Akshay Chaudhari presents his key takeaways from the ISMRM-RSNA Co-Provided Workshop on High-Value MRI, elaborating on the assessment of value in MRI and the challenges of fully utilizing artificial intelligence in diagnostic imaging.
I recently had the opportunity to attend and present some of my research at the ISMRM-RSNA Workshop on High-Value MRI. Hosted at the Capitol Hilton in Washington D.C. in the middle of February, I was welcomed to the city with some beautiful weather that consisted of a 35°F (2°C) evening with pouring rain. But it did not seem as if this weather dampened any spirits given the next day we dove directly into some great discussions.
The first part of the workshop was focused on understanding the current landscape of healthcare costs and reimbursements. As someone who is primarily a MR researcher, these sessions were eye-opening because for the first time, I was able to get a sense of the complexities that exist in the health reimbursement system (at least in the United States). Researchers typically focus on developing new MR methods and implementing them successfully and robustly in the clinics. However, it was interesting to learn that there is a whole world that exists beyond such an implementation. New methods typically require Current Procedural Terminology (CPT) codes to be issued by the American Medical Association. Such CPT codes are used for providing reimbursements to patients, hospitals, and healthcare providers where implementing new CPT codes is a relatively long and arduous process. Interestingly enough, MRI is not necessarily a high-cost modality, but rather, a high-charge modality. I’m still trying to make sense of how that may guide future research, so if you have thoughts, let me know!
With some of the other talks and the keynote session, the discussion then started moving into talking about how to define value in MRI. The concept of ‘value’ in and of itself can be a bit nebulous and is quite certainly different for different stakeholders. Overall, my conclusions were that value can be evaluated in two different ways: 1. Doing the same as the status quo with fewer resources or 2. Doing more than the status quo with the same resources. In the context of MRI, this can be broken down into the notion of providing actionable insights for diagnostic exams with a reduced scan time and/or interpretation time, or the notion of utilizing the same amount of scan time to gather more information that could be used to guide patient therapy. As MR researcher, this can be summarized as performing comparative effectiveness research so that we can evaluate new methods to find “what works best” in evaluating health-related outcomes.
One of the other overarching themes was that of managing disruptive forces such as artificial intelligence (AI) in the field. AI has the potential to considerably enhance the routine workflow that radiologists have and to allow them to be more patient-facing. However, one of the major challenges regarding AI is the issue of standardization. Will a sequence from one vendor have the same appearance on another vendor’s platform? How repeatable do quantitative measurements have to be intra-scanner and inter-scanner? How will neural networks be able to decipher through these variations and will the variations affect the accuracy of such networks? Unless we have a better grasp of such questions, it will be challenging to fully utilize the power of AI in diagnostic imaging.
Overall, this was a didactic workshop to attend as it gave me exposure to concepts that I usually do not deal with on a day-to-day basis. Interacting with the leadership of ISMRM and RSNA and hearing their thoughts on the state and trajectory of the field will certainly shape how I pose research questions in the future and how I determine useful outcomes for them. It was also encouraging to hear positive feedback on my presentation that pertainedto a five-minute quantitative diagnostic knee MRI protocol, which received a 2nd place award for the ‘Best Science’ as well as ‘Best Value’ category. I am curious and excited to see what the discussions from this workshop will lead to and how it may affect the way we perform diagnostic MRI.
Akshay Chaudhari is a Postdoctoral Research Fellow at Stanford University and works for Skope as Technology and Application Specialist. Akshay attended the workshop in his capacity as Postdoctoral Research Fellow.