Structural magnetic imaging study on autism

• Main Author: Teo, Eunice Shu Juan.
• Other Authors: School of Computer Engineering
• Format: Final Year Project
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer applications::Life and medical sciences
• Online Access: http://hdl.handle.net/10356/55021
• Institution: Nanyang Technological University

Autism is one of the top 10 mental illnesses. It is how some people are born with. The diagnosis takes place as soon as a child starts to display symptoms of autism. The traditional diagnosis can take a lot of time and subjective as it would be a series of questionnaires and observations. Not only time consuming and subjective but also it could be inaccurate if the patient is too young to answer the questionnaires. Autism if detected early and with proper interventions, can be coped and managed and the patient would have no problems living like a normal person. However, there is no cure for autism, the answer to that is to learn how to manage it. In this project, I hoped that using a MRI image analysis approach would not only expedite the diagnosis process but also to give a more objective result. Using the MRI images, significant features were extracted. With the significant features, it was fed into a PBL-McNN classifier to see if the features extracted were accurate enough to be used as an autism detection criterion. With the extracted features, the regions of interest were also identified and mapped against and actual brain. In order to validate the value of the regions detected I did another step of analysis by verifying the region’s brain function. As important as the result of the analysis, the preprocessing of the raw data was important too. The preprocessing required segmentation, smoothing and normalizing. All these were done in SPM8. In this project, 3 covariant were used for the analysis. They were grey matter, gender and BMI. Of the 3, BMI yielded the best accuracy during the classification at 72%. However, it should not be the only criterion in autism detection. On top of that, the mapping of the regions of interest to the brain was relatively fruit as well. Most of the region’s functions were as expected as they reflect the symptoms of autism.