To investigate the utility of texture analysis of ADC value in differentiating benign and malignant myxoid soft tissue tumors (STTs).

Study subjects

40 consecutive subjects with 40 myxoid STTs (23 benign and 17 malignant) who were pathologically diagnosed with myxoid STTs by tissue samples from image-guided biopsy and/or surgery and underwent musculoskeletal MRI including DWI before the procedure were included. Acquisition of MR data All 40 patients were examined using a 3.0-T MRI system (Achieva TX; Philips Healthcare, Best, The Netherlands). Various radiofrequency coils and MR parameters were used, based on the anatomical location of lesions. A single-shot spin-echo echo-planar DWI sequence was performed in the axial plane. Sensitizing diffusion gradients were sequentially applied in the x, y, and z directions with b values of 0, 400, and 800 s/mm².

Image data analysis

ADC maps were computed by exponential fitting of the local signal intensity using three b values (0, 400, and 800 s/mm²). The region of interest (ROI) was manually drawn on each section of the ADC maps to contain the entire tumor by a single radiologist.

For texture analysis, all DWI images were transferred to a personal workstation and analyzed using in-house software written using MATLAB v. 7.6. (Mathworks, Natick, MA). For global features, a histogram was derived from the distribution of voxels’ ADC values within the ROI. From the histogram, we computed mean, standard deviation, skewness, and kurtosis. For regional and local texture analysis, the ADC values within ROI were resampled to yield 64 discrete values ranging from 1 to 64. For regional features, the grey level size zone matrix’s (m, n) was defined by the number of homogenous regions given the homogeneous tumor size (n) to their intensity (m). From the grey level size zone matrix, the intensity variability and size-zone variability was computed.¹ For local features, the gray level co-occurrence matrix (GLCM) was created for each 13 direction with 1-voxel distance.² To create rotation-invariant GLCM, we make an average of each direction of GLCM, and from the averaged GLCM, 7 local features (energy, entropy, correlation, contrast, homogeneity, variance and maximum probability) were computed. In summary, we extracted 4 global, 2 regional, and 7 local features from the ROI.

Statistical analysis

Student’s t-test was used to test the difference between group means. ANCOVA was performed with adjustments for age, gender and tumor volume.

Regarding global features, malignant myxoid STTs had significantly higher kurtosis (P = 0.040). Regarding local features, malignant myxoid STTs had significantly higher energy, correlation and homogeneity and significantly lower contrast and variance compared with benign myxoid STTs (P = 0.034, < 0.001, < 0.001, 0.003 and 0.001 respectively). Other parameters were not significantly differentTexture analysis of ADC values suggests that ADCs of malignant myxoid STTs are significantly more homogenous compared to those of benign myxoid STTs. This study reveals potential utility of texture analysis, especially local features, of ADC map based on entire tumor volume for differentiating benign and malignant myxoid STTs.