ABSTRACT
Objective:To investigate the value of histogram analysis of ultrasound gray scale in differential diagnosis of triple negative breast invasive ductal carcinoma (TN-IDC) and non-triple-negative breast invasive ductal carcinoma (NTN-IDC).Methods:Totally 195 patients with invasive ductal carcinoma confirmed by pathology in Hubei Cancer Hospital from September 2017 to July 2020 were retrospectively analyzed.According to immunohistochemical results after surgery, 195 patients were divided into TN-IDC group ( n=44) and NTN-IDC group ( n=151). All cases were retrospectively analyzed to observe the ultrasonoscopy histogram features of tumors and obtain the histogram parameters, including mean, variance, skewness, kurtosis and percentile gray-scale values. The histogram parameters of TN-IDC were compared with those of NTN-IDC. The ROC curves were constructed to observe the efficiency of differential diagnosis. Results:The values of variance, 90th and 99th percentiles in TN-IDC group were much lower than those in NTN-IDC group (all P<0.05). The values of mean, skewness, kurtosis and other percentile (1st, 10th, and 50th) values between the two groups were not significantly different(all P>0.05). The optimal cutoff value for the accurate identification of TN-IDC and NTN-IDC groups was 552.85 for variance[sensitivity 75.0%, specificity 79.5%, positive predictive value(PPV) 51.6%, negative predictive value(NPV) 91.6%, accuracy 78.5%, and area under curve 0.829, respectively]. The optimal cutoff value for the accurate identification of TN-IDC and NTN-IDC groups was 74 for 90th percentile (sensitivity 68.2%, specificity 57.6%, PPV 31.9%, NPV 86.1%, accuracy 60.0%, and area under curve 0.648, respectively). The optimal cutoff value for the accurate identification of the two groups was 107 for 99th percentile (sensitivity 75.0%, specificity 77.5%, PPV 49.3%, NPV 91.4%, accuracy 76.9%, and area under curve 0.772, respectively). Conclusions:Histogram analysis of ultrasound gray scale can provide certain value for the differential diagnosis of TN-IDC and NTN-IDC, the variance and 99th percentile values could perform better.