Are signal intensity and homogeneity useful parameters for distinguishing between benign and malignant soft tissue masses on MR images? Objective evaluation by means of texture analysis.

摘要: Abstract Objectives To objectively identify possible differences in the signal characteristics of benign and malignant soft tissue masses (STM) on magnetic resonance (MR) images by means texture analysis to determine value these for computer-assisted lesion classification. Method Fifty-eight patients with histologically proven STM (benign, n =30; malignant, =28) were included. was analyzed routine T1-weighted, T2-weighted short tau inversion recovery (STIR) obtained heterogeneous acquisition protocols. Fisher coefficients ( F ) probability classification error average correlation (POE+ACC) calculated most discriminative features separation STM. >1 indicated adequate power features. Based features, k-nearest-neighbor (k-NN) artificial neural network (ANN) performed, accuracy, sensitivity specificity calculated. Results Discriminative only two derived from gray-level histogram STIR (first 10th percentiles). Accordingly, best results achieved using information images, an accuracy 75.0% (sensitivity, 71.4%; specificity, 78.3%) k-NN classifier, 90.5% 91.1%; 90.0%) ANN classifier. Conclusion Texture revealed small MR images. Computer-assisted pattern recognition algorithms may aid characterization STM, but more data is necessary confirm their clinical value.