Feature Selection in Spectroscopy Brain Cancer Data

In cancer diagnosis, classification of the different tumor types is of great importance. An accurate prediction of different tumor types provides better treatment and toxicity minimization on patients. Predicting cancer types using non-invasive information –e.g. \(^1\)H-MRS data– could avoid patients to suffer collateral problems derived from exploration techniques that require surgery. Two Feature Selection Algorithms specially designed to be use in \(^1\)H-MRS Proton Magnetic Resonance Spectroscopy data of brain tumors are presented. These two algorithms take advantage of two distinctive aspects: first, metabolite levels are quite different between types of tumors and two, \(^{1}\)H-MRS data possess a quasi-temporal series shape. Experimental readings on an international data set show highly competitive models in terms of accuracy, complexity and medical interpretability.