This series highlights the insights and learnings from radiology residents engaged in the use of artificial intelligence for their medical practice
Artificial intelligence (AI) has become less of a science fiction movie and more of a reliable tool that can support radiology residents in diagnosing more patients, achieving more goals and enhancing their skills. It has become an invaluable asset, an invisible teammate that never sleeps or needs to rest but that’s constantly working to ensure the best possible patient outcomes. Already, AI has proven itself in the radiology environment, offering up additional insights and allowing for medical professionals to gain deeper control over workloads and to improve patient prioritization. For Dr. Komal Chughtai, a third-year resident at the University of Rochester (UR) Medical Center, the journey to AI-empowered radiology has been an interesting one with highly beneficial results.
Aidoc: Can you tell us a bit more about yourself?
Dr. Komal Chughtai: I was born in Pakistan, grew up in England, and moved to Ann Arbor in Michigan, USA when I was 16. I became interested in radiology because of my father, he’s a cardiothoracic radiologist and his passion for his work has kept this profession on my radar for some time. I’m familiar with the workloads, how radiology works, and the job. In medical school, I tried quite a few different pathways, including emergency medicine, but I circled back to radiology as I’m fascinated by how broad its scope can be. I realized that clinicians depend on radiology and that imaging has become a central component of patient diagnosis. I like being part of a field that’s constantly growing and expanding and that has so much to offer both patients and medical practitioners. The radiologist is often referred to as the ‘doctor’s doctor’ because they are often consulted when medical practitioners need to figure out what’s going on.
Currently, I am in my third year of residency – I completed my medical degree in Michigan, moved to UR for my residency, and I’m applying for my fellowship in body radiology as it embodies why I became involved in this field in the first place, the multiple modalities and procedural components.
Aidoc: What was your reaction when you first encountered discussions around AI? Were you familiar with using AI before it was introduced at UR as the university has a significant AI program?
Dr. Komal Chughtai: I had always thought of AI in the same way as science fiction, honestly. This idea that one day all images would be read by a computer and that AI would do all the work. So, when I heard about AI it was a very distant concept that I didn’t really understand. When I was in my second year of residency, Aidoc came to UR and was implemented as a tool to flag intracranial hemorrhage on non-contrast CTs. The residents were invited to learn more about the tool from a research standpoint. I was rotating through ER and AI wasn’t on my radar, but my chief resident suggested I get involved, get my arms around something that would allow me to do research in a meaningful way. At that meeting where we were taught about Aidoc and AI, I realized the potential that AI and this system offered.
Aidoc: How has this technology influenced your day-to-day routine as a radiology resident?
Dr. Komal Chughtai: When I was first going through neuroradiology, we started a project where we would turn the Aidoc lab on and off randomly to assess the impact on our work list. I started to see these non-contrast head CTs appear with an AI flag attached to them. They had caught a hemorrhage on a CT and flagged it for me. This draws my attention to scans that needed to be read earlier. If the flag wasn’t there, I would still get the same result, but I would have done other work first. It has changed the dynamic of how we triage and operate the workflows.
Aidoc: Do you feel that the system offers an extra layer of security, helping you catch cases you may have missed?
Dr. Komal Chughtai: That’s a complex question because now we’re getting to the accuracy of Aidoc in flagging intracranial hemorrhage. It goes into one of the projects I am working on with Aidoc right now. What I have found is that the system flags items with high sensitivity and does a good job of picking things up. What Aidoc does is help me – I know there is another system that has also checked the scans. Ultimately, it will always be the radiologist whose name is on the report so it will come down our skills, but AI supports these.
Aidoc: Do you believe that AI offers a tangible benefit to radiology residents?
Dr. Komal Chughtai: Radiology is one profession where AI will be implemented in our workflows, whether head CTs or PE studies or nodule research. To some extent, AI and computer-assisted diagnosis are the future of radiology so it is important for people to come into contact with AI and understand how it works. When we graduate and have careers in radiology, we are going to be working with AI to some extent so exposure to AI training is essential. We need to not be afraid of it and work with it as if it is a team member, not a black box we don’t understand.
We need to teach residents to be critical when they look at studies independently, regardless of the results as this will be how we learn and diagnose. As I mentioned earlier, with or without the AI, the radiologist is the final diagnosis and their name goes on the report.
Aidoc: Do you believe that there will be changes in the curriculum at medical school to teach the basics of AI and machine learning?
Dr. Komal Chughtai: At medical school, I don’t believe that this is going to happen now. We need to educate people about radiology itself, period. We’re not quite there when it comes to teaching medical students about machine learning, maybe one day. We need to do a better job of teaching medical students about radiology as a career. That said, I think that a curriculum of machine learning could be incorporated into residency. We are not currently in a place where machine learning is an essential part of the curriculum, but we may do so in the future if AI becomes a more commonly used tool. The future is AI and machine learning, they are a growing part of how radiology practices in the future, so teaching residents about these technologies and how they work is essential.
Aidoc: How can AI be most effectively used to help transform your workflow?
Dr. Komal Chughtai: Oncology is definitely an area where AI can have a huge impact. We spend a lot of time looking for nodules, so if we had a program that could find these and present them as suspicious it would certainly make workflow much faster. The population is aging and a lot more people are surviving with complex diseases like cancer, for longer. This means they have more imaging. At UR we have a huge cancer population and it plays a significant role in how we practice radiology.
Aidoc: What are your final thoughts on AI and its impact on your profession?
Dr. Komal Chughtai: I think that AI is going to get smarter and already we’ve had some ‘Wow’ moments. On the other hand, there have been moments where we’ve had false positives with AI. I’m sure improvements in the future AI can help us get through the worklists more efficiently.
Aidoc has been of immense help when we’ve been on call and trying to get through studies fast. You may think a CT is negative and then AI will flag it and say otherwise, finding a subtle hemorrhage for us. I think we are looking for studies where AI can have the biggest impact as we have this ever-growing volume of CTS and the need to go faster. If AI can support our results. That’s the holy grail of what it can do in this setting.