Skip links

Radiologists Share First Hand Experience on Artificial Intelligence in Practice

Aidoc now holds the most FDA clearances for acute pathologies

Last week we received Food and Drug Administration (FDA) clearance for an additional product in our expanding suite of AI-based workflow orchestration solutions. This means that we’re one step closer to our goal: providing life-saving AI solutions to the broadest base of patients throughout the world.

Why is this significant?

While it’s true the FDA has cleared quite a few Artificial intelligence solutions recently, Aidoc’s recent PE clearance means that we now hold the most FDA clearances for acute pathologies, establishing Aidoc a leader in providing AI value in the emergency setting. Up to 200,000 people a year die due to PE. Undetected or late-detected PE is one of the most common causes of preventable death in hospitalized patients. PE diagnosis can be highly challenging due to its variable and non-specific presentation, making AI-driven workflow triage especially beneficial.

From the early days as a company, we believed in comprehensiveness – being able to provide value across use-cases. We know our customers want and need broad solutions. Solutions that would enable them to have a single, seamless user experience across workflows. Solutions that would enable them to have a single vendor supplying a critical mass of high-quality products.

Naturally, the crucial question to check this assumption is that of the bottom line: to what extent do we help the radiologists in the field, considering the nature of the solution.

Accordingly, we always seek continuous user feedback. I would like to share some insightful comments made by radiologists with hands-on experience using our technology.

Dr.-Barry-Pressman

Q. Can you share a bit about your decision to add the PE solution on top of your existing ICH deployment at Cedars?

A. In addition to the significant value provided to the department by Aidoc’s ICH solution, we recently added the PE module to the workflow. I was impressed by the fact that the coverage continuously grows, allowing us to add this product in the workflow of more radiologists, becoming part of our daily work.

Q. Can you share more your department’s use of Aidoc and the research you’re performing towards assessing the value of Aidoc’s technology?

A. Everyone who’s working with it likes it. The usage is very high – it’s not just our emergency radiologists who are using it but also our neuroradiologists, residents, and fellows.

When I first signed on, I wasn’t sure how many people would interact with the system, and now we see that the prevalence of use is overwhelming.  At the same time, our analysis shows that use of Aidoc software induced improvements in our TAT. The fact that our turnaround times have improved dramatically at the implementation site when compared to other sites in our health system, means that Aidoc not only saves lives but also saves money.

We’ve also seen a decrease in time that patients stayed in the emergency department. This decrease is slight but still significant for a patient in an acute setting.

All of these results demonstrate direct benefits brought by the Aidoc system. However, we are also exploring significant secondary benefits  For instance, we’re seeing that residents do perform better with the AI solution, which, to my judgment,  is very intuitive. I’m interested to see how this will help our residents learn. I think that it’s going to be a multi-step multi-year process.  

Another indirect benefit is in the area of quality control. Peer review is an essential part of our workflow. Aidoc provides us an unbiased platform that facilitates the review process. 

Q. Any other interesting use cases you’ve seen so far while using AI?

A. As soon as we launched the Aidoc solution, maybe a couple of days later, we had an outpatient that a routine head CT, due to the headache.   It was the end of the day. Typically such case would have been left for Monday or the following day. However, Aidoc detected significant bleeds even before the patient had time to leave our outpatient facility.  As a result, the patient was immediately admitted to the hospital.

radiology professor

Q. Can you share more about Aidoc’s impact on quality at UZB?

A. Often we have younger residents covering the night shift.  Aidoc is now notifying them indicating that they must look at something, alerting them that something may be wrong with the patient.

Now they get more signs, which causes them to start thinking, is it a pathology or not? It’s good for training, making residents look at all the things that seem to be abnormal. Our training is becoming much better, and our diagnostic quality is improving because missed lesions are immediately brought to the physician’s attention. As a university hospital, it’s important that we start to introduce AI as soon as possible. If we miss this train now,  we’ll have to catch on in 1,2 or 3 years. At that time the change would be dramatic.

Dr.-Hans-Nieboer

A. How have your residents adjusted to the new AI system?  

Q. The residents are paying more and more attention to this widget. Of course, they have to get used to this extra information coming in but more and more they see the value that Aidoc brings. This system helps them to reach the correct decision. In our hospital, we use Aidoc as a second reader and indeed it ensures that we really didn’t miss something.

Professor-Eli-Konen

Q. Can you share a bit more about your user adoption within the department?

A. This is a game changer. As a department director, I am used to radiologists being opposed to the introduction of any new gadget. We must push the radiologist to adopt any modification of the established practices. Now I have physicians that actually ask for this new software.

We’ve had the PE solution in the last few months and it works quite well and reliably. And I’m talking about small PEs. We had a case and the resident said that the PE scan was not good enough in terms of the timing of the contrast injection and a new test was ordered. However, Aidoc managed to detect a PE even in the low-quality scan.

Join the Discussion