Towards Fully Automated Tumor and Organ-at-Risk Detection and Segmentation from PSMA PET and SPECT Scans of Prostate Cancer Patients

Prostate cancer is the third deadliest cancer in men and early detection is crucial. PSMA is a protein that is highly present in prostate cells, making it a promising target for imaging and treatment. Total metabolic tumor volume (TMTV) is a measure of tumors’ characteristics, but it is currently not measured in clinical settings due to the labor-intensive and time-consuming process of manually delineating the borders of all tumors in PET images. AI can automate this process, but it struggles with low-quality images and small tumors. Our proposal is to develop AI-based object detection methods to locate lesions before segmenting them to improve accuracy. PSMA can also be used for personalized radioligand therapy, where radioactive drugs attached to PSMA molecules are injected into the patient to kill cancer cells. AI can aid in automating the process of manually delineating the borders of tumors and organs at risk in PET images, simplifying existing protocols, and predicting patient response and outcome. Personalized radioligand therapy could maximize cancer irradiation while minimizing toxicity to healthy organs, leading to better efficacy.

Faculty Supervisor:

Arman Rahmim

Student:

Partner:

Microsoft Canada Development Centre

Discipline:

Life Sciences

Sector:

Technology; Health and Related Sciences & Technology; Artificial Intelligence

University:

The University of British Columbia

Program: