Automatic Gleason Grading of Prostate Cancer Using Shearlet Transform and Multiple Kernel Learning

Description: The process is a method of classifying prostate tumors as cancerous or benign. It classifies the tumors according to the Gleason grading scale to determine the cancerous nature of the tumor. The process utilizes a shearlet transform, as well as three other features, and combines them via multiple kernel learning. The shearlet transform is used to represent the local structure of image textures. Multiple kernel learning is used to "fuse" color channel histograms, co-occurrence matrix features, statistics from discrete shearlet coefficients, and morphological features. While current methods of detecting prostate cancer can be inaccurate and often can cause more harm than good, this noninvasive method can detect cancer and classify tumors with remarkable accuracy.