View
466
Download
2
Tags:
Embed Size (px)
DESCRIPTION
In this paper, we introduce a lesion abnormality score based on computerized comparison of the 3D texture properties between brain hemispheres in T1 MRI. Overlapping cubic texture blocks extracted from user–defined 3D regions of interest (ROI) are expressed in terms of energies of 3D steerable Riesz wavelets. The abnormality score is defined as the Hausdorff distance between the ROI and its corresponding contralateral region in the brain, both expressed as ensembles of blocks in the feature space. A classification based on the proposed score allowed an accuracy of 85% with 10 control subjects and 8 patients with epileptogenic lesions. The approach therefore constitutes a valuable tool for the objective pre–surgical evaluation of patients undergoing epilepsy surgery.
Citation preview
Manual annotation Contralateral ROI Image Registration
Texture analysis using 3D Riesz wavelet transforms is performed in 323 blocks inside the ROIs. Riesz energy coefficients are computed as an objective texture measure.
Results • The Hausdorff distance between the affected and unaffected
side in the brain is compared with the average textural difference in control subjects.
• For 6 out of 8 patients a higher abnormality score obtained.
Control ROIs Epileptogenic lesions
Epileptogenic Lesion Quantification in MRI Using Contralateral 3D Texture Comparisons
Contact and more information: [email protected], http://iig.hevs.ch/
Oscar Alfonso Jiménez del Toro1, Antonio Foncubierta-Rodríguez1, María Isabel Vargas Gómez2, Henning Müller1,2 and Adrien Depeursinge1,2
1University of Applied Sciences Western Switzerland (HES–SO), Sierre, Switzerland; 2University and University Hospital of Geneva, Switzerland
Summary 1. Adresses the important clinical problem of identifying the epileptic focus in patients prior to surgery. 2. The approach compares the 3D texture measured using 3D Riesz wavelet transforms of suspicious brain tissue
with the contralateral hemisphere
Introduction • Lesions may not be apparent simply by visual inspection of
structural MRI so a more sophisticated analysis is required. • Precise localization of the lesions has a strong influence on the
outcome of epilepsy surgery. • MRI image processing methods have outperformed visual
assesment. • These methods can still rely on subjective visual confirmation
and personal experience of the interpreter.
Dataset • 8 cases with seizures caused by focal lesions like
dysplasia, cavernoma, ganglioglioma and tuberous sclerosis.
• Complete MRI series including T1weighted, T2-FLAIR and DTI. Lesions sometimes visible only in 1 sequence
Methods • Epileptogenic lesion were manually delineated on T2 or DTI
MRI (where the lesion was most visible) by a neuroradiologist for 8 patients.
• Affine registrations between the hemispheres allowed to compute the contralateral anatomical region.
• For each patient, both the lesion and the contralateral regions were registered to 10 control subjects.
• The method is applied to compute an "abnormality score" defined as the Hausdorff distance computed in the feature space between a possible epileptogenic lesion and the corresponding contralateral healthy anatomical region.
• Textural difference is then compared to the average textural difference in the same anatomical regions, in healthy subjects