Research collaborated with a team from The Korea Advanced Institute of Science and Technology (KIAST) and the Yonsei University College of Medicine leveraged on deep learning methods to diagnose autism.
Diagnosis of autism spectrum disorders is currently through observation of an individual’s behaviour or the use of validated tools such as the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). However, the variability between individuals in the spectrum and causes of autism still remains largely unexplained. Accurate diagnosis of autism as well as prognosis of patients is challenging.
Researchers from KIAST and Yonsei University College of Medicine recently published their findings from applying deep learning to recognize patterns for diagnosis of autism. Deep learning is a type of artificial intelligence (AI) function that makes use of artificial neural networks mimicking the human brain to identify patterns. This technique has been used in a number of other fields for example, voice recognition, translation, autonomous vehicles, and drug discovery.
"There was something as to what defines autism that human researchers and clinicians must have been overlooking," said Cheon Keun-Ah, one of the two corresponding authors and a professor in Department of Child and Adolescent Psychiatry at Severance Hospital of the Yonsei University College of Medicine.
"And humans poring over thousands of MRI (magnetic resonance imaging) scans won't be able to pick up on what we've been missing, but we thought AI might be able to." She added.
The researchers used MRI scans of autistic individuals to pinpoint structures of the brain that were associated with autism. They were able to identify that abnormal grey and white matter volume together with irregularities in the cerebral cortex activation and connection was associated with autism.
Brain image data was collected from an open-sourced dataset of more than one thousand MRI scans by the Autism Brain Imaging Data Exchange (ABIDE) initiative and five different categories of deep learning models were applied. Another 84 high-resolution MRI images were also taken form the Child Psychiatric Clinic at the Severance Hospital, Yonsei University College of Medicine. The team used both structural MRIs for examining the anatomy of the brain and functional MRIs to examine the activity of the brain at different regions.
The models allowed the team to explore the structural bases of ASD brain region by brain region, focusing in particular on many structures below the cerebral cortex, including the basal ganglia, which are involved in motor function (movement) as well as learning and memory.
"Understanding the way that the AI has classified these brain structures and dynamics is extremely important," said Sang Wan Lee, the other corresponding author and an associate professor at KAIST. "It's no good if a doctor can tell a patient that the computer says they have autism, but not be able to say why the computer knows that."
From this study, the researchers were able to demonstrate the use of deep learning models in describing the reasons behind the development of autism. This AI tool can be applied in the future to assist psychiatric physicians in the diagnosis of autism and understanding its severity.
"Doctors should be able to use this to offer a personalized diagnosis for patients, including a prognosis of how the condition could develop. Artificial intelligence is not going to put psychiatrists out of a job. But using AI as a tool should enable doctors to better understand and diagnose complex disorders than they could do on their own." He explained.