Radiomics analyses are being increasingly employed to investigate tissue heterogeneity present within the prostate gland. We present a method for improving the repeatability of radiomics features extracted from T2-weighted images using a deep normalization technique based on fully convolutional networks (FCNs). We test the repeatability of select radiomics features on a previously published test-retest prostate dataset. We demonstrate that the intraclass correlation coefficient of first-order statistics features extracted from images normalized using the FCN-based pre-processor is consistently higher than for features extracted from non-normalized images.
This abstract and the presentation materials are available to members only; a login is required.