Breaking News! 'Shootout' at the MRI Corral: Bots and AI beat expert radiologists

By Howard Wolinsky

The first large-scale “shootout” or “bake-off” between AI bots, on one side, and radiologists specializing in prostate MRIs, on the other, was reported June 11 in Lancet Oncology.

The bots won. Bigtime.

The study showed AI saved 20% of men from diagnosis with Gleason 6 and from unnecessary Active Surveillance and revealed 7% extra cases of Gleason 3+4 and above that human radiologist eyes missed. Plus AI triggered false positives 50% less often.

Dr. Anindo Saha, of the Diagnostic Image Analysis Group, Department of Medical Imaging, Radboud University Medical Center, Netherlands, and colleagues, report in Lancet Oncology: “An AI system was superior to radiologists … on average, at detecting clinically significant prostate cancer and comparable to the standard of care.”

They described AI as having “the potential to be a supportive tool within a primary diagnostic setting, with several associated benefits for patients and radiologists. Prospective validation is needed to test clinical applicability of this system.”

But researchers urged caution, calling for more studies.They stopped shot of turning over the keys to the MRI to bots and artificial intelligence.

(The famed gunfight at the O.K. Corral made names for Wyatt Earp and Doc Holliday. Will the new study be the road toward AI running the MRI show? https://www.britannica.com/topic/Gunfight-at-the-OK-Corral)

Science Daily touted the “PI-CAI” MRI study as the first large-scale research in which an international team transparently evaluated and compared AI with radiologist assessments and clinical outcomes.

Increasingly, the future of prostate cancer diagnosis and surveillance appears it may be in the “hands” of robots and AI.

MRI are being used moe and more to screen men before they undergo biopsies that can saddle them with unnecessary cancer diagnoses in the case of low-risk prostate cancer, who may in reality not need to know that they have a harmless “cancer.”

Ine expert after another has told me if my PSA was rising today as it was in June 2010 I would would have been asked to do an MRI—or have a biomarker like MyProstateScore 2.0—and likely would have avoided being diagnosed with prostate cancer in today’s world. The numbers of men with low-risk p[rostate cancer going on AS has been declining because of these tech improvements. AI should advance the situation even more.

Men who never needed to know that they had a lame, slow-growing cancer would never have known the difference.

Is this a wake-up call for radiologists? Could the bots be coming after their jobs?

Peter Carroll, MD, of UCSF, one of the developers of Active Surveillance in the late 1990s, mentioned in passing at a recent webinar I moderated for AnCan that he advises young medical students to rethink specializing in pathology or radiology. Why? Because AI keeps getting better and better, and can potentially play a huge role in diagnosing prostate and help find cancers the humans missed and help men with low-risk lesions avoid being diagnosed.

https://ancan.org/webinar-is-pre-treatment-active-surveillance-for-prostate-cancer-safe/

I’ve spoken to the gurus in radiology and pathology, and they insist that AI will not be replacing them—at least anytime soon.

Antonio Westphalen, MD, an ace prostate radiology expert and a columnist for TheActiveSurveillor.com, told me: “Ha! Fear mongering! Will it happen? Who knows! Certainly not abruptly. There will be plenty of time for radiologists and pathologists to adapt.

“A little over 100 years ago there were no radiologists, CT scans have only been available since the early 70’s. MRI’s since the early 80’s. And so on. New diagnostic tools will become available and for AI to be able to work with new tools, data will be required. A lot of data. And that takes time! And we are not even close to having AI replace plain films, ultrasound, CT, MRI, FDG PET, etc. I am pretty sure the next few classes of radiologists and pathologists will do just fine!”

Jonathan Epstein, MD, former ace pathologist at Johns Hopkins, told me in another webinar for ASPI that AI may serve as a back-up or to screen cases, but humans like him will remain firmly in charge.

(Inside scoop: More coming soon on Dr. Epstein.)

https://aspatients.org/meeting/as-101-episode-5-second-opinions-and-biopsies/

Science Daily takes a more middle-of-the-road stance: “Radiologists face an increasing workload as men with a higher risk of prostate cancer now routinely receive a prostate MRI. Diagnosing prostate cancer with MRI requires significant expertise, and there is a shortage of experienced radiologists. AI can assist with these challenges.”

Here’s SD’s description of the new study:

“AI expert Henkjan Huisman and radiologist Maarten de Rooij, project leaders of the PI-CAI study, organized a major competition between AI teams and radiologists with an international team. Along with other centers in the Netherlands and Norway, they provided over 10,000 MRI scans. They transparently determined for each patient whether prostate cancer was present. They allowed various groups worldwide to develop AI for analyzing these images. The top five submissions were combined into a super-algorithm for analyzing MRI scans for prostate cancer. Finally, AI assessments were compared to those of a group of radiologists on four hundred prostate MRI scans.”

The PI-CAI community brought together over 200 AI teams and 62 radiologists from 20 countries. They compared the findings of AI and radiologists not only with each other but also with a gold standard, as they monitored the outcomes of the men from whom the scans originated. On average, the men were followed for five years.

Huisman observes that society has little trust in AI. 'This is because manufacturers sometimes build AI that isn't good enough,” he said.

Doctors and patients alike are distrustful of the idea of relying on AI and reducing the human factor. In society at large, we are facing a huge change in just about everything we do because of these bots. AI already is playing a role in our daily lives.

One of my favorite takes on this idea comes from Flight of the Conchords’s tune, “Robots.”

The parody the aftermath of a robotic uprising stresses: “The Humans are Dead.” Also, the elephants. And there is only robotic dancing. The bots made a big mistake.

Fortunately, it’s in the interest of AI bots for now to keep us alive so they can determine if we have prostate cancer or other diseases—or not. They need us. I think that we and the doctors need them.

Me to AI: Let’s be friends.

ASPI video on AI: https://aspatients.org/meeting/how-ai-will-decrease-overdiagnosis-and-overtreatment-of-prostate-cancer/

Affairs of the Heart and Prostate: ASPI seminar on June 22

By Howard Wolinsky

Kevin T. McVary, MD, FACS, is a professor of urology and director of the Center of Male Health Stritch School of Medicine, Stritch School of Medicine, Loyola University Medical Center in Maywood, Illinois, outside Chicago.

McVary is a rare researcher on the heart and prostate gland and related issues.

He will be speaking at an Active Surveillance Patients webinar on Saturday, June 22, 2024, at noon Eastern. The program is entitled “Matters of the Heart—and the Prostate.”

Register here: https://zoom.us/meeting/register/tJ0rdeGtrDIpEtynJ_U3A1rpWYkiCOHRMagt

Your heart is related to your prostate on many levels