On a cloudy day in a nondescript building in Los Angeles, Decrypt was invited to take a closer look at a technology that purports to use artificial intelligence to help doctors better identify cancer cells in the body. The tech is being developed by New York-based medical technology company Ezra, and the company offered a first-hand, hands-on experience with the Ezra MRI scanner.
With pervasive medical conditions like cancer seemingly spreading further and wider each year, Ezra says it’s targeting a narrower—but important—constituency. Early detection of cancer is crucial in treating and surviving the deadly disease, and Ezra says it is making diagnosing and treatment more accessible and faster using full-body scans and artificial intelligence.
“[Early detection] makes a huge difference for every patient, and it also makes a huge difference for the medical system,” Ezra’s medical advisor, Dr. Daniel Sodickson, told Decrypt. “It’s been shown that the five-year survival rates for cancer jumped from 20% to 80%, on average, if it’s caught early, so if you do the numbers, that’s something like a billion lives saved.”
Sodickson says the key to early detection is a full body scan, rather than just focusing on one area of the body or symptom. According to Ezra, the scanner targets several significant areas, including the brain, thyroid, uterus, prostate, lungs, liver, and kidneys.
The nearest Ezra facility to me was on the busy corner of Wilshire Blvd and Robertson in the heart of Beverly Hills. As I entered the office where the scan took place, the first thing that caught my attention was how brightly lit and surprisingly busy the place was for a Monday morning. Waiting for my 9:30 a.m. appointment, I looked around the office to find it filled with people of various age groups, from kids to senior citizens.
“We get a lot of people who are claustrophobic and can’t lay in the standard MRI machine,” a technician told me.
“[Ezra] is a compilation of the best-established exams for each body area,” Sodickson continued. “There’s actually something called active surveillance, which is a new and evolving paradigm in medical care where—if you have a risk of prostate cancer or breast cancer, for example—we’ll bring you back each year to check how you’re doing and to make sure your risk hasn’t changed,” he said.
Ezra explained that it documents each of these exams for each body area, and later put them together one after the other. But while the Erza scan does a comprehensive scan of the body, Sodickson acknowledged that the machine does not do a comprehensive scan of the spiral column.
“In part because spinal tumors are not among the most common,” Sodickson said. “So it’s a time budget; the longer we keep you in the scanner, the more inconvenience for you and the more expensive it is.”
Dressed in my gown, I laid on the bed that fed into the MRI scanner. With my head held in place by a brace, a soft pad over my chest, and a pad to elevate my feet, I was put inside the MRI machine for the faster “Ezra Flash” scan.
What followed was 30 minutes of sounds ranging from buzzing, pounding, whistles, and me trying to remain still. The position wasn’t uncomfortable; the larger MRI machine tube Ezra uses (70 cm bore size), compared to the standard model (60 cm bore size), helped me not feel claustrophobic.
The worst but necessary part of the experience was being unable to move as the collar on my gown rode up my neck, which was very irritating. During the experience, as banging, chirping, and buzzing continued, I meditated and said what mantras I could remember to keep myself calm. I tried not to think about the time and to remember to breathe.
Time stood still as I lay in the tube, the 30 minutes feeling like hours. Finally, though, it was over, and I was brought out of the machine to await my hopefully negative results.
A week after my scan, I received an email from Ezra that my results were processed and uploaded to my Ezra account. The Ezra Flash report indicates places where the scan found items that should be reviewed via a follow-up appointment with my primary care physician.
The report includes a “Risk Score” drop-down that explains the level of severity included in each section, ranging from 1 (no follow-up necessary) to 5 (emergency action needed).
I have to be honest; seeing the words “potential cancer” made me do a double-take, and while I did start to worry, I read further and ran the report through ChatGPT to explain what all of the medical jargon means. While there is still cause for concern, knowing that the spots could be benign gave me hope that the issue was more cautionary than anything.
A follow-up appointment will definitely happen, however.
When a report is processed, Ezra also provides customers the ability to download the report to be viewed as a PDF and as a digital file for medical professionals to review. The most interesting aspect of the report is the digital report, which includes a 360-degree shot of the scanned area and allows for adjusting the depth of the image.
Along with its MRI scans, Ezra utilizes artificial intelligence to produce higher-quality, faster, and cheaper scans.
“The way we make it faster is we can actually take less data than would normally be required,” Sodickson said. “To get a perfect crisp MRI scan or MRI image, AI can be used to clear away the fuzz, almost like wiping a shower door clean, restoring it to full quality in a shorter time.”
The company said the Ezra AI was trained on a large set of images from thousands of patients – both healthy and with various diseases – with the results evaluated by human radiologists.
According to Ezra, its proprietary AI model, Ezra Flash, takes around 30 minutes to complete a full-body MRI scan. Along with Ezra Flash, Ezra Reporter AI creates a digestible translation of radiology reports that assist medical professionals and patients in understanding their results.
“The only people who have access to your scanned information are the medical team and Clinical Operations team at Ezra,” an Ezra spokesperson said in an email. “Ezra is governed by HIPAA and all the associated confidentiality and security regulations. Ezra can delete your information upon request as per the privacy policy.”
Thanks to consumer-facing AI models like Anthropic’s Claude and OpenAI’s ChatGPT, patients and their families have been able to use the technology to decode complex medical terms and jargon that otherwise would have confused and left the families unable to advocate for their sick family members properly.
During the annual Healthy Longevity Global Innovator Summit in October, Peter Lee, Microsoft Corporate Vice President of Research and Incubations, detailed how his family used ChatGPT to decide on care for their elderly father.
“The ability for GPT to give us guidance just brought the temperature down and really kept family harmony,” Lee said. “The thing that I always gravitate towards is the empowerment of people themselves, as we live longer, and the people who care for us as we age, for that, I really have a great deal of hope.”
With the rise of artificial intelligence, medical professionals have leveraged the technology to detect and diagnose difficult-to-fight illnesses like COVID-19, Parkinson’s Disease, and cancer.
In October, a new study published in the international journal Nature introduced an AI-powered concept called Sturgeon, which is aimed at changing how surgeons approach diagnosing and removing central nervous system (CNS) tumors.
“The problem [the Sturgeon researchers] are trying to solve is that surgeons don’t have real-time or rapid knowledge of diagnoses of tumors during cases,” Dr. Gabriel Zada, neurosurgery specialist at Keck Medicine of USC, told Decrypt.
Dr. Zada said Keck and other medical and academic institutions are increasingly turning to AI, adding that AI tools are used across the board before, during, and after surgery.
“The complexity of the surgery doesn’t change; the risks don’t necessarily change,” Zada said. “But one could imagine that if you get some information that augments the surgical approach, it could one day make something safer if you have additional information that would change what you do.”
Edited by Ryan Ozawa.
