Analysis of Brain White Matter Hyperintensities using Pattern
Recognition Techniques
Mariana Bento
1
, Leticia Rittner
1
, Simone Appenzeller
2
, Aline Lapa
2
, Roberto Lotufo
1
1
School of Electrical and Computer Engineering, University of Campinas - UNICAMP, 13083-852,
Campinas (SP), Brazil
2
Department of Medicine, Rheumatology Unit, University of Campinas - UNICAMP, 13083-970,
Campinas (SP), Brazil
maripb@dca.fee.unicamp.br
ABSTRACT:
The brain white matter is responsible for the transmission of electrical signals through the central nervous system.
Lesions in the brain white matter, called white matter hyperintensity (WMH), can cause a significant functional deficit.
WMH are commonly seen in normal aging, but also in a number of neurological and psychiatric disorders. We propose
here an automatic method for WHM analysis in order to distinguish regions of interest between normal and non-normal
white matter (identification task) and also to distinguish different types of lesions based on their etiology: demyelinating
or ischemic (classification task). The method combines texture analysis with the use of classifiers, such as Support
Vector Machine (SVM), Nearst Neighboor (1NN), Linear Discriminant Analysis (LDA) and Optimum Path Forest
(OPF). Experiments with real brain MRI data showed that the proposed method is suitable to identify and classify the
brain lesions.
KEYWORD LIST: White Matter Hyperintensity; Brain White Matter; Magnetic Resonance Imaging; Lesions Etiology;
Demyelinating; Ischemic; Texture analysis; Classifiers
1. INTRODUCTION
White matter hyperintensities (WMH) are commonly found in brain Magnetic Resonance Imaging (MRI) in
both asymptomatic and neurologic symptomatic patients [1]. Their etiology varies according to age, but ischemic and
demyelinating are more frequently observed. The analysis of white matter intensities in the brain through MRI, however,
is a non-trivial task, due to the complexity of underlying factors: variable staining procedures and practices, diversity in
imaging devices, and the ultimate goal of the analysis. The specialist usually takes into account additional clinical
information from patients, such as age, physical exam and history and also medical images from different modalities to
manually accomplish the classification task. Thus, in order to automatically analyse white matter hyperintensities it is
necessary to combine methods from different research areas, such as digital image analysis and pattern recognition.
Klöppel presents a comparison of different methods for the detection of WMH in MRI based on intensity
features and Support Vector Machine (SVM) and k-nearest neighbor (kNN) classifiers [2]. Anbeek proposes a method to
automated segmentation of white matter lesions in cranial MR images [3]. The method generates probability maps
representing the probability per voxel being part of a white matter lesion. Wu, on the other hand, proposes a fully
automatic method for quantifyng and localizing lesions in the brain white matter on MRI in the elderly [4]. Another work
compares automatic methods to detect multiple sclerosis lesions in the brain MR images [5].
Those and other related works in the literature generally propose methods to automatically identify white matter
lesions caused by a specific disease with a known etiology. There is no work in the literature that combines image
processing and pattern recognition techniques to analyze lesions according to their etiology.
We present here a technique based on texture analysis and a classification procedure to distinguish between
normal and non-normal white matter tissue, denoted identification task, so as to distinguish white matter lesions based on
their etiology, called classification task. Texture analysis is a branch of image processing [6] that has been used in many
Medical Imaging 2013: Image Processing, edited by Sebastien Ourselin, David R. Haynor,
Proc. of SPIE Vol. 8669, 86693P · © 2013 SPIE · CCC code: 1605-7422/13/$18
doi: 10.1117/12.2006924
Proc. of SPIE Vol. 8669 86693P-1
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 04/05/2013 Terms of Use: http://spiedl.org/terms