Meeting Banner
Abstract #0828

Automatic Lesion Detection in White & Grey Matter Using T1-Weighted and FLAIR Images

Javier Gonzlez-Zabaleta1, Norberto Malpica1, Ana Ramos2, Juan lvarez-Linera3, Juan Antonio Hernndez-Tamames1

1Neuroimaging lab., Center for Biomedical Technology, Universidad Politcnica de Madrid and Universidad Rey Juan Carlos, Pozuelo de Alarcn, Madrid, Spain; 2Hospital 12 de Octubre, Madrid, Spain; 3Hospital Ruber Internacional, Madrid, Spain

Cortical lesions in multiple sclerosis are characterized by simultaneous hiperintensity in FLAIR images and hypointensity in T1 images. We have developed a method for automatic lesion detection that includes tissue segmentation correction, lesion detection and quantification. We have evaluated the technique in a study on Multiple Sclerosis, including 21 controls and 15 patients. All subjects were correctly classified with our method. All the algorithms have been included in a plugin for 3D Slicer.

Abstract PDF Poster PDF

Keywords

allowed appear apply applying artificial assessed automate automatic available axes background biography biomedical brain brains classical classified classify classifying clinical combining command compute consider controls correction cortical detected detecting detection developed directly distributions edition essence evaluate executed extracting fail fifteen finally flair gold graphical gray guide hospital human identification identified impact improve inside insight intense intensity interface international intervention involving keep label least lesion lesions load local macro manual mapping mike morphological noise normalized numerous objects parametric partial particular patients platform plug priori problem procedure quantification refine registration remove reported requires residual respect rest routine sclerosis segmentation segmented sensibility simultaneous simultaneously slicer software specificity staging statistical still studies subjects symposium system table technology thresholding tissue tool toolkit twenty usually version vision visual volume volumes white