Setting the stage for AI in medical imaging - a Q&A with Prof. Dr. Erik Ranschaert
In recent years, artificial intelligence (AI) for medical images has made significant progress, mainly thanks to advances in Machine Learning (ML) and Deep Learning (DL) coupled with a massive increase in available computing power. As a result, AI-driven products have entered the market and enabled computer-aided image analysis of radiological images, primarily for classification and anomaly detection.
However, algorithms can also be used for a variety of non-interpretive applications that support the radiological workflow for purposes such as planning, prioritization, quality, patient safety and operational efficiency.
We’re still very much in the early stages of AI implementation in radiology. This means that many are still looking for ways to easily integrate AI into existing workflows, and to find the necessary funding needed to help further progress in this field.
Osimis spoke about the best path forward for AI in radiology with Prof. Dr. Erik Ranschaert, Radiologist at ETZ Tilburg, Visiting professor at Ghent University, EuSoMII Past-President
Q - Why use AI in the field of medical imaging?
Erik Ranschaert - For a start, the radiological workload is continually growing. The increasing age of our population is one reason for that. We’re also confronted with a shift towards preventive medicine, which also means that a lot of imaging is being performed in a pre-emptive capacity. There is a growing tendency towards screening with imaging: we already screen for breast cancer, and more and more instances of imaging are being performed for detecting prostate or lung cancer for example.
Imaging is also increasingly used for staging of oncological diseases, such as rectal cancer. The treatment for cases such as these is often dependant on accurate imaging for staging and evaluation of the disease.
It’s also important to consider the fact that radiologists are confronted with more and more complex imaging tasks, not only to detect diseases or to determine the treatments, but to evaluate the treatment of patients. This also means that radiologists are becoming indispensable in making decisions about the treatment of patients in a growing number of multidisciplinary meetings.
As a result, radiologists are always searching for tools to assist them in dealing with this increasing workload and this is where AI comes in: empowering radiologists so they can perform these tasks with a greater capacity and capability.
Radiologists are looking for a co-pilot, able to assist them in providing the best information in a timely way, with the final goal being the improvement of patient services. AI can and will play a role, not only for image-related tasks such as detection and evaluation, but also for operational tasks such as prioritization of examinations in the ever-lengthening worklists they’re confronted with.
In a time where healthcare is under even more pressure to perform, AI in radiology will help save time, improve the diagnostic process and increase confidence levels, all the while working towards that vital goal: enhancing the quality of patient care.
“Whether you want it, need it or dread it, AI will overhaul radiology. It is the co-pilot we all need to direct us towards better patient care. That calls for a well thought-out roadmap.”
– Prof. Dr. Erik Ranschaert, Radiologist at ETZ Tilburg, Visiting professor at Ghent University, EuSoMII Past-President
More on 3 key challenges and 5 essential benefits of AI in medical imaging?
