AI Is Superior To Doctors In Detecting Prostate Cancer In MRI

Illustration of AI use in the world of medicine (photo: x @et_insights)

JAKARTA - Innovative research found that artificial intelligence (AI) is more accurate than doctors in hospitals to detect prostate cancer.

The computer system developed by these experts was trained and then tested on more than 10,000 prostate MRI tests in patients. The study revealed that the use of AI resulted in fewer than half of positive errors and reduced clinical detection of prostate cancer by up to a fifth compared to radiological.

Doctors believe that AI can help reduce overdiagnosis and prevent unnecessary surgery in cancer that most commonly attacks men. This will be very beneficial for future screening programs.

Researchers estimate the use of AI to help scan readings is important in handling the increasing demand for medical imaging worldwide.

This computer system is trained and then tested at more than 10,000 prostate MRI tests in patients.

More than 52,000 men are diagnosed with prostate cancer every year in the UK, making it the most common cancer in men. About 12,000 men die each year from this disease - the equivalent of one person every 45 minutes.

However, until now there is still limited scientific evidence as to whether AI is really working effectively, which halts the widespread adoption of the AI system for prostate cancer diagnosis.

Researchers want to test whether advanced AI models trained use thousands of patient examinations are as good as radiologists in detecting clinically significant prostate cancer using MRI.

In the first study of the like, the international team developed technology using 10,207 MRI examinations from 9,129 patients in the Netherlands.

This technology is then tested on 1,000 further patient scans to determine whether they have cancer and if so, how aggressive the cancer is.

This technology is pitted with 62 radiologists from 20 countries, each with an average experience of five to ten years in interpreting prostate MRI scans.

The diagnosis is then matched with the results using histopologies - cell analysis under a microscope - and how the patient underwent an average of four to six years later.

AI is shown to be as effective as radiologists in detecting more severe prostate cancer and providing half the number of positive errors, which can lead to unnecessary biopsy.

AI also detects 20 percent fewer cases where cancer is at a very low level, so it is likely not to interfere with patients throughout their life, according to research published in The Lancet Oncology.

Global researchers, including from Radboud University Medical Center, Netherlands, concluded: "The AI system, when trained and adequately validated for the target population with thousands of patient cases, has the potential to support a diagnosis pathway for prostate cancer management."

"Clinical trials are needed to determine whether such systems can be translated into increased workflow efficiency, equivalence of health care, and the outcome of patients," the study wrote.

This comes as an important prostate cancer screening trial is ongoing in the UK, where the national screening program is likely to follow.

Transform will test the effectiveness of diagnosis techniques including MRI when compared to current standards, prostate specific antigen blood tests (PSA).

Prostate cancer is the most common cancer among men and the second most deadly, leading to about 12,000 deaths per year.

The sooner diagnosed, the better the result, where the majority of men were still alive five years later when diagnosed at the earliest stage, compared to just half in stage four, when cancer had spread elsewhere.