Portal de Periódicos da CAPES

Extraction of Characteristic Breast Dimensions of North Vietnamese Girl-Students Using Random Forest Algorithm

2022; Springer International Publishing; Linguagem: Inglês

10.1007/978-981-19-3394-3_25

2367-3370

Nguyen Thi Le, Luu Thi Hong Nhung, Nguyen Nhat Trinh,

Infrared Thermography in Medicine

Women’s breasts are measured with many different dimensions, which are quite diverse and complicated depending on the purpose of use. The aim of this study is to select the typical dimensions of girl-student breasts to obtain important dimensions for breast classification, thereby reducing noise, reducing costs, and contributing to building a basis for improving the sizing system, designing, and choosing the right woman’s bras. 460 girl students in Northern Vietnam between the ages of 18 and 24 with a body mass index of 14.5–24.3 kg/m2 are selected in this study. Their breasts were 3D scanned to identify 3 parameters of the body (height, weight and body mass index) and 18 breast parameters. The Principal Component Analysis method, Random Forest, and Learning Vector Quantization algorithms are applied to extract the characteristic breast dimensions. The results show that the accuracy reaches 0.9615, the sensitivity reaches 0.989 and the specificity reaches 0.977 with the Random Forest method. The most important dimensions for breast classification are breast volume, the difference between bust and underbust circumference, outer breast, bust, upper bust, inner breast, and bust circumference. Height, weight, body mass index, prolapse, and thoracic arch are the least important parameters when classifying girl-student breasts.