Article Text

PDF
THU0335 Support vector machines classification of texture parameters of white matter lesions in childhood-onset systemic lupus erythematosus. Possible mechanism to distinguish between demyelination and ischemia
  1. A.T. Lapa1,
  2. M.P. Bento1,
  3. L. Rittner1,
  4. H.H. Ruocco1,
  5. G. Castellano1,
  6. B.P. Damasceno1,
  7. L.T.L. Costallat1,
  8. R. Lotufo1,
  9. F. Cendes1,
  10. S. Appenzeller2
  1. 1State University of Campinas
  2. 2Medicine, State University of Campinas, Campinas, Brazil

Abstract

Background Texture analysis (TA) is a branch of image processing which seeks to reduce image information by extracting texture descriptors from the image. White matter hyperintensities (WMH) are frequently observed in childhood-onset systemic lupus erythematosus (cSLE); however the etiology is still unknown. Ischemic and demyelination have been proposed as possible etiologies. Support vector machines (SVM) are a group of supervised learning methods that can be applied to classification or regression

Objectives To determine etiology of WMH in cSLE patients using texture analysis of magnetic resonance (MR) parameters basedon neural network

Methods TA was applied to axial FLAIR magnetic resonance images (MRI) in 43 patients with cSLE (mean age 17.25 years (SD ± 3.57), 30 patients with multiple sclerosis (MS), 30 patients with stroke and 30 normal age and sex-matched controls. The TA approach used was based on the Gray Level Co-occurrence Matrices (GLCM). The WMH were manually segmented for each subject, classified in periventricular, subcortical, deep white matter and cortical WMH and 256 texture parameters were computed for each lesion. A SVM classifier previously developed and validated (accuracy 93%) was used to classify WMH in patients with cSLE. Nature of the classified WMH, demographic, clinical and laboratory features were included in a regression model to determine which variables could support the possible nature of WMH in clinical practice

Results In cSLE, of the 125 of periventricular lesions, 59% were classified as ischemic and 41% as demyelinating. Of the 738 subcortical lesions 63% were classified as ischemic and 37% as demyelinating. Of the 64 cortical lesions, 61% were classified as ischemic and 39% as demyelinating. Of the deep white matter WMH, 72% were classified as ischemic and 28% as demyelinating in nature. In controls all lesions were classified as ischemic. Positive antiphospholipid antibodies (aPL) (OR=3.2; 95%CI 1.2-7.3) and higher total corticosteroid dose (OR=2.1; 95%CI 1.2-3.7) were variables associated with stroke, whereas cutaneous vasculitis (OR=4.2; 95%CI 2.1-7.2), anti-SM antibodies (OR=2.9; 95%CI 1.3-8.7) and disease activity (OR=3.7; 95%CI 1.3-7.8) were associated with demyelination in all cerebral regions.

Conclusions In cSLE, the majority of WMH were classified as ischemic in nature; however aproximately 30% of the lesions were demyelinating. Positive aPL and higher total corticosteroid dose were variables associated with stroke, whereas cutaneous vasculitis, anti-SM antibodies and disease activity were associated with demyelination in all cerebral regions.SVM of TA is a useful method to help to determine etiology of WMH in cSLE.

Disclosure of Interest A. Lapa Grant/Research support from: FAPESP 2010/13639-1, M. Bento: None Declared, L. Rittner: None Declared, H. Ruocco: None Declared, G. Castellano: None Declared, B. Damasceno: None Declared, L. Costallat: None Declared, R. Lotufo: None Declared, F. Cendes: None Declared, S. Appenzeller Grant/Research support from: FAPESP 2008/02917-0; Conselho Nacional Pesquisa Desenvolvimento-Brasil CNPq (300447/2009-4)

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.