Interview with Dr. Kai Vilanova, Director MRI Clínica Girona, Coordinator Radiology, Faculty of Medicine, University of Girona. Director School of MRI-ESMRMB (IDIBGI) and Prof. Dow-Mu Koh, MD., M.R.C.P., F.R.C.R. Professor in Functional Cancer Imaging, Consultant Radiologist in Functional Imaging Royal Marsden Hospital, Sutton, UK
How the clinicians see that the models that have been developed and validated, could affect and modify current patient management
How do clinicians anticipate AI models will transform the diagnostic process for prostate cancer?
Dr. Kai Vilanova: AI models are revolutionizing the diagnostic process for prostate cancer, especially through the use of MR. AI can quickly and accurately identify potential lesions and suspicious areas in MRI images, reducing the time radiologists need to spend on each scan. AI can integrate data from different MRI sequences (such as T2-weighted images, diffusion-weighted imaging, and dynamic contrast-enhanced imaging) to provide a more comprehensive assessment of prostate cancer. This integration allows AI to assess both the structural and functional characteristics of the prostate, leading to more nuanced diagnostic insights. Moreover, these models can detect subtle differences and patterns that may not be visible to the human eye, improving early detection of prostate cancer.
Prof. Dow-Mu Koh: Emergent studies have shown that AI models can perform as well as human radiologists for the detection of significant prostate cancer on MRI studies. Thus, AI models could be used as decision support tools towards the early and better detection of prostate cancer on MRI. There is also significant interest and development in using AI tools to define regions of interests on MRI prostate studies to guide targeted prostate biopsy for the confirming significant prostate cancer. In the future, AI models may also be used in combination with prostate MRI for screening men at risk of developing prostate cancer.
How could AI models improve patient care, treatment outcomes, and the patient experience in prostate cancer management?
Dr. Kai Vilanova: The AI will provide detailed diagnostic information which can differentiate between benign and malignant areas with high precision, reducing false positives and negatives. By analyzing MRI features, AI can predict the grade of the tumor, which helps in determining the most appropriate treatment plan. These models can incorporate patient data along with MRI findings to provide a personalized risk profile, aiding in more tailored patient management. AI models allow to perform standardizing readings, minimizing inter- and intra-observer variability among radiologists
Prof. Dow-Mu Koh: AI models developed using imaging, clinical, laboratory and relevant genomics data could in the future help to direct personalised patient care. Likewise, powerful predictive models could emerge from the use of sophisticated AI models to identify patients who might be prone to local disease recurrence following surgery or the development of metastatic disease. There is also huge potential for AI to improve the patient experience in prostate cancer management by accelerating the diagnostic workflow, there by reducing waiting time, improving patient throughput; as well as creating new patient-centric communication of results, education and support.
How could the use of AI models impact the workload and efficiency of radiologists and other healthcare providers?
Dr. Kai Vilanova: AI models will significantly reduce the workload of radiologists by automating time-consuming tasks such as image analysis and reporting, reducing the clinicians’ burden. It will allow radiologists to focus on more complex cases and direct patient care, potentially leading to higher productivity and a better use of clinical expertise. It has been recently published the results of how AI can provide better results than expert radiologists, Figure (Saha A, et al. Artificial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international, paired, non-inferiority, confirmatory study. The Lancet Oncology. 2024 Jun). In the end, this leads to better patient outcomes and more efficient healthcare delivery.
Prof. Dow-Mu Koh: In the current situation where there is rising workload and global radiological workforce shortage, the use of AI models for automated image analysis is seen as a way of helping radiologists identify potential abnormalities on scan, which can reduce image reading times and/ or improved diagnostic accuracy. Using AI models can also help to automate and streamline the radiological workflow, by providing automatic segmentation and measurements (e.g. of prostate and disease volumes), as well as the ability to track treatment-related changes across studies. For other healthcare providers, AI models that integrate relevant clinical data can enhance treatment decision support and provide novel prognostic information that leads to more personalised treatment and optimised use of healthcare resources.
