Abstract
P1567
Introduction: Structured reporting improves the consistency and reproducibility of the Nuclear Medicine physician while supporting readability and clarity for the referring physician and patient. In addition to these performance and communications benefits, structured reporting at the level of structured content exponentially facilitates data mining and machine learning. In Nuclear Cardiology, an established method of structured data entry currently exists in the form of polar plots. The purpose of this study was to create a web-based graphical user interface (GUI) that allows for structured manual data entry of stress/rest perfusion imaging findings and the automatic generation of a standardized nuclear cardiology report.
Methods: In collaboration with Imaging Informatics and Cardiology, Nuclear Medicine established the fundamental elements of a Nuclear Cardiology Report. This included the findings of stress and rest segmental defects, stress and rest ejection fraction, left ventricular size, thickening and wall motion, imaging artifacts, transient ischemic dilatation, and calcium score. A GUI was created based on the AHA 17-segment left ventricular model using onclick events for all functions except numerical text entry for ejection fraction and calcium score (image 1). Stress and rest defects were graded based on a 5-point system from 0-4, 0 - normal uptake, 1 - mild decrease, 2 - moderate decrease, 3 - severe decrease, 4 - absent uptake. Left ventricular size was graded on a 4-point scale, normal, mildly, moderately, and severely enlarged, and thickening was either normal or abnormal. Wall motion was graded on a 6-point scale, normal, mild, moderate, severe hypokinesis, akinesis, dyskinesia, and paradoxical motion. There are 5 choices of artifacts, breast attenuation, diaphragmatic attenuation, RV insertion, apical thinning, and splanchnic activity. And finally, if TID is present or absent with numerical ratio. The instructions state, 1) select artifacts, LV size, and TID value, 2) select wall motion and wall thickening on Stress/Rest polar plots, 3) click on segment score to change perfusion values, 4) enter LVEF values and calcium scores, 5) click SUBMIT (image 2). Each onclick option from every domain and all iterations of numerical entry fields are specifically mapped to designed reporting vernacular. Clicking submit automatically generates a complete report which can be copied and pasted into Powerscribe ( image 3).
Results: We successfully created a web-based software application GUI for structured data entry of stress/rest nuclear cardiology imaging studies that automatically generates a standardized nuclear cardiology report. Not including the calcium score values, there are 52 object sets and 28 sample sets with a possibility of 4.26E+14 combinations. Images 4 and 5 demonstrate the data input and report generated, respectively, for an abnormal case. Each report generated is anonymous, with a novel ID associated with each unique report for future data mining. The report can be copied with a single onclick function and pasted into a Powerscribe report. While this is currently manual, this could also be automated and integrated in the future. The web-based platform allows flexibility of use throughout the hospital and at remote desktops, which are more frequently used after the COVID-19 pandemic. Integration of this tool into daily clinical workflows allows the passive collection of structured data for future research and can also be used as an active research tool to collect data for a specific project. Once a large database of structured report data is collected, we intend to use these data to train models to automatically evaluate and score the polar plots.
Conclusions: We created a web-based GUI that allows for structured manual data entry of stress/rest perfusion imaging findings and the automatic generation of a standardized nuclear cardiology report.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.