In this episode of the podcast series “Keeping Up With the Radiologist,” Aunt Minnie.com Collaboration with Penn Radiology, Saurabh (Harry) Jha, MD. Mitchell Schnall, MD. Tessa Cook, MD. and Chuck Kahn, MD, address the shortcomings of AI in radiology in a lively discussion.
Dr. Prashant Warier, co-founder and CEO of AI software development company QureAI, has been at the receiving end of radiologist's constructive criticism. The temperature rises a bit in this episode. Like other medical imaging AI companies, QureAI is focused on the needs of time-strapped radiologists who may question the value of their current AI services or remain skeptical and cautious. I am trying to grasp the.
Mr. Cook is a cardiovascular radiologist and is board certified in clinical informatics. She also serves as co-director of Penn Medicine's Center for Radiology Practice Transformation, which develops and pilots innovative initiatives, including AI, to improve care delivery in radiology. Kahn is a radiologist and imaging informatician at the Perelman School of Medicine at the University of Pennsylvania.She also serves as an editor Radiology: Artificial Intelligence.
Cook and Kahn said during the show that each radiology practice has unique needs, and Warrior was not defensive about the current state of his industry.
- Should you build radiology AI in-house or buy off-the-shelf?
- What problem does Schnall encounter 500 times a day that an AI hasn't offered to solve?
- What is private industry doing as part of its radiology AI strategy and scale?
“That's the problem,” Cook said, adding, “The needs of the market are very diverse. What are the challenges that we face and what we've been talking about with our industry partners is, well, that's fine. However, this will not yield any great results.” It puts a dent in our workflow. ”
However, Cook added that he “knows he needs help” in different practice settings.
AI companies are happy to step up to the plate for potential relief. Where does it work? AI is also used to treat strokes and lung nodules.
In the podcast, Cook advises: “Especially if you find a vendor who is open to feedback and willing to improve the performance of their systems, find a vendor that works well and has a good relationship with them, adapt it to your workflow, and go at their pace. Please practice. “
Cook added that data drift will occur. Model performance will be degraded. “AI is just like anything else…it adjusts and checks.”
However, how to “monitor” for AI performance degradation. The panel has the following proposals up for discussion:
- Post-implementation monitoring…who will do the QC? That's still being worked out, Cook said. “We need to engage our industry partners on that part as well. We can't rely solely on our customers to make it successful.” Hear what Warrior has to say.
- “Probably the best model for this is to ask the laboratory pathologist,” Schnall added. [Clinical Laboratory Improvement Amendments] CLIA passed to document and validate the system. This has to become part of our DNA as a radiology department. ”
Radiologist's AI wish list: Cook asked for automated measurements of the aorta. That's what computers are designed for, she says. Jha also added about the line on the chest x-ray and the location of the tube. However, it's not as easy as it seems for everyone. why? As a professional radiologist, you probably already know this, but we'll explain it in detail in this episode.
When it comes to AI in radiology, there is a lot of energy to innovate, and university radiology departments have achieved amazing results. According to Kahn, the challenge then is not only to maintain the product, but also to take what has been developed as a “project” and turn it into a “product.” So he said corporate partners are valuable when expanding and commercializing AI to more sites.
Busy radiology departments considering how to implement AI now or in the near future will want to listen to this special episode of the Keeping Up With the Radiologist podcast about building or buying AI. Listen now.
Thoughts on this episode:
{02:57:00} Make and buy
{04:50:08} FDA clearance
{05:11:23} Corporate partners
{07:47:07} AI backend
{10:32:00} Economies of scale
{13:01:02} Transfer to a commercial organization
{13:55:01} Commercial system adaptation
{15:12:01} Unique needs
{16:36:17} Overfitting
{18:21:18} Post-implementation monitoring
{19:13:18} QC
{26:15:24} AI overhead
{27:16:08} The value of AI
{29:33:11} Comprehensive features
{33:01:05} Marketplace approach
{35:19:14} Previous image processing context
{39:55:14} Improved efficiency
{41:01:05} QureAI in the US
{48:29:19} What the industry needs
{49:01:19} Medical needs of radiology department
Mentioned:
Saurabh (Harry) Jha
Mitchell Schnall
pen medicine
University of Pennsylvania
tessa cook
chuck kern
Prashant Warrier
Cure eye
Special guest:
Prashant Warrier Co-founder and CEO of QureAI. Warier is a data scientist by training and has been building algorithms for about 22 years.
Features:
Saurabh (Harry) Jha, Marylandis an associate professor of radiology at the Hospital of the University of Pennsylvania. Jha earned a master's degree in health policy studies from the Leonard Davis Institute at the University of Pennsylvania. He received his medical degree from the United School of Medicine and Dentistry at Guy's Hospital, King's Hospital and St. Thomas' Hospital. Jha has developed his Value of Imaging, a set of radiology education resources.
Mitchell Schnall, MD, is a physician in Penn Medicine's Abdominal Imaging Services Program. Schnall, chair of the Department of Radiology and the Eugene P. Pendergrass Professor of Radiology at the Perelman School of Medicine, has served as group co-chair of the ECOG-ACRIN Cancer Research Group since its creation in 2012. He is an international leader. You will work throughout your career to bridge the gap between basic imaging science and clinical medicine in translational biomedical and imaging research, ensuring effective integration of radiology research with other medical disciplines.
Tessa Cook, MD, He is a cardiovascular radiologist and a certified clinical informaticist at Penn Medicine. Cook is co-director of Penn Medicine's Center for Radiology Practice Transformation, which develops and pilots innovative initiatives aimed at improving care delivery in radiology. That includes AI.
Dr. Chuck Kahn, a radiologist at the University of Pennsylvania. Khan serves as the department's deputy director, responsible for issues related to informatics and information technology. His personal interests include information systems, decision support, artificial intelligence, and health services research.
This episode tells about the current situation of the radiologist. Aunt Minnie.com Collaborative research with Penn Radiology. The series is also available on Spotify and Apple Podcasts. Check out the new episode!
