Identification of Disease Biomarkers from Brain fMRI Data using Machine Learning Techniques: Applications in Sensorineural Hearing Loss and Attention Deficit Hyperactivity Disorder
Year: 2015
Type: article
Abstract: Machine learning techniques have received intense attention in the neuroimaging field during the past decade.Nevertheless, machine learning analysis of Magnetic Resonance Imaging (MRI) data is still in its early stage, especially for functional MRI (fMRI) images.The research in this thesis was focused on the automatic diagnosis and prognosis of brain disorders using fMRI data.We had two major research goals: (1) we attempted to design novel machine learning models for the automatic diagnosis of brain disorders as well as for the prognosis of disease treatment; (2) we would like to derive disease biomarkers, which might help us to better understand the disorder and brain function.The first brain disorder that was investigated in this thesis was sensorineural hearing loss (SNHL).Children with SNHL receive little or no benefits from hearing aids and face challenges in developing language abilities.Cochlear implant (CI) is an electronic device that can be inserted into the cochlea for direct stimulation of the auditory nerve and has been demonstrated to be effective in restoring hearing in patients suffering from SNHL.However, variability in speech and language outcomes for CI remains high and individual outcomes may be difficult to predict.To tackle this problem, our collaborators have acquired fMRI images from a group of SNHL patients before the surgery as well as a group of age-matched controls with normal hearing.Using this dataset, we first developed a model for the classification of normal hearing infants vs. hearing impaired infants with SNHL, which provided a proof of principle that it is possible to accurately determine the functional, developmental status of the central auditory system in congenitally hearing impaired children based on MR images alone by utilizing machine learning techniques.Second, we trained a machine learning model to predict CI outcomes, which has validated the hypothesis that pre-implant cortical activation patterns First of all, I would like to express my sincere appreciation to Dr. Jason Lu, who is the supervisor of my PhD study.He supported my dissertation research, both intellectually and financially.Without his support, I would not be able to complete my PhD study.We discussed research progress every week, and I could always get help from him whenever I needed it.Most importantly, he introduced me to the field of Data Science and helped me to find my intense interests in this field.Dr. Lu is open-minded and always willing to listen to opinions that are different from his.I enjoyed working with him very much.I would also like to convey my heartfelt gratitude to Dr. Scott Holland.Without his guidance, I would not be able to find a dissertation topic so quickly.He spent countless hours discussing the research work with me.He provided precious guidance to
Author: Lirong Tan
Institution: University of Cincinnati
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Keywords: sensorineural hearing loss, attention deficit hyperactivity disorder, disease biomarkers, functional MRI, cochlear implant outcomes, congenital hearing loss, central auditory system, cortical activation patterns, machine learning classification, hearing-impaired children, language development, neuroimaging biomarkers, prefrontal cortex, parietal cortex
Open Access status: green
OpenAlex ID: W2320748006
