Discriminating Value on Molecular Typing of Breast Cancer by Ultrasonic Elastic Imaging Intelligent Diagnosis System Based on Deep Learning
ZHAO Jie
FAN Ruirui
LIU Huaqing
Abstract：Objective To construct an ultrasonic elastography intelligent diagnosis system based on deep learning,and analyze the value of the intelligent diagnosis system in the differentiation of pathological types of breast cancer.Methods The clinical data of 595 patients with breast cancer diagnosed in Zhengzhou Yihe Hospital from January 2018 to December 2022 were retrospectively collected.Before treatment,shear wave elastography was performed,and there were clear molecular typing results.The ultrasonic elastic imaging intelligent diagnosis system based on deep learning was constructed by Resnet convolutional neural network using shear wave elastic imaging inspection information and molecular typing information.All patients were randomly divided into a training set and a validation set according to 4∶1 to train and validate the intelligent diagnosis system.Results The shear wave elastic modulus parameters of different molecular types of breast cancer patients were significantly different(P＜0.05).There were 29 cases of Luminal A type,52 cases of Luminal B type,13 cases of HER-2 overexpression type,and 25 cases of triple negative type among 119 patients in the validation set.The ultrasonic elastic imaging intelligent diagnosis system based on deep learning could output the molecular typing diagnosis results of breast cancer,and the validation set output 26 cases of Luminal A type,53 cases of Luminal B type,12 cases of HER-2 overexpression type,and 28 cases of triple negative type through the intelligent diagnostic system.The sensitivity of the intelligent diagnosis system in diagnosing Luminal A,Luminal B,HER-2 overexpression,and three negative types of breast cancer was 68.97％,78.85％,61.54％,and 84.00％,respectively.The specificity and accuracy were higher than 80％,and it had good consistency with the gold standard(Kappa=0.648,P＜0.001).Conclusion The ultrasonic elastography intelligent diagnosis system based on deep learning can be used for molecular typing and classification of breast cancer,and it has good value for molecular typing.
Keywords：breast cancerdeep learning algorithmultrasound elastographyintelligent diagnostic systemmolecular typing
Publication Date：2025-06-13
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：5( 1990-1994 )