specifically designed for biomedical objects, enhancing detection performance and rotation invariance.

Findings

Superior detection performance on pathological images

Enhanced rotation invariance compared to bounding box methods

Reduced degrees of freedom in object representation

Abstract

Box representation has been extensively used for object detection in computer vision. Such representation is efficacious but not necessarily optimized for biomedical objects (e.g., glomeruli), which play an essential role in renal pathology. In this paper, we propose a simple circle representation for medical object detection and introduce CircleNet, an anchor-free detection framework. Compared with the conventional bounding box representation, the proposed bounding circle representation innovates in three-fold: (1) it is optimized for ball-shaped biomedical objects; (2) The circle representation reduced the degree of freedom compared with box representation; (3) It is naturally more rotation invariant. When detecting glomeruli and nuclei on pathological images, the proposed circle representation achieved superior detection performance and be more rotation-invariant, compared with the bounding box. The code has been made publicly available: https://github.com/hrlblab/CircleNet