With Professor Thomas Frauenfelder
Deputy Director of Diagnostic and Interventional Radiology
University Hospital of Zurich
It says a lot when a solution works so well for a radiology department that they decide to perform a study to quantify its benefits. That is exactly what happened at the University Hospital of Zurich (USZ): USZ set up a study on the clinical and workflow benefits of Riverain™ Technologies ClearRead™ CT after implementing the solution into its chest CT workflow.
In this Imaging Wire Q&A, we sat down with Professor Thomas Frauenfelder, Deputy Director of Diagnostic and Interventional Radiology at USZ, to discuss how ClearRead CT improved his team’s chest CT reading performance. The study they performed quantified efficiency and accuracy along with key observations to aid other radiology teams looking to bring new CAD solutions into their workflows.
The Imaging Wire: Tell us about your team and how you handle Chest CT reading volume?
Professor Frauenfelder: The Institute of Diagnostic and Interventional Radiology at the University Hospital of Zurich consists of about eighteen staff radiologists and twenty residents. Last year we performed around 35,000 CT scans, 40% of which were chest CTs. For reading, we mainly use a standard PACS system.
Since we do not have a lung cancer screening program, most CT scans are related to either trauma, vascular pathologies, tumor diagnosis and follow-up, or interstitial lung diseases. During daytime shifts, about three staff radiologists read up to 70 CT scans.
The Imaging Wire: Why did you start using ClearRead CT and how do you use it?
Professor Frauenfelder: Several years ago, we evaluated a number of applications for lung nodule detection. Although many applications had a very high detection rate, we seldom used them because our radiologists were forced to open a second application just to see the results. Even then, it was common that when our radiologists opened the second application, the cases had not been read by the system.
The advantage of ClearRead CT is that it sends the “nodule-only” images back into the PACS, where they can be reviewed side by side with the “normal” lung window by forming specific hanging protocols. Our radiologists liked this type of display because they were able to stay in the system and quickly get an overview of possible lung nodules.
The Imaging Wire: Is that what inspired you to perform your study?
Professor Frauenfelder: We found that radiologists were able to review cases much more efficiently and safely with this type of display, especially the young residents. Since there was limited scientific data on the use of the software, we decided to conduct a study to confirm ClearRead CT accuracy and efficiency.
For the study, we created vessel-suppressed reconstructions of 100 patients’ contrast-enhanced chest CTs using ClearRead CT. The two sets of images were read by two groups of three radiologists, finding that vessel-suppressed CTs had 21% greater nodule detection rates, much higher interreader-agreement rates, and about 20% shorter average read times.
The Imaging Wire: What were the most compelling takeaways?
Professor Frauenfelder: Well, we expected that the results would be in favor of ClearRead CT concerning the detection rate and reading time, but it was surprising that the advantages were so significant.
The Imaging Wire: What was your experience with respect to ClearRead CT’s ease of installation and integration into the workflow?
Professor Frauenfelder: ClearRead CT was very easy to install for our ICT. The advantage is that we can adapt many parameters on our own, especially if CT protocols are changing. This gives us a lot of flexibility.
Because all post-processed images are directly stored into the PACS, they are accessible without changing applications. This saves a lot of time. We can also access the results in more detail by using the Web interface, if needed.
Overall, it keeps workflow running very smoothly.
The Imaging Wire: Based on your research and experience with ClearRead CT, what do you see as the most important qualities to look for in a CAD product?
Professor Frauenfelder: Well, many products today are very accurate for the depiction of pulmonary nodules. Some might be too sensitive. Since we do not have a lung cancer screening program, it is important that the system fits into our existing workflow and that it assists the radiologist by providing a nodule-specific recommendation about follow-up. Furthermore, the results should be easily transferable into reports.
The Imaging Wire: Do you have experience with any other ClearRead applications (e.g., ClearRead Xray| Bone Suppress) and if so, can you share about the other ClearRead applications you’ve used?
Professor Frauenfelder: We also use ClearRead Xray with both bone suppression and image enhancement. Our first impression is that ClearRead Xray helps us see pathologies more clearly and more accurately. ClearRead Xray installation and workflow were also very easy, and we’ve benefited from being able to integrate the images in specific hanging protocols on our existing PACS review station.
We actually also performed a study evaluating the use of ClearRead Xray for COVID-19 diagnosis that we’ll publish in the future. In the retrospective study, we evaluated the diagnostic accuracy of conventional radiography (CXR) and enhanced CXR (eCXR/ClearRead Xray) for the detection and quantification of disease-extent in COVID-19 patients compared to chest CT. Our initial findings show that the use of ClearRead Xray increases interreader agreement and has a higher sensitivity for the detection of the consolidation. So it seems that ClearRead Xray improves the detection of COVID-like pneumonia. However, further analysis is needed.
About Professor Frauenfelder. About Professor Frauenfelder. Thomas Frauenfelder is a professor of radiology at the University Hospital of Zurich (USZ), as well as its head of chest imaging, and deputy director of the Institute for Diagnostic and Interventional Radiology. He has a special interest in medical imaging and architecture of PACS in the hospital environment.
