Related Experiment Video
Updated: Dec 23, 2025
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Diagnostic Performance of the Support Vector Machine Model for Breast Cancer on Ring-Shaped Dedicated Breast Positron
Yoko Satoh, Daiki Tamada1, Yoshie Omiya1
1Department of Radiology, University of Yamanashi, Shimokato, Chuo City, Yamanashi Prefecture, Japan.
Objective:
The aim of this study was to evaluate the diagnostic ability of support vector machine (SVM) for early breast cancer (BC) using dedicated breast positron emission tomography (dbPET).
Methods:
We evaluated 116 abnormal fluorodeoxyglucose (FDG) uptakes less than 2 cm on dbPET images in 105 women. Fluorodeoxyglucose uptake patterns and quantitative PET parameters were compared between BC and noncancer groups. Diagnostic accuracy of the SVM model including quantitative parameters was compared with that of visual assessment based on FDG-uptake pattern.
Results:
Age, maximum standardized uptake value, peak standardized uptake value, total lesion glycolysis, metabolic tumor volume, and lesion-to-contralateral background ratio were significantly different between BC and noncancer groups. Area under the curve, sensitivity, specificity, and accuracy for FDG-uptake pattern of visual assessment were 0.77, 0.57, 0.77, and 0.71, respectively; those of an SVM model including age, maximum standardized uptake value, total lesion glycolysis, and lesion-to-contralateral background ratio were 0.89, 0.94, 0.77, and 0.85, respectively.
Conclusions:
Support vector machine showed high diagnostic performance for BC using dbPET.
