Collaborative research cracks the autism code and makes the neurodivergent brain visible

A multi-university research team co-led by University of Virginia engineering professor Gustavo K. Rohde has developed a system that can spot genetic markers of autism in brain images with 89 to 95 percent accuracy.

Their findings suggest that doctors could one day detect, classify and treat autism and related neurological disorders using this method, without having to rely on or wait for behavioural cues. That means this truly personalised medicine could lead to earlier interventions.

“Autism is traditionally diagnosed based on behavior, but it has a strong genetic basis. A genetics-driven approach could transform the understanding and treatment of autism,” the researchers wrote in a paper published in the journal Scientific progress.

Rohde, a professor of biomedical, electrical and computer engineering, collaborated with researchers at the University of California, San Francisco and the Johns Hopkins University School of Medicine, including Shinjini Kundu, a former doctoral student of Rohde and first author of the paper.

While working in Rohde’s lab, Kundu, now a physician at Johns Hopkins Hospital, helped develop a generative computer modeling technique called transport-based morphometrics, or TBM, that is central to the team’s approach.

Using a new mathematical modeling technique, their system reveals patterns of brain structure that predict variations in certain regions of an individual’s genetic code, a phenomenon called “copy number variations,” in which segments of the code are deleted or duplicated. These variations are linked to autism.

TBM allows researchers to distinguish normal biological variations in brain structure from those associated with deletions or duplications.

“We know that some copy number variations are associated with autism, but their relationship to brain morphology – that is, how different types of brain tissue, such as gray or white matter, are arranged in our brain – is not well understood,” Rohde said. “Understanding how CNV relates to brain tissue morphology is an important first step in understanding the biological basis of autism.”

How TBM cracks the code

Transport-based morphometrics differs from other machine learning-based image analysis models because the mathematical models are based on mass transport, which is the movement of molecules such as proteins, nutrients, and gases in and out of cells and tissues. “Morphometry” refers to the measurement and quantification of the biological shapes created by these processes.

According to Rohde, most machine learning methods have little or no connection to the biophysical processes that generate the data. Instead, they rely on pattern recognition to identify anomalies.

But Rohde’s approach uses mathematical equations to extract mass transport information from medical images, creating new images for visualization and further analysis.

Then, using a different set of mathematical methods, the system analyzes the information associated with autism-related CNV variations relative to other “normal” genetic variations that do not cause disease or neurological disorders, what the researchers call “confounding sources of variability.”

These sources previously prevented researchers from understanding the “gene-brain-behavior” relationship, limiting healthcare providers to behavior-based diagnoses and treatments.

According to Forbes magazine, 90% of medical data is in the form of imaging, which we don’t have the means to exploit. Rohde believes that TBM is the key.

“Such major discoveries from such vast amounts of data could be made if we use more appropriate mathematical models to extract this information.”

The researchers used data from participants in the Simons Variation in Individuals Project, a group of subjects with a genetic variation linked to autism.

Control subjects were recruited from other clinical settings and matched for age, sex, handedness, and nonverbal IQ, while excluding those with associated neurological disorders or family history.

“We hope that the results, the ability to identify localized changes in brain morphology related to copy number variations, could point to brain regions and possibly mechanisms that can be exploited for therapies,” Rohde said.

Additional co-authors are Haris Sair of the Johns Hopkins School of Medicine and Elliott H. Sherr and Pratik Mukherjee of the Department of Radiology at the University of California, San Francisco.