Please use this identifier to cite or link to this item:
https://scholarbank.nus.edu.sg/handle/10635/135188
DC Field | Value | |
---|---|---|
dc.title | AUTOMATIC CLASSIFICATION OF SCHIZOPHRENIA, AUTISM AND HUMAN AGE USING FUNCTIONAL MRI | |
dc.contributor.author | TANG BUU VINH | |
dc.date.accessioned | 2017-03-31T18:00:54Z | |
dc.date.available | 2017-03-31T18:00:54Z | |
dc.date.issued | 2016-09-28 | |
dc.identifier.citation | TANG BUU VINH (2016-09-28). AUTOMATIC CLASSIFICATION OF SCHIZOPHRENIA, AUTISM AND HUMAN AGE USING FUNCTIONAL MRI. ScholarBank@NUS Repository. | |
dc.identifier.uri | http://scholarbank.nus.edu.sg/handle/10635/135188 | |
dc.description.abstract | Individual variation in neurodevelopmental disorders as well as brain maturity might be predicted by their brain functional connectivity. In this research, our objective is to identify schizophrenic or autistic patients from healthy controls and perform human age prediction. We attempt to tackle this problem using different machine learning techniques, including support vector machine (SVM), random forest (RF) and k-nearest neighbors (KNN) for the classification task and support vector regression for subject’s age estimation. Healthy controls and patients are first matched based on their age, gender and motion during their resting state functional magnetic resonance (rs-fMRI) scans. Dimensionality reduction is further employed using different brain parcellations ranging from 100 to 400 regions of interest. Each subject is therefore represented by a connectivity matrix corresponding to a brain parcellation which is used as input for various machine learning classifiers. Our study provides a systematic exploration of how different data preprocessing techniques combine with various brain parcellations and machine learning classifiers could support the classification of neurological disorders and human brain maturity assessment. We found that different preprocessing approaches, parcellations and machine learning algorithms give a range of classification accuracy rates from around 60 to 70 percent and mean age prediction error from 10 to 13 years. Although there is no particular brain parcellation that outperforms the others in our classification and regression problem, we found that SVM performs best followed by RF and KNN. This thesis contributes the first steps towards improving the predictability using resting-state fMRI in clinical diagnosis of neuro-disorders patients and human brain activity. | |
dc.language.iso | en | |
dc.subject | resting-state fMRI, schizophrenia, autism, machine learning, support vector machine, random forest | |
dc.type | Thesis | |
dc.contributor.department | ELECTRICAL & COMPUTER ENGINEERING | |
dc.contributor.supervisor | YEO BOON THYE THOMAS | |
dc.description.degree | Master's | |
dc.description.degreeconferred | MASTER OF ENGINEERING | |
dc.identifier.isiut | NOT_IN_WOS | |
Appears in Collections: | Master's Theses (Open) |
Show simple item record
Files in This Item:
File | Description | Size | Format | Access Settings | Version | |
---|---|---|---|---|---|---|
TANGBV.pdf | 1.81 MB | Adobe PDF | OPEN | None | View/Download |
Google ScholarTM
Check
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.