TEXTURAL CLASSIFICATION OF MULTIPLE SCLEROSISLESIONS IN MULTIMODAL MRI VOLUMES

Storey, Ian (2019) TEXTURAL CLASSIFICATION OF MULTIPLE SCLEROSISLESIONS IN MULTIMODAL MRI VOLUMES. MRes thesis, University of Lincoln.

Documents
TEXTURAL CLASSIFICATION OF MULTIPLE SCLEROSISLESIONS IN MULTIMODAL MRI VOLUMES
Thesis
[img]
[Download]
[img] PDF
Storey, Ian - Computer Science - November 2019.pdf - Whole Document

20MB
Item Type:Thesis (MRes)
Item Status:Live Archive

Abstract

Background and objectives:Multiple Sclerosis is a common relapsing demyelinating diseasecausing the significant degradation of cognitive and motor skills and contributes towards areduced life expectancy of 5 to 10 years. The identification of Multiple Sclerosis Lesionsat early stages of a patient’s life can play a significant role in the diagnosis, treatment andprognosis for that individual. In recent years the process of disease detection has been aidedthrough the implementation of radiomic pipelines for texture extraction and classificationutilising Computer Vision and Machine Learning techniques.

Eight Multiple Sclerosis Patient datasets have been supplied, each containing one standardclinical T2 MRI sequence and four diffusion-weighted sequences (T2, FA, ADC, AD, RD).This work proposes a Multimodal Multiple Sclerosis Lesion segmentation methodology util-ising supervised texture analysis, feature selection and classification. Three Machine Learningmodels were applied to Multimodal MRI data and tested using unseen patient datasets to eval-uate the classification performance of various extracted features, feature selection algorithmsand classifiers to MRI volumes uncommonly applied to MS Lesion detection.

Method: First Order Statistics, Haralick Texture Features, Gray-Level Run-Lengths, His-togram of Oriented Gradients and Local Binary Patterns were extracted from MRI volumeswhich were minimally pre-processed using a skull stripping and background removal algorithm.mRMR and LASSO feature selection algorithms were applied to identify a subset of rankingsfor use in Machine Learning using Support Vector Machine, Random Forests and ExtremeLearning Machine classification.

Results: ELM achieved a top slice classification accuracy of 85% while SVM achieved 79%and RF 78%. It was found that combining information from all MRI sequences increased theclassification performance when analysing unseen T2 scans in almost all cases. LASSO andmRMR feature selection methods failed to increase accuracy, and the highest-scoring groupof features were Haralick Texture Features, derived from Grey-Level Co-occurrence matrices.

Keywords:MULTIPLE SCLEROSISLESIONS, MRI
Subjects:G Mathematical and Computer Sciences > G400 Computer Science
Divisions:College of Science > School of Computer Science
ID Code:39936
Deposited On:24 Jan 2020 14:37

Repository Staff Only: item control page