Breast Cancer Diagnostic Decisions from Multi-Source Data

To evaluate the efficacy of support vector machines (SVM) in diagnostic decisions on the benign or malignant nature of the ultrasound Breast Imaging Reporting and Data System (BI-RADS) category 4 breast nodules in a multi-source diagnostic context.

An experimental study. Place and Duration of the Study: Department of Ultrasound Imaging, Yichang Central People's Hospital, Yichang, China, from January 2020 to November 2023.

This study involved patients with ultrasound BI-RADS category 4 breast nodules. Conventional ultrasound diagnostics, S-Detect technology, and medical quasi-intelligent software were used to analyse the pre-treatment ultrasound results with pathological diagnoses serving as the reference standard for accuracy. Principal component analysis (PCA) was applied to extract the principal components from the multi-source breast imaging parameters, which were then integrated with SVM for evaluating its feasibility in classifying category 4 breast nodules as benign or malignant across various breast imaging modalities.

PCA extracts two principal components from a 12-dimensional feature parameter matrix measured from the multi-source breast imaging. The SVM, when combined with PCA, demonstrated a high level of reliability in multi-source breast cancer diagnostics, achieving a decision accuracy rate of 94.5%.

The integration of SVM with PCA principal component analysis has proven to be highly valuable in the diagnostic decision- making process for the multi-source breast cancer diagnostics, offering a robust method for distinguishing between benign and malignant category 4 breast nodules.

Support vector machine, Principal component analysis, Breast imaging, Multi-source diagnostics.