Collaborators from the Children’s Hospital Los Angeles (CHLA) and the radiation oncology software company Voxel Healthcare have been awarded a $2.1 million grant to advance the company’s AI-driven clinical decision support product for pediatric brain radiation therapy planning.
The product, ClickBrain RT, uses deep learning technologies to automatically identify and outline sensitive brain areas on MRI scans. This helps radiation oncologists better plan where to direct brain radiation, which in turn reduces damage to healthy brain tissue leading to a better quality of life.
Radiation therapy is the standard of care along with surgery and chemotherapy for many pediatric brain tumors.
“Today’s advanced techniques allow us to precisely shape the radiation beams during each treatment session,” said Arthur Olch, PhD, chief of the Physics, Radiation Oncology Program at CHLA. “This helps minimize radiation to healthy tissues and has greatly reduced side effects. Nevertheless, there is still more progress to be made.”
He told Inside Precision Medicine that the goal “is to create a tool that helps doctors plan treatments in a way that reduces long-term detrimental effects.” His team of researchers are specifically looking at how different doses of radiation impact the brain’s most sensitive areas in children.
Several brain substructures have well-recognized implications for cognition, especially in growing children. ClickBrain RT is trained to segment the brain based on hundreds of MRI images that have been annotated by neuroanatomy experts to highlight these critical brain substructures.
“When radiation oncologists create treatment plans, our focus is to zero in on the tumor and target it as precisely as possible,” said Olch. “The process is meticulous, but our focus hasn’t been on segmenting specific brain sub-structures like a corpus callosum or hippocampus.”
Even if radiation oncologists were experts in the finer points of brain substructures, Olch points out that not enough research has been conducted to create standards for optimal tolerable doses in these areas. Tolerance dose-driven radiation treatment is a common practice in radiation that exists for the kidney, heart and other key organs, but not yet for specific regions of the brain.
The $2.1 million grant, from the Small Business Innovation Research program at the National Institutes of Health will enable CHLA and Voxel Healthcare to evolve ClickBrain RT to not only auto-segment key brain areas associated with neurocognitive effects, but also to deliver clinical decision support on dosing based on age, chemotherapy dose, tumor type, grade, and location.
To train the technology’s machine learning algorithm, researchers will supply it with a body of demographic and neuropsychological data from pediatric brain tumor patients who were previously treated with radiation to better correlate radiation doses to their effects.
Voxel healthcare was started by two CHLA investigators, Natasha Lepore, PhD, and Niharika Gajawelli, PhD.
“We realized that our expertise in software for medical imaging could be used to help patients directly in the clinic and not just in a lab setting,” said Leporé. “We then participated in the National Science Foundation I-Corps program for customer discovery. This allowed us to interview more than 350 experts in the clinical medical imaging environment to find out what the most pressing needs were. Voxel Healthcare was then founded to solve some of these pressing issues.”
She explained that AI-driven clinical decision support is becoming increasingly important in brain radiation therapy because it enhances the ability to deliver targeted, effective, personalized treatment, which improves outcomes and quality of life for patients.
Although there are AI-driven tools for adult treatment planning, none of them can handle children’s brains. “Hence, this is the reason for developing our product,” said Leporé.
She believes that the collaboration between CHLA and Voxel healthcare is “essential” to provide clinicians tools to reduce cognitive late effects in children treated with radiation therapy for their brain tumors.
“CHLA will provide invaluable data and expertise in the clinical care of children with brain tumors. They will also pilot the software and give feedback for continuing improvements. Simultaneously, Voxel will use its expertise in AI-driven pediatric medical imaging software to build this product.”
Olch said the work will “introduce a new level of personalization to brain tumor treatments.”
He added: “By guiding treatment planners on how to deliver radiation more safely to specific areas of the brain, we are creating a new approach that goes beyond traditional methods that takes into account each child’s individual needs.”
