Has the time come for AI and MRIs?
Also, get the latest on PCa screening move in UK. Brits, don't be "fobbed off" by your GP.
(Editor’s note: We’ve heard a lot about AI and pathology reports.
But MRIs--another key factor in Active Surveillance, helping diagnose and plan treatment for prostate cancer-- have not been left out of the red-hot AI game.
Is AI the future? Radiologists and pathologists I know are skeptical—after all, we all want to feel needed. They see AI as a possible back-up, not a prime player.
However, consider this: Historically, physicians often prepared medications for their patients. Pharmacists did, too. However, with the rise of the pharmaceutical industry and mass production of drugs in the 20th century, the role of pharmacists and physicians in medication preparation changed significantly.
Deja news: Could be happening again with AI?
Dr. Peter Carroll, the AS pioneer from UCSF, last year told a webinar I moderated that he warns medical students searching for residencies to think twice about entering pathology or radiology.
I recently interviewed Randall W. “Randy” Jones, D.E. (PhD,MBA), founder of Bot Image Inc. He has developed ProstatID®, an MRI system using AI.
ProstatID produces a tailored diagnostic report developed from an AI-driven analysis of 64 radiomic features to detect, classify, and segment prostate lesions with a high degree of accuracy. With a proven specificity and sensitivity exceeding 93%, it is approved to support diagnosis, biopsy targeting, and surgical planning. The system is installed in over 20 facilities, including Allina Health System (MN), UTMB (TX), Advanced Urology (GA), Sperling Medical Group (Fl), and Vantage Health (Fl).
The process works with almost all existing MRI systems and enhances clinician confidence with pathology-grounded intelligence. It feels like a game-changer.
Jones has partnered with a commercial service, Precision Prostate Consulting (PPC), to make ProstID more widely available.
Precision Prostate Consulting’s AI-supported “second opinion” consultation provides insights and recommendations that aid in early detection of prostate cancer. The patient or physician can simply upload the MRI “.dicom” file to the individual patient portal and receive a confidential analysis of MRI images.
Jones is himself a prostate cancer patient. Ten years ago, he was diagnosed with Gleason 4+3 at age 59. He opted for focal laser ablation at the University of Texas Medical Branch (Galveston). This was near the beginning of Jones’ education and journey and before Active Surveillance was really a thing or much offered, so like many he chose to eliminate the cancer promptly. However, unlike everyone, he was able to barter his prostate imaging MRI antenna (medical device made by ScanMed LLC, also his company at the time) in exchange for the laser ablation.
Jones has agreed to be a speaker at The Active Surveillor’s upcoming program on Jan. 3, 2026, “AS 2026: Year in Review.” Details to come.—HW)
One MRI, four different readings: AI joins the party with a new opinion
By Howard Wolinsky
When Bill Peck was diagnosed with prostate cancer last August at age 65, he expected difficult decisions ahead. What he didn’t expect was that a single MRI scan could produce four very different readings — ranging from “no lesions” to “high likelihood of clinically significant cancer.”
Randy Jones, PhD, MBA, CEO of Bot Image Inc. in Omaha, Nebraska, which is pioneering cancer detection with AI and MRIs and CTs, told The Active Surveillor: “This is an all-too-common occurrence. Yet the medical community largely ignores it or treats it with an ‘Oh well, we’re doing the best we can’ attitude instead of seeking feedback and improvement.’”
Peck’s diagnosis started in August 2024 with a PSA of 9.7, which led his local urologist to recommend a biopsy. The results: mostly Gleason 3+3, which some experts called “non-cancer,” with one core at 3+4 — but importantly, with large cribriform pattern absent, per renowned pathologist Jonathan Epstein.
Epstein said of the one "3+4 core" that "Large cribriform gland Gleason 4 is absent"
Peck said: “Note that my original biopsy report had two "3+4" cores, but in two subsequent 2nd reviews (one from Epstein), both only had one "3+4" core.”
(Bill Peck)
Then came the MRI reads starting in September 2024, and recently BOT Image AI. Peck summarized the results:
MedStar – Bel Air, Md.: No lesion suspicious for malignancy.
Johns Hopkins: PI-RADS 3 and 4 lesions — high likelihood of clinically significant cancer.
University of Virginia: Clinically significant cancer unlikely.
Bot Image AI / Precision Prostate Consulting: 3 lesions, all PI-RADS 4 — likely clinically significant cancer.
All from the same images — interpreted by different readers.
“How could the second one be completely different?” Peck still wonders.
AI and imaging expert Silvia Chang, MD, an abdominal radiologist at Vancouver General Hospital and Associate Professor of Radiology at the University of British Columbia, noted: “The Bot Image program called 3 lesions, whereas two sites (MedStar and UVa) did not identify a suspicious lesion. This goes along with the known variability in interpretation of prostate MRI.”
Jones, a medical physicist, MRI scientist, medical device inventor, and self-described “healthcare disruptor,” said Peck’s story is less an anomaly than a symptom of a bigger problem.
He led the development of ProstatID, the first AI software with FDA clearance for detection, diagnosis, screening, lesion segmentation, and classification of prostate cancer from MRIs. Built from a prototype tested at the NIH, it is now being used around 500 times a month by dozens of healthcare providers nationwide.
Jones said, “When conflicted, it is the patient’s final decision as to what path and what associated risks there are to take. Putting myself in your place (which is where I have been by the way) is that with the AI calling out more than one PI-RADS 4 lesion, and getting agreement from even one group (especially one as highly esteemed as Hopkins), I would definitely want to confirm with a guided biopsy – and ONLY a guided biopsy using targeting provided by the software’s detected lesions either directly (available only through locations with a Koelis Interventional system or with translation by the radiologist to a Philips UroNav system) or indirectly (cognitive fusion – mental alignment with any ultrasound system in the urologists office). I know that I threw a lot out there in italics but the key is to have the urologist use those AI outputs as targets the best way they can.
“Although false positives are certainly possible (meaning they aren’t clinically significant lesions), at this point, the minor risks (but discomfort) of a biopsy are worth catching the disease now or knowing for certain (at least >90%) that you do not have prostate cancer.”
Michael Liss, MD, PhD, a urologic surgery researcher at the University of California San Diego, has used ProstID on a few patients.
He said ProstatID has an advantage because it uses AI to process prostate MRI scans that have already been acquired. He contrasts that with other systems such as OnQ Prostate, which he is researching, that have to be built into the MRI acquisition.
He said ProstatID could be particularly useful in rural areas where high-level prostate MRI expertise may be unavailable and could serve as a “second opinion” for urologists and radiologists.
He noted: “There is no long-term surveillance outcome data yet on how Bot Image impacts patient management over time.”
He said Bot Image gets impressive results, but stressed that it should not be used as “the sole decision-maker.
Liss maintained: “AI should augment, not replace human experts. Radiologists can integrate AI with broader clinical context and detect additional findings AI may miss.”
Liss warned: “Who do we talk to when it’s wrong?” — saying the process ought to be like a car’s check engine light, which still requires a mechanic’s judgment.
He envisions pairing AI-enhanced imaging with biomarkers to predict cancer progression risk, reduce unnecessary biopsies, and make active surveillance more patient-friendly
Peck, a data analyst at the U.S. Naval Academy, has a different perspective.
He said, “I would definitely recommend AI. My general take is that the computer is better than the human. With my results so different, I’d do AI first, but definitely follow with a radiologist’s second opinion — maybe two. Given different results, I would lean toward AI.”
Jones added: “ ProstatID is much better than most radiologists, but still requires a trained physician's oversight. Nothing is perfect, so combine the report with a radiologist's interpretation to ensure the best results.”
Q&A with Randy Jones, Bot Image CEO
&A with Randy Jones, CEO of Bot Image
Q: What is Bot Image, and how long has it been around?
A: I founded Bot Image, Inc. in 2019. Our flagship product, ProstatID™, uses artificial intelligence to help radiologists and trained urologists interpret prostate MRIs with greater accuracy and confidence. It’s designed to detect, segment, and risk-score prostate lesions in 3D, ultimately improving diagnostic accuracy and reducing unnecessary procedures.
Q: What problem does your AI address?
A: Interpreting prostate MRIs is challenging. Even with high-quality images, cancers can be missed, and false positives can lead to unnecessary biopsies. ProstatID improves detection accuracy, identifies cancers earlier, and reduces false positives — saving lives, avoiding unnecessary interventions, and lowering costs.
Q: How well does it work compared to radiologists?
A: In a double-blind clinical study of 150 cases, the highest-performing radiologist achieved 74% accuracy. ProstatID alone reached 93.6%. When physicians used the AI alongside their own readings, accuracy improved by more than 5%, though some underutilized the software.
Q: How does it work without contrast?
A: ProstatID uses a biparametric MRI (bpMRI) protocol, eliminating the need for gadolinium-based contrast agents used in traditional multiparametric MRI (mpMRI). It relies on:
T2-Weighted Imaging (T2W): High-resolution anatomical detail, delineating prostate zones and suspicious lesions.
Diffusion-Weighted Imaging (DWI): Measures water molecule motion; restricted diffusion often signals malignant tissue.
Apparent Diffusion Coefficient (ADC) Maps: Quantitative diffusion metrics that improve lesion characterization and reduce inter-reader variability.
Our AI extracts 64 quantitative radiomic features — covering intensity, texture, shape, and diffusion characteristics — and processes them using a combination of Random Forest, Deep Learning Neural Networks, and Parallel Boosted Random Forest models. This delivers mpMRI-level accuracy without contrast, making scans safer, faster (15–20 minutes), more comfortable, and more cost-effective.
Q: Can this help men on Active Surveillance?
A: Absolutely. The risk index shows lesion size, location, and risk level, enabling low-risk patients to avoid unnecessary biopsies and track changes over time through simple follow-up scans.
Q: How much can it reduce unnecessary procedures?
A: Published studies show a 30% reduction in false positives, often translating into an equivalent decrease in unnecessary biopsies.
Q: What’s the cost?
A: For facilities, ProstatID is $2,500/year for licensing plus $40 per patient scan. Our direct-to-patient second-opinion service is $259. Compared to the thousands a biopsy or contrast MRI can cost, it’s highly cost-effective.
Q: Who are your customers?
A: Our clients include radiology groups, hospitals, imaging centers, and an increasing number of urologists who value our 3D visual output for targeted biopsies and diagnosis confirmation. Examples include: Advanced Urology, Agiliti Health, Impression Imaging, 3T Radiology, Allegiance Imaging, Bartlett Hospital, Blackford, Concursive, Ferrum Health, Regional Medical Imaging, Sperling Medical, Synthesis Health, Todd Dorfman MD, UC Health, University of Texas Medical Branch (UTMB) Galveston, Vantage Health, and Wasatch.
Q: Who are your competitors, and how do you compare?
A: The AI prostate MRI market is small but growing. Competitors include:
DeepHealth (U.S.) – AI prostate MRI; not FDA-cleared for detection/classification.
RSIP Vision (U.S./Israel) – Segmentation tools, no full automation.
Lucida Medical (UK) – CE-marked, but lacks FDA clearance and full lesion classification.
Quibim (Spain) – Imaging biomarkers, no FDA prostate clearance.
Mediaire (Germany) – CE-marked, no U.S. clearance.
OnQ Prostate (UC San Diego) – Requires specialized MRI hardware.
Why we lead:
First and only FDA-cleared AI for prostate cancer detection, classification, and segmentation, with CE Mark approval expected August 2025
Fully automated — no special acquisition sequences required
Works with both bpMRI and mpMRI, across all MRI scanner types
High published performance metrics — 95% sensitivity, 72.5% specificity, AUROC 0.936
Approved 3D modeling for treatment planning
Integrates directly into PACS, no proprietary viewer needed
Q: What is your long-term vision for this technology?
A: We aim for ProstatID to become as routine for men’s health as colonoscopy is for colorectal cancer — a safe, non-invasive screening tool starting at age 40. Our priorities are:
Early detection at scale — finding cancers before symptoms appear.
Global accessibility — deployable with existing MRI systems, even in resource-limited settings.
Healthcare cost efficiency — reducing scan time, avoiding contrast, and cutting unnecessary biopsies by up to 30%.
Beyond prostate cancer, we plan to adapt our AI to other tumor types — breast, liver, pancreas, kidney, brain — using the same radiomic feature extraction and machine learning framework to deliver multi-cancer, fully automated lesion detection and risk scoring.
How did you find your prostate cancer specialist?
By Howard Wolinsky
I recently shared how I found my AS-friendly doctors here.
But I want to know how you found your doctor. Please answer this poll: https://forms.gle/Hhhdds7rSFz1NcgP6 to help other patients navigate this road.
Chicago Area Active Surveillors will be discussing how to find a new doctor at the meeting at 11:30 am, Aug. 26, at Seasons 52 in Oakbrook Center.
Call in UK for more AS monitoring to avoid overtreatment
By Howard Wolinsky
It’s a bit ironic.
From a U.S. viewpoint, Britannia rules the AS (Active Surveillance) waves with high uptakes,
The United Kingdom and Sweden lead the world in uptake of Active Surveillance in low-risk patients, with 95% of low-risk patients following the monitoring path.
This far outdistances the uptake of 60% in low-risk patients in the U.S., where 40% of patients undergo unnecessary surgery and radiation with potential side effects.
Rates are lower in patients with favorable intermediate risk PCa (Gleason 3+4/Grade Group 2).
Despite the UK’s lead, Prostate Cancer UK, a leading men’s health charity, is still calling out the overtreatment of lower-risk men, especially those with favorable intermediate-intermediate risk (Gleason 3+4/Grade Group 2) cancers.
PC UK reports on the BBC that the National Health Service, UK’s national health service, “is over-treating men for prostate cancer … with around 5,000 a year undergoing treatment for cancers unlikely ever to cause harm. While most prostate cancer cases need treatment, around one in four are so slow growing men can opt for regular monitoring instead, avoiding the side-effects of surgery and radiotherapy such as incontinence and erectile dysfunction.
“Of the 56,000 diagnosed in the UK each year, around 6,500 men opt for this, but an analysis by Prostate Cancer UK said another 5,000 could benefit.”
The problem is that many men with favorable intermediate-risk prostate cancer are being treated when they could be surveilled and treated years later if necessary.
PC UK attributes this to UK guidelines, which prevent GPs from recommending PSA screening, forcing patients to request it. The National Institute for Health and Care Excellence (NICE), which produces guidelines, said it was reviewing its advice.
NICE recommends that monitoring, using blood tests and scans, should be offered to the lowest risk cases (Gleason 6/Grade Group 1) in England and Wales, where nine in 10 will have no signs of cancer spreading within five years.
But research has suggested this could be extended to the next lowest risk group (Gleason 3+4/Grade Group 2) where eight in 10 men will have no signs of cancer spreading.
Prostate Cancer UK has found that many hospitals have started offering monitoring to this broader group of patients, but a quarter have not.
According to the analysis of data in England, this appears to have resulted in over-treatments rates of between 2% and 24% across different hospitals.
But the charity said if the NHS was more active in offering monitoring it could help strengthen the case for prostate cancer screening, which has gained traction since the diagnosis of Olympic cyclist Sir Chris Hoy. (See article below.)
Doctors in the UK and many other countries akso have been slow to adopt PSA screening because it is unreliable and has resulted in the epidemic of overtreatment of lower-risk patients in the U.S. when it was introduced starting in the 1990s. This problem led to the development of Active Surveillance.
Amy Rylance of Prostate Cancer UK, told the BBC: "To reduce the harm caused by prostate cancer and build the foundations for a screening program, we need to both save lives and prevent unnecessary treatment."
Push on for PCa screening in UK like PM Thatcher did for breast cancer
By Howard Wolinsky
The founder of Prostate Cancer UK has urged the British Health Secretary Wes Streeting to “channel” former Prime Minister Margaret Thatcher and simply push forward screening for prostate cancer, the London Telegraph reported.
Dr. Jonathan Waxman, founder and president of the charity Prostate Cancer UK, says the Health Secretary should learn from Thatcher’s decision to introduce routine mammograms for women, which he said had saved 3,000 lives a year since they were introduced in 1988.
The Telegraph recently launched a campaign to adopt routine prostate cancer screening. In the UK, men have to ask their GPs for an order for the test, rather than the GPs initiating the discussion amidst an emerging epidemic of PCa. Men who ask their doctors for PSA screening often are rebuffed because their doctors are concerned about overdiagnosis and overtreatment.
It’s really a bit of a mess.
Writing for The Telegraph, Waxman said: “There are interesting parallels between the prostate cancer and breast cancer screening stories.
“Mrs. T., regardless of the medical consensus, waved her handbag at the doctors, ignored the consensus, and launched a screening program for breast cancer. And the result? Screening mammography, together with treatment advances, has led to a massive improvement in survival from breast cancer.
“Before screening, just 65 per cent of patients with breast cancer survived, and now over 80 per cent are cured as a result of early detection and better treatment.
“And how does this improvement manifest in the real world? Around 3,000 fewer women die each year in the UK from breast cancer than before the screening program was introduced.”
The Telegraph noted that Waxman said the NHS should rewrite its advice so that GPs are told that they should bring up prostate cancer with all patients who might be at extra risk.
As an old newspaperman, I love to see a publication campaigning on an issue impacting public health. It’s not likely to happen, but I’d like to see some regional or national newspapers in the U.S. follow suit and put a spotlight on a men’s health issue, especially prostate cancer.
Telegraph readers with stiff uppers are urging their mates not to be “fobbed off” by clinicians reluctant to order PSAs: https://www.telegraph.co.uk/health-fitness/conditions/cancer/readers-prostate-cancer-stories/
Micro-Ultrasound Today: Where it Stands in Active Surveillance
Could a faster, more accessible scan help monitor prostate cancer just as well as an MRI? A new study shows that micro-ultrasound may be a safe alternative for some men on Active Surveillance. But how does it work—and is it right for you?
Active Surveillance Patients International’s August webinar features the lead author of this important new research to help explain what micro-ultrasound is, how it compares to MRI, and where it might fit into the care of men with low-risk prostate cancer.
The session will run from noon to 1:30 pm Eastern time on Saturday, August 23.
There will be a Question and Answer session following the presentation. Please send questions in advance to: contactus@aspatients.org
Please register for the meeting here.
The featured speaker is Dr. Adam Kinnaird, surgeon-scientist and the Frank and Carla Sojonky Chair in Prostate Cancer Research in the Division of Urology, Department of Surgery at the University of Alberta. He is a leader in prostate cancer imaging and diagnostics, with a research focus on new drugs, technologies, and devices.
Please send questions in advance to: contactus@aspatients.org
In other news:
Please answer this questionnaire on transperineal vs. transrectal biopsies: https://forms.gle/GShpHwegEPtAVgTs9 I have a relevant story coming, and you can take a few minutes and help if you haven’t already.
Please respond to my survey on your experience with MRIs: https://forms.gle/kuQVP4bDhEsm3RLV9
New Research Study at NYU Langone Health for Hispanic Patients with Prostate Cancer: Dr. Stacy Loeb and Colleagues are conducting a new study “Perspectives on Genetic Testing for Prostate Cancer” for Hispanic patients with prostate cancer. Participation involves an anonymous 40-minute virtual interview, in English or Spanish, to share perspectives about genetic testing. A $40 honorarium is provided for participation. For more information, please contact the study coordinator Tatiana Sanchez Nolasco (Tatiana.Sancheznolasco@nyulangone.org or 646-501-2550).
While I am a bit of a technology Luddite, this does sound promising.
However, reduced costs for shorter MRI scans (due to bpMRI) and potential unnecessary biopsies may not appeal to many hospitals who are looking at their bottom lines. Just sayin'!
If Jones’s diagnosis was 4+3 at age 59, AS was probably not a good option for him even if he had found a doctor who favored AS.