AI Tool Streamlines Prostate Cancer Diagnosis

The rising reliance on blood tests for detecting prostate cancer has resulted in increased workloads for healthcare providers. To alleviate this pressure, researchers at NTNU have developed an AI diagnostic tool designed to assist in this area.

Artificial intelligence-based diagnostic tools are beginning to be integrated into hospitals across Norway. These AI systems are capable of independently interpreting X-ray images to identify bone fractures and evaluating tumors in breast and prostate tissues.

“AI technologies can efficiently manage the identification of straightforward cases, freeing up doctors to concentrate on more complex situations,” stated Tone Frost Bathen, a professor at the Norwegian University of Science and Technology (NTNU) and project manager for the AI-driven prostate cancer analysis tool named PROVIZ.

Initial tests conducted at St. Olavs Hospital have shown promising results for this tool.

“AI can help radiologists evaluate more quickly and accurately whether a patient requires a biopsy and, if so, the specific location in the prostate from which it should be obtained,” Bathen explained.

Recent studies indicate that patients are more likely to trust medical test results when they receive confirmation from an experienced physician.

“For AI to be effectively utilized in the diagnosis of prostate cancer, trust in medical professionals is essential. Technology alone does not suffice; human interaction and expert judgment remain vital,” remarked Simon A. Berger, a PhD research fellow at NTNU.

Prostate Cancer and Ageing

Prostate cancer ranks as the most prevalent cancer among men in Western nations.

Screening has shown that 10 percent of men at age 50, 50 percent at 60, and around 70 percent of those over 80 have prostate cancer.

This indicates a strong association between the disease and aging.

“Most men live with prostate cancer rather than from it,” added Berger.

The Prostate-Specific Antigen (PSA) blood test plays a critical role in identifying prostate cancer. As more men are opting for this test, the annual incidence of new prostate cancer diagnoses has surged, now reaching approximately 5,000 cases.

As more individuals are screened for a condition that many may naturally develop as they age, the follow-up medical procedures, post-blood test, must also increase to provide a comprehensive clinical understanding of prostate cancer severity.

Trusting Medical Professionals

Currently, the follow-up procedure involves MRI scans, which furnish detailed images of the prostate gland and adjacent tissues. These images require manual interpretation by skilled radiologists. The substantial rise in imaging demands innovative methods for efficient diagnoses.

Through the PROVIZ initiative, NTNU researchers have crafted an AI-supported tool aimed at assisting doctors in interpreting prostate MRI scans. While currently available only within the research framework, efforts are ongoing to patent the tool and eventually make it available in the commercial sector.

For this technology to enhance efficiency in routine medical practice, patients must also have faith in the results produced by AI.

Berger conducted interviews with 18 men diagnosed with prostate cancer utilizing PROVIZ. The findings reveal that trust in healthcare professionals is crucial for patient acceptance of AI technologies in medical diagnostics.

“Patients tend to trust AI in lower-risk situations, like identifying bone fractures, but are hesitant in higher-stakes scenarios like cancer diagnoses. In these cases, they prefer confirmation from specialized doctors,” Berger explained.

Doctors as Trusted Guides

In his research, Berger identified three critical dimensions of patient trust.

1. Foundational trust in the healthcare system: many patients had previously positive experiences, establishing a favorable basis for trust.
2. Interpersonal trust in health professionals: patients relied on their doctors and their evaluations, which were essential for embracing AI, as doctors articulated and endorsed the technology.
3. Conditional trust in AI: while patients acknowledged AI’s potential, they insisted on human assessments, particularly in prostate cancer diagnoses, due to concerns regarding accountability, professional judgment, and AI’s ability (or lack thereof) to consider the entire clinical picture.

“The relationship between patients and physicians remains paramount. For AI to be effectively integrated into clinical settings, healthcare professionals must communicate actively and assure patient safety. For doctors to serve in this capacity, they need to understand how AI reaches its conclusions to validate its assessments. Patients are likely to embrace AI within a framework of trust they already have,” Berger concluded.