Texture‐based classification of focal liver lesions on MRI at 3.0 Tesla: A feasibility study in cysts and hemangiomas

Abstract

Purpose: To determine the feasibility of texture analysis for the classification of liver cysts and hemangiomas, on nonenhanced, zero‐fill interpolated T1‐ and T2‐weighted MR images. Materials and Methods: Forty‐five patients (26 women and 19 men; mean age, 58.1 ± 16.9 years) with liver cysts or hemangiomas were enrolled in the study. After exclusion of images with artifacts, T1‐weighted images of 42 patients, and T2‐weighted images of 39 patients, obtained at 3.0 Tesla (T), were available for further analysis. Texture features derived from the gray‐level histogram, co‐occurrence and run‐length matrix, gradient, autoregressive model, and wavelet transform were calculated. Fisher, probability of classification error and average correlation (POE+ACC), and mutual information coefficients were used to extract subsets of optimized texture features. Linear discriminant analysis (LDA) in combination with k nearest neighbor (k‐NN) classification, and k‐means clustering, were used for lesion classification. Results: LDA/k‐NN produced misclassification rates of 16–18% on T1‐weighted, and 12–18% on T2‐weighted images. K‐means clustering yielded misclassification rates of 15–23% on T1‐weighted, and 15–25% on T2‐weighted images. Conclusion: Texture‐based classification of liver cysts and hemangiomas is feasible on zero‐fill interpolated MR images obtained at 3.0T. Further studies are warranted to investigate the value of texture‐based classification of other liver lesions, such as hepatocellular and cholangiocellular carcinoma, on MRI. J. Magn. Reson. Imaging 2010;32:352–359.