Table 2 A brief description of the intensity and radiomic features and their significance in quantifying the appearance of prostate cancer on MRI
Feature category | Description | Significance |
|---|---|---|
Signal intensity | T2w, ADC signal intensity | Regions of low signal intensities are indicative of prostate cancer lesions |
First-order statistics | Mean, standard deviation, median and range; first order differentials computed using Sobel operators | Localize regions with significant intensity changes; Gradients detect edges and quantify region boundaries |
Co-occurrence features | A co-occurrence matrix is a compilation of the different spatial relationships of neighboring pixel intensities. Several metrics of this matrix are computed to obtain useful features, such as Haralick features | Distinguish homogeneous regions of low intensities of cancer from hyper-intense normal regions |
Gabor features | A Gabor filter is a convolution of Gaussian function with a Fourier transform at different orientations and frequencies | Gabor features quantify the appearance of cancer lesions at multiple orientations and image scales |
Texture energy | Laws’ texture energy features are computed by convolution of the image with local masks that are obtained from vectors which capture local average, edge, spot, wave and ripple patterns | Texture energy quantifies the variation of pixel intensities within a fixed region of the image. Regions of image containing cancer lesions typically contain lower texture energy |