Breast Symmetry Classification: A Proposal of Image Analysis Based Indices

The study aims to develop a method for evaluating breast symmetry using profile images. This shape-based analysis assesses differences between the two profiles with the help of simple indices, enabling an external symmetry evaluation independent of the image acquisition method. Initially, thermograms of 62 volunteers were analyzed, leading to the proposal of four asymmetry indices based on measurements of breast profile. Subsequently, images from 30 additional volunteers were examined to determine whether a multimodal pattern emerged across different imaging techniques, including infrared (IR) cameras, three-dimensional (3D) scan models, and photographic (RGB) cameras. To validate the method and assess the feasibility of applying the proposed indices to these imaging methods, the classified images (as symmetric or asymmetric) were evaluated by 28 specialists. The comparison between the proposed indices and expert assessments resulted in an accuracy rate of 60%.