Author: Rafał Jóźwiak

The proof-of-concept version of the AI4AR platform—an innovative radiology structural reporting tool—is now available for users to explore. The version features a structural report form for describing mpMRI prostate scans, which adheres to the PI-RADS v2.1 standard.

In Phase I of the INFOSTRATEG-I project, AI-enhanced Radiology – Detection, Reporting, and Clinical Decision-Making in Prostate Cancer Diagnosis, a reference database of mpMRI prostate scans was created. The database included 500 annotated examinations and the relevant medical data. The dataset was developed in collaboration with the Lower Silesian Oncology, Pulmonology and Hematology Center in Wrocław. The data was annotated by a team of four qualified radiologists, and each examination was described independently by three experts. The examinations were verified for internal correlation of suspicious lesion locations and compared with historical data from fusion prostate biopsies. Anatomical annotations of the prostate gland and its zones, as well as those of suspicious lesions, are essential for conducting research on computer-assisted prostate cancer diagnosis. These annotations facilitate various diagnostic processes, including prostate segmentation, detection of suspicious lesions, and classification of the clinical relevance of lesions. One notable accomplishment of Phase I was the development of a preliminary version of the universal AI4AR structural reporting system. AI4AR facilitates the efficient and seamless creation of descriptions for mpMRI prostate examinations, in compliance with the PI-RADS 2.1 standard, which outlines the terminology, structure, and categories for the final assessment of suspicious lesions. At present, the system offers basic features that enable users to view examinations, create structural reports, and generate text descriptions of examinations automatically. The system can be developed further to integrate with image analysis algorithms that can be used in a variety of scenarios, such as prostate segmentation, detection of suspicious areas, automatic PI-RADS classification, and automatic classification of clinical relevance. During the examination description process, a range of assistance that utilises the innovative concept of AI-enhanced structural reporting will be incorporated.