Browse

Usefulness of Quantitative Voxel-based Morphometry of MRI and FDG-PET in detection of Focal Cortical Dysplasia

Cited 0 time in Web of Science Cited 0 time in Scopus
Authors
이소정
Advisor
정천기
Major
자연과학대학 뇌인지과학과
Issue Date
2018-02
Publisher
서울대학교 대학원
Keywords
Focal cortical dysplasia (FCD)Voxel based morphometry (VBM)Magnetic resonance imaging (MRI)FDG-PET
Description
학위논문 (석사)-- 서울대학교 대학원 : 자연과학대학 뇌인지과학과, 2018. 2. 정천기.
Abstract
Its important to detect the epileptogenic lesion in pre-surgical evaluation for refractory epilepsy surgery, since non-lesional epilepsy is known as a predictor of surgical failure. Focal Cortical Dysplasia (FCD) constitutes the most common non-lesional epilepsy. Voxel-based morphometry (VBM) of MRI has been useful in enhancing visualization of FCD compared to the conventional visual analysis. However, previous studies were based on qualitative methods to localize the lesion prone to subjectivity, depending on person. Thus, we investigated the usefulness of applying thresholds quantitatively through VBM of MR T1 and T2-weighted image in objective way rather than subjective method. Also, T1-weighted image and T2-weighted image are known to be sensitive to different structural information, since they have different relaxation time and echo time. Moreover, FDG-PET detects the metabolic abnormality of FCD as a hypometabolic lesion, while MRI detects FCD structural abnormality. As different modalities have different sensitivities of various FCD characteristics, multimodal analysis in diagnosis may improve detection sensitivity of neuronal disease. As a result, we investigated additive values of combining quantitative voxel-based analysis of T1-weighted image, T2-weighted image and FDG-PET in FCD detection.

55 patients who underwent epilepsy surgery and histologically proven as FCD between 2004 and 2016 (30 female, mean age at MRI 29.5 years, range 16-55 years) at Seoul National University Hospital were retrospectively recruited. Preoperative T1, T2 MRIs and FDG-PET were analyzed with voxel-based methods in SPM 12, Matlab. The gray-white matter junction of each patient was extracted and compared to those from healthy controls (26 subjects including 14 female, mean age at MRI 33 years, range 23-45 years). Thereafter, the junction image was made to highlight the blurred gray-white matter. Two observers chose potential lesion in each junction image, which made the blurred region stand out. For quantitative analysis, we applied threshold and among the over-threshold clusters, the 3 clusters with maximum size of voxels were selected as candidates of the lesion.
FDG-PET data were compared to those in 50 normal controls (37.8 years), using two sample t-test (p<0.001). Among the clusters over the threshold, the area with maximum standard deviation was selected as a candidate lesion. The area chosen in MRI and PET analysis was compared with the area surgically resected on post-operative MRIs, taking the surgical outcome into account.

Out of 46 patients whose seizure control after the surgery showed improvement, visual assessment alone in MRIs detected FCD in 16 patients. However, by voxel-based analysis, the FCD lesions were found in 21 patients. Furthermore, when conventional method and voxel-based method were combined, FCD lesions were detected in 26 patients with MRI T1 and T2-weighted image.
Consideration of MRI and PET images together detected 30 patients lesion correctly, while MRI alone detected 26 patients lesion with combination of conventional visual analysis and voxel-based morphometric method. Furthermore, the more concordant suspected lesion in between the modalities, the better the surgical outcome presented. There were no significant differences in sensitivity concerning the gender, age at surgery and scanners difference respectively.
Our results suggest that voxel based analysis based on the blurring of gray and white matter junction not only helps visualization qualitatively, but also can be used quantitatively additional to the conventional visual analysis (p<0.000). Also, consideration of both MRI and PET leads better detection of FCD with good surgical outcome.
Language
English
URI
https://hdl.handle.net/10371/142427
Files in This Item:
Appears in Collections:
College of Natural Sciences (자연과학대학)Brain and Cognitive Sciences (뇌인지과학과)Theses (Master's Degree_뇌인지과학과)
  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse