Utilizing AI in Prostate Cancer Care and Management

For patients with prostate cancer, artificial intelligence (AI) appears to be en route to becoming a key part of the cancer journey, as experts who spoke to CURE explain.

“Right now, AI is definitely being folded into different levels of the entire journey for patients,” says Dr. Soroush Rais-Bahrami of Wake Forest University School of Medicine in Winston-Salem, North Carolina.

AI caught attention at the annual meeting of the American Urological Association, held in April in Las Vegas, where data were presented from a UCLA-led study showing that AI-generated cancer mapping was able to accurately predict seminal vesicle invasion (SVI), or the spread of cancer cells from the prostate gland into a patient’s seminal vesicles.

“AI-generated cancer mapping accurately predicted SVI, despite never being specifically trained for this task. AI dramatically improved upon the SVI predictions of MRI, with the potential to support early diagnosis and treatment of invasive [prostate cancer],” wrote researchers, including Dr. Wayne Brisbane, an assistant professor of urology at UCLA Health, in study findings published in the Journal of Urology.

As Dr. Nitin K. Yerram, director of urologic oncology at Hackensack University Medical Center in New Jersey, explains, the technology’s potential utility in prostate cancer care and management is ever-evolving.

“This role, it’s relatively new over the [past] few years, and it just keeps changing every day,” Yerram says.

“But there [are] two main places that I think AI is used more on a day-to-day basis: one of them is more a diagnosis, and the other one is treatment stratification.”

Regarding diagnosis, Yerram says AI is now sometimes used to read an MRI of a patient’s prostate in order to understand whether there are potentially suspicious areas of the organ.

“That’s going to be prime time in the next few years,” he says, “but what it’s currently being used in right now is, when you do a biopsy — and these slides are sent to a pathologist — [AI] is augmenting the pathologist and helping them determine the areas that need further investigation. ... So it’s really making the pathologist more accurate as well as probably more efficient.”

Concerning treatment stratification, Yerram explains that when a man is diagnosed with prostate cancer, he is placed into a category of low, interme-diate or high risk. “Based on those categories, we’re able to either offer them surveillance for low-risk patients — because that cancer is really slow-growing and probably not going to be a major issue in their lifetime — or, for high risk, we’re saying, ‘Hey, we need maybe surgery or radiation, or maybe both.’

“... Now we’re actually using [AI], and ... they’re able to now look at that slide and understand very minute patterns over a course of many slides, or all the tissue that pathologists probably can’t really perceive but a machine learning algorithm never forgets, [and it] can really understand and see those patterns and can help decipher a little bit more of what is high risk, what is low risk, maybe what’s intermediate risk.”

Such work, Yerram explains, is currently being done by the team at Artera via its ArteraAI Prostate Test, for example.

“We’re able to send the patient’s tissue to this company, and they’re able to scan the whole slide and be able to give us an output saying, ‘Hey, this patient is probably low risk,’ or maybe it’s high risk, or maybe it’s intermediate risk. So, we’re able to further identify patients and stratify them better to help make treatment decisions.”

In the realm of treatment, Yerram says machine learning algorithms are playing a role, for example, by making robotic surgery more efficient and less harmful for patients in order to effectively treat cancer and improve functional outcomes after surgery.

“You know, not everyone needs surgery,” Yerram adds. “We have different options, including ultrasound ablation ... and AI is able to help us determine, when you have a whole prostate and you want to just treat one certain area, how we can use [AI] to find that area, give a good treatment margin [and] help us determine what the most effective treatment course [is]. It’s helping the physicians augment that treatment pattern, and I think we’re going to be able to see leaps and bounds of improvement in this area for the years to come.”

However, questions and issues remain, as researchers note.

“The long-term success of AI in prostate cancer depends on our ability to solve multiple challenges spanning ethical, legal, regulatory and technical realms,” wrote Dr. Irbaz Bin Riaz, of the Mayo Clinic in Phoenix, and colleagues in the 2024 American Society of Clinical Oncology Educational Book. “Critical issues include protection of the privacy of the patients’ data, liability management between the [mistakes and biases] that can come with the AI systems, and the interpretability and adaptability deficit of the AI systems with respect to current clinical workflows.

“The effectiveness of the AI models depends highly on the quality and diversity of the training data; therefore, models often do not have their intended effect across different hospitals and healthcare systems settings because of data limitations. Additionally, there is concern about healthcare providers potentially over-relying on AI, which might diminish the value of human clinical judgment and intuition. The absence of standardized AI guidelines, coupled with high deployment costs and fears about job security and loss of professional autonomy, further complicates AI adoption in clinical practice.”

“Right now, [AI] has a complementary role, as far as it can make our jobs easier and [make] sure that things are not falling through the cracks,” says Richard Boyajian.

An advanced practice registered nurse and nurse practitioner in the Department of Radiation Oncology with the Dana-Farber Brigham Cancer Center in Boston, Boyajian founded the Virtual Prostate Cancer Clinic at Brigham and Women’s Hospital in 2016.

He is also a member of the CURE advisory board.

“The quicker you can present all the data succinctly to the provider, they’re able to make a decision based on these multiple points that should be considered,” Boyajian says. “If you don’t present all of it, it’s human nature to ... [say], ‘OK, I know we need to focus on [prostate-specific antigen]. We need to focus on what’s going on ...’ But there may be other things that are in play that AI can calculate to build the picture so we can put whatever’s going on into a context, if you will.”

Aside from skin cancer, prostate cancer is the most common cancer among men in the United States, with the American Cancer Society estimating that there will be approximately 313,780 new cases of prostate cancer in the United States in 2025. Incidence rates of the disease have increased by 3% per year since 2014, and approximately 1 in 8 men will be diagnosed with prostate cancer in their lifetime.

“With prostate cancer, we’re going to cure, what, 90% of these guys? They’re the biggest male population of cancer survivors in the world, in the country,” Boyajian says. “So, if you think of what’s going on now ... we’re being asked to
see more people than humanly possible in a certain time frame. It gets overwhelming because the human mind can’t process everything quick enough.”

Boyajian says he is now working with Cancer Insights on a National Cancer Institute grant to use their AI platform to understand what phase of care
a patient is in based on medical records data, with eventual plans to develop monitoring modules for each phase of care.

“It will make most of my colleagues’ lives infinitely easier, especially in the survivorship and follow-up community,” he says, “because we’re just drinking at a fire hose right now, where there [are] just so many patients.

“The other thing I worry about is [that] our human structures are breaking down. I’m not in love with AI, don’t get me wrong, but I’m seeing more and more that humans are fallible, and when you overwhelm the healthcare system — like it seems like it is now — there’s a lot of stuff falling through the cracks. And in other businesses, that might be acceptable, but falling through the cracks here means there’s a patient [who] hasn’t got an answer on something, stressing out of their mind, wondering, ‘What’s happening? Are they just letting me die?’ or things like that. That’s the communication issue and ... we need to find ways around it.”

When it comes to treating the prostate, Brisbane explained in a 2024 interview with CURE that precision is key.

“There are some very important factors that run in and around
the prostate,” Brisbane said. “One is the urethra, and the others are the neurovascular bundles, which are like telephone wires from the brain to [the] phallus, saying it’s time for an erection, and those are running right alongside the prostate. The bladder is right behind the prostate, and then there are the pelvic floor muscles, which are responsible for urinary continence and fecal continence, right at the apex of the prostate or the nose of the prostate.

“The prostate is kind of... high-stakes real estate, so where the tumor is relative to all those structures really indicates what kind of side effect profiles patients are going to experience when they’re getting treatment. So, knowing that with accuracy is very important to know [when asking], ‘Can we treat this cancer?’ or ‘How do we treat this cancer best while minimizing side effects?’”

Tumors, Brisbane explained, are shaped like crabs, in that they have a central body as well as “a lot of legs that go out beyond the portion [of the tumor] that is visible on imaging, and that’s true for all imaging modalities [such as] MRI, [prostate-specific membrane antigen] and these newer imaging modalities like microultrasound.”

As of late, Brisbane has been utilizing the cancer-mapping tool Unfold AI.

As explained by AI health care company Avenda Health, Unfold AI “combines patient-specific data from prostate imaging, biopsies and pathology into deep learning algorithms to create a unique and tailored cancer estimation map. Its 3D, AI-generated map visualizes the location of the cancer for physicians to use in treatment decision-making and interventional planning.” It was cleared for use by the Food and Drug Administration in late 2022, according to Avenda Health.

Brisbane and his colleagues have utilized Unfold AI to support the use of precision medicine (also known as focal therapy) to treat cancer with the TULSA procedure, or TULSA-PRO, which uses ultrasound to create high temperatures and destroy prostate tissue.

“The use of AI is certainly building and only becoming more augmented,” Rais-Bahrami says. “As we move forward, there’s definitely going to
be an important role for AI in a lot of the processes we use, and I would say ... the diagnostic and treatment decision-making journey for patients with prostate cancer, we have yet to see where it’s going to land, but I certainly see it developing [and] becoming more commonplace.”

Rais-Bahrami is the William H. Boyce Endowed Chair in Urology and Professor and chair in the Department of Urology at Wake Forest University School of Medicine. He is currently using the Unfold AI tumor mapping tool retrospectively to compare its accuracy to diagnosis without AI assistance.

“Specifically, with Unfold AI, it’s integrating the inputs from patient demographics, clinical parameters and patient imaging — so the primary imaging from prostate indication, pelvic MRI, multiparametric MRI input [and] the pathology output from biopsies that we get,” Rais-Bahrami says. “It’s giving us cancer estimation maps that have been very valuable in my practice to scale the degree of radical prostatectomy surgical planning, [as well as] planning for focal therapy patient candidates.”

Rais-Bahrami also explained that the use of AI can help level the playing field, so to speak, for patients and providers. “Understanding each clinical case and personalizing treatment algorithms, decision-making [and] patient counseling is going to be key, and that’s where the nuances of AI will most likely normalize some of the personalization that was largely based on clinician and provider expertise,” he says. “AI is taking huge data sets of input with deep learning and machine learning, whereas that would take an entire generation of learning, a senior practitioner and a senior urologist with hundreds or thousands of cases of experience and background.

“So certainly, it levels and normalizes the playing field. And early on,
it normalizes the experience that practitioners can provide to their patients in personalizing the care recommendations. And then even
for very seasoned and experienced practitioners [who] have had their own subset of patients over career-long practice, AI will also bring in infinitely more data points of experience than even one practitioner — even seasoned ones — will have seen.”

For more news on cancer updates, research and education, don’t forget to subscribe to CURE®’s newsletters here.