Discriminant Convex Non-negative Matrix Factorization for the classification of human brain tumours

•Brain tumours can be diagnosed on the basis of magnetic resonance spectroscopy (MRS).•A new method to introduce class information into a convex variant of NMF is presented.•Novel techniques for diagnostic predictions of unseen MRS are described.•The new method and techniques are experimentally asse...

Full description

Saved in:
Bibliographic Details
Published in:Pattern recognition letters Vol. 34; no. 14; pp. 1734 - 1747
Main Authors: Vilamala, Albert, Lisboa, Paulo J.G., Ortega-Martorell, Sandra, Vellido, Alfredo
Format: Journal Article Publication
Language:English
Published: Elsevier B.V 15.10.2013
Subjects:
Aplicacions de la informàtica
Bioinformàtica
Brain tumours
Diagnosis
Discriminant Convex Non-negative Matrix Factorization
Informàtica
Magnetic resonance spectroscopy
Models matemàtics
Source separation
Tumors
Àrees temàtiques de la UPC
Magnetic resonance spectroscopy
Discriminant Convex Non-negative Matrix Factorization
Source separation
Brain tumours
ISSN:0167-8655, 1872-7344
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Brain tumours can be diagnosed on the basis of magnetic resonance spectroscopy (MRS).•A new method to introduce class information into a convex variant of NMF is presented.•Novel techniques for diagnostic predictions of unseen MRS are described.•The new method and techniques are experimentally assessed with real MRS data.•The new methods are predictive and generate very tumour type-specific MRS sources. The medical analysis of human brain tumours commonly relies on indirect measurements. Among these, magnetic resonance imaging (MRI) and spectroscopy (MRS) predominate in clinical settings as tools for diagnostic assistance. Pattern recognition (PR) methods have successfully been used in this task, usually interpreting diagnosis as a supervised classification problem. In MRS, the acquired spectral signal can be analyzed in an unsupervised manner to extract its constituent sources. Recently, this has been successfully accomplished using Non-negative Matrix Factorization (NMF) methods. In this paper, we present a method to introduce the available class information into the unsupervised source extraction process of a convex variant of NMF. Novel techniques to generate diagnostic predictions for new, unseen spectra using the proposed Discriminant Convex-NMF are also described and experimentally assessed.
ISSN:0167-8655
1872-7344
DOI:10.1016/j.patrec.2013.05.023