Classification of calf muscle MR images by texture analysis

Aim: To evaluate a method of texture analysis (TA) for the description of magnetic resonance (MR) images of healthy and diseased calf muscles and to compare this method with standard radiological evaluation. Methods: A total of 93 subjects (20 controls, seven healthy children of hypertonic parents, five diabetic patients and 61 subjects with muscle malfunction of various origin) underwent MR imaging of the calf muscle and texture analysis of images was performed. The results of TA were analysed by t-statistics and principal component analysis. Images of subjects were divided into four groups according to the assessment of three radiologists and this categorization of subjects was compared with the results from TA. Results: We extracted seven features (from a total number of 282) which were successfully used for the description of the texture of T1w MR images of calf muscles. The results of classification by TA are in 80% agreement with the categorization made by the radiologists. In some cases, TA is able to describe changes not apparent by visual inspection. Conclusion: The TA of MR images of calf muscles can be used for the objective description of changes in muscles and could help radiologists to distinguish between healthy and diseased tissue.