Skip to main content

Table 3 Performance of different feature combinations using the multilayer perceptron method

From: Performance of machine learning software to classify breast lesions using BI-RADS radiomic features on ultrasound images

Features Sensitivity (%) Specificity (%) Area under the curve
{5} 67.5 78.7 0.759
{5, 2} 68.4 79.2 0.789
{5, 2, 1} 68.8 84.1 0.799
{5, 2, 1, 3} 66.2 71.7 0.806
  1. {1}: area difference with equivalent ellipse; {2}: orientation; {3}: average of difference vector; {5}: average of distance vector