For Satish Viswanath, a researcher and associate professor at Case Western Reserve University, nearly a decade of work has led to this point.
"It's been a long time coming to get to this stage where, we can validate it on a larger scale," Viswanath said.
Last week, Viswanath, along with other researchers from Cleveland Clinic and University Hospitals, received a five-year $2.78 million grant from the National Institute of Health and the National Cancer Institute. This grant would allow them to continue their work on an AI model for treating rectal cancer. 46,000 people are
The model would mine information from MRIs and CT scans and provide patients and doctors with detailed information on their tumors. They would be able to use the model to see what the tumor looks like, the likelihood of it growing or shrinking in the future, and the next steps for treatment.
Viswanath has been working with Case Western since 2012. When he first arrived at the university, he spent his time talking to clinicians and researchers across different cancer specialties to find a common issue. Through these conversations, they found a critical gap in diagnosing and treating rectal cancer tumors.
"The question is which patients have [a] tumor dying out and which patients don't," Viswanath said. "The current clinical approaches... are not as accurate as we need them to be."
This led Viswanath to sit down with a group of clinicians centered around treating the gastrointestinal tract. He found that doctors had a desire to personalize their treatment. The typical treatment process for someone who has rectal cancer consists of radiation, chemotherapy, and surgery. Some patients can receive enough radiation and chemotherapy for the tumor to completely die off, which is called a complete clinical response. However, some patients may require surgery, which can lead to life-altering complications. The issue? Clinicians are having a hard time telling the difference.
This led Viswanath to join forces with different clinicians to help inform surgeons and doctors on how to ensure more effective treatment. Since 2015, he's been working on the model and is hopeful that it will be used commercially in the coming years.
"It's [ the grant] a critical step if we actually want to try and get it into the clinic. We need to be able to confirm that these models work in very large settings," Viswanath said.
The grant will fund a nationwide validation study. Using information from a previous clinical study, this new study will use medical images from over 900 patients in Cleveland and nationwide to better understand how these tumors respond to therapy. Then, a reader's study will provide the technology directly to radiologists to see if it helps them make a diagnosis. The final step will be looking into FDA regulations and potential commercial use.
With the rapid advancements of AI in other parts of society, Viswanath sees this model as a tool, not a replacement.
"We as a society are getting to see AI as a brand new technology being developed in front of our eyes," Viswanath said. "We really have to take our lessons from the past and use them here."
Emily Steinhagen joined the project as the co-principal investigator under Viswanath in 2017. Steinhagen serves as a colon and rectal surgeon for University Hospitals, and knows about the complications of diagnosis first-hand.
"A generation ago, all patients with rectal cancer all ended up with colostomy bags," Steinhagen said. "One of the things that this work will do will help us figure out who might not need surgery at all and who can keep their rectum after surgery."
Something else Steinhagen has seen is a rise in colon and rectal surgery with younger populations. According to a journal from the American Cancer Institute, since the mid-90s, cases among people under 50 have increased by about 50%. It's one of the deadliest cancers in the age group. Steinhagen hopes the AI model will be able to help.
"Hopefully, we can figure out what's driving these tumors, why they're happening, who they're going to happen to, and how to detect them earlier," Steinhagen said.
The study will begin within the next month. For now, the group is focused on ironing out the final details.
"We're at the very beginning of the potential of what it can do," Viswanath said.
