Why are Radiologists Embracing AI? Performance Building Trust

Posted by Alyssa Watanabe, MD, FACR on Aug 27, 2020 4:21:45 PM
Alyssa Watanabe, MD, FACR


Computer-Aided Detection (CAD) software has been a fixture in digital imaging centers in the U.S. since the late 1990s. While it’s been widely implemented in the U.S., studies have shown that traditional CAD systems lack specificity and display numerous false flags that are distracting and rarely beneficial. 


As a result, confidence in traditional CAD has waned, and radiologists are in need of solutions that they can trust and that will help with efficiency. With the advent of artificial intelligence (AI) in healthcare, CAD products are now more robust and perform at sensitivity levels that are equal to or above radiologists’ performance levels.  Also, with AI-based triage platforms, overall performance metrics and efficiency are improved—which can help radiology practices rebound after COVID-19 closures.


How, then, can AI based solutions be integrated in imaging centers? The key is trust in AI technology. Now that radiologists have been validating AI software and realizing its benefits in their practice, their perceptions are changing. We asked several radiologists about how they perceive AI is helping their practice, what they view as valuable in selecting an AI partner, and what they think the future of AI can bring to radiology. Here’s what they had to say.




Dr. Michelle Melany, FACR, FSRU is Chief of Women’s Imaging at Cedars-Sinai Medical Center, where she has been practicing for over 20 years. Dr. Melany is Clinical Professor of Radiology at UCLA Medical Center and Chief of Ultrasound at West Los Angeles VA Medical Center. She has expertise and currently practices in the areas of breast imaging, ultrasound imaging and biopsy, and body CT and has authored book chapters, scientific articles and lectured in the field of ultrasound for the past 25 years.


Dr. Mohamed Eghtedari, MD, PhD is a board-certified diagnostic radiologist with UC San Diego Health who specializes in breast imaging. Since completing a fellowship at the MD Anderson Cancer Center at the University of Texas, he has served as associate professor of radiology at the UC San Diego School of Medicine.


Dr. Kiren Jain, MD is a breast imaging specialist and the Medical Director of Women’s Imaging for Inview Imaging. A graduate of Washington University in St. Louis, she completed a radiology fellowship at the Yale University School of Medicine, where she also served as a junior faculty member. She has 23 years of extensive radiology experience, in addition to fellowship and residency training. 


Dr. Ty Vachon, MD is a radiologist and founder of ORA Informatics. A graduate of USC and the Uniformed Services University of the Health Sciences, and former U.S. Navy physician, he serves as a senior medical adviser for multiple organizations and academic institutions.




Dr. Melany: “Having a sorted worklist is a paradigm shift, increasing my efficiency as a radiologist and improving the metrics for the practice. With triage, normal cases can be read more efficiently and with more confidence. Doctors can read through the cases with more focus and improved workflow.”


Dr. Jain: “With AI, radiologists who aren’t necessarily breast imaging specialists but who have to read a lot of mammograms will feel more comfortable signing off on cases that are likely benign, reducing the need to recall patients. Additionally, practices that use AI can market themselves to patients as being able to read their images even more accurately with the help of AI technology.”


“AI can also aid in detection of additional suspicious lesions that warrant sampling during a scheduled biopsy, helping a radiologist understand the full extent of the disease and facilitate treatment planning.”


Dr. Eghtedari: “Quality control is a major area in which AI can be useful. It takes a lot of time to generate the routine documentation required to ensure our practice is in compliance with MQSA. AI can save the time that we spend populating these compliance documents; it cuts down on paperwork and helps practices with workflow management.”


Dr. Vachon: “I have worked on remote naval bases which, like small towns, can have a small radiology staff. If you’re in a small town or don’t have other colleagues to ask in the practice, AI is a nice security blanket to have, a second opinion where you couldn’t get one otherwise.”




Dr. Jain: "I would look for ease of use. The features need to be easy to implement, and the company needs to be able to confirm reproducibility and accuracy of the results they advertise. Cost is an important factor too and they need to have a fee structure that matches your practice volumes. For example, do they charge for use of their system per mammogram, or is it a flat rate? Look at the average volume of mammograms per day at your practice, factoring in busier versus slower times of the year, and use that to calculate the cost/benefit analysis of each AI-based software.”


Dr. Vachon: “First, make sure they can understand your technological needs. Not all AI systems are the same. Everyone thinks that the same plug goes into the same socket, but their tech people need to speak the same language as yours. Every medical AI company who is FDA-cleared is reasonably familiar with this, but that's something you need to check.”


“Second, know the metrics of your practice so that you can leverage the AI to support the goals of your practice.”


“Finally, make sure everyone at your practice is invested. There has to be interest from top to bottom.”


Dr. Eghtedari: “Having a very good customer support team that can help the hospital IT team troubleshoot is a really important factor. There are so many AI products out there, and it’s hard to know which is better, so this is one thing to keep in mind.”




Dr. Jain: “I would emphasize that AI will be a game-changer in your practice; it will make you more efficient and accurate and overall lead to higher patient outcomes and satisfaction. It will markedly improve your cancer detection rate while at the same time reducing the number of false positives.”


“Traditional CAD has been around for a while, but it still picks up things that are insignificant. It doesn’t help with the process of interpretation and basically slows you down as a radiologist. AI-based CAD can deliver something far superior to what most radiologists are currently using, and AI-based triage can streamline my workflow and make my clinical work more efficient, so I can spend more time with patients and focus on scrutinizing images that are potentially suspicious.”


Dr. Vachon: “I think AI will be indispensable. And as more practices start to implement the technology, we will realize the benefits throughout the healthcare industry.”


Dr. Eghtedari: “As AI is integrated into the entire clinical workflow—from analyzing a patient’s intake form responses to image quality through diagnosis, reporting, notification, and follow-up scheduling—the entire practice becomes optimized, creating a true 21st-century practice.”


CureMetrix’s cmTriage™ and cmAssist® are AI-based mammography solutions that are making an impact. We are also proud to announce our recent collaborations with Ambra Health and Nuance.

To learn more about how AI can benefit your radiology practice, follow the link below.

Request A Demo

Topics: radiology, AI, breast cancer, mammogram, cmTriage, artificial intelligence, mammography, screening, CAD, AI-CAD, Deep Learning, Machine Learning