Eirini Polymeri: AI as a tool for individualized prostate cancer treatment
AI-based image analysis can contribute to improved assessment of prostate cancer patients and optimized radiotherapy planning. Eirini Polymeri’s research shows that this method can be used in clinical practice.
EIRINI POLYMERI
Dissertation defense: 4 April 2025 (click for details)
Doctoral thesis: Artificial intelligence-based organ and tumour segmentation in prostate cancer patients. Studies on PET/CT and pre-treatment CT scans
Research area: Radiology and Imaging
Sahlgrenska Academy, The Institute of Clinical Sciences
Prostate cancer is the most common form of cancer among men in Sweden. To assess the extent of a tumor, positron emission tomography (PET) and computed tomography (CT) are used (see fact box below). However, the association between tumor spread on imaging and patient prognosis has not been adequately explored, potentially complicating individualized treatment decision-making.
“Our studies show that AI-based measurements of the prostate tumor in relation to the volume of the prostate gland are associated with patient survival,” says Eirini Polymeri, senior consultant in radiology at Sahlgrenska University Hospital and doctoral candidate at the Institute of Clinical Sciences.
AI can identify and measure tumors
Eirini Polymeri’s research shows that AI can automatically identify and measure prostate tumors in PET/CT images. By partially replacing manual measurements with AI-based analyses, variability in physicians’ image interpretation can be minimized. This, in turn, can result in more consistent and reliable treatment decisions.
Improved radiation treatment planning
Radiation treatment planning requires great precision to minimize damage to healthy tissue. Today, this process is performed manually by radiation oncologists, which is both a time-consuming and complicated process.
“Our studies also show that AI can be used to identify and delineate the prostate and surrounding risk organs in medical images. Most of the AI-generated delineations could be applied clinically in radiation treatment planning, with only minor manual adjustments needed from specialists,” says Eirini Polymeri.
By automating parts of the radiation planning process, using AI, the workload for specialists can be reduced, while treatment can become more individualized.
Challenges and lessons learned
Conducting research in an interdisciplinary field has been both challenging and rewarding for Eirini Polymeri.
“It was fascinating to collaborate with experts in radiation oncology, radiation physics, and computer science. Learning to think differently from what we usually do in clinical practice and combining these perspectives has made me a better doctor and researcher,” she says.
The pandemic, however, posed several challenges. Remote work, travel restrictions, and canceled conferences delayed the research process.
“Dealing with manuscript rejections was another challenge. But that is part of the research process and something I have learned a lot from.”
Text: Jakob Lundberg
Positron emission tomography (PET) is an imaging technique that uses small amounts of radioactive substances to detect pathological processes in the body, such as cancer or inflammation. PET is often combined with computed tomography (CT) to provide a more detailed picture of where the disease is located. For cancer patients, PET/CT is used to detect tumor spread, monitor treatment, and detect possible recurrence.