X-ray texture analysis in differential diagnosis of phyllodes tumor and fibroadenoma in breast / 中国医学影像技术

ABSTRACT

Objective: To investigate the value of X-ray texture analysis in differential diagnosis of breast phyllodes tumors (BPTs) and breast fibroadenoma (BF). Methods Preoperative X-ray data of 32 patients with BPTs and 32 patients with BF were retrospectively analyzed. MaZda 4.6 software was used to manually draw ROIs on the maximum level of mammography images. The optimized texture parameters of texture analysis were selected with Fisher coefficient, probability of classification error and average correction coefficient (POE+ACC), mutual information (MI) as well as combination of the above three methods (MI+PA+F), respectively. The diagnostic results obtained by radiologist or using texture analysis techniques were contrasted with pathology. The misclassification rates of BPTs and BF were compared between texture analysis and subjective diagnosis by radiologist. Results: The misclassification rate of subjective diagnosis, Fishers, POE-ACC, MI and MI+PA+F was 20.31%(13/64), 14.06% (9/64), 23.44% (15/64), 25.00% (16/64) and 3.12% (2/64), respectively. The misclassification rate of MI+PA+F was lower than that of subjective diagnosis, POE-ACC and MI, respectively (all P<0.05). Conclusion: It is feasible to use texture analysis on X-ray films for differentiation of BPTs and BF.