Automatic CT repeat software applied to pulmonary embolism protocols
May 11, 2023
By Anna Romanyukha, PhD and Niki Fitousi, PhD, MPE, Research Department, Qaelum
In order to achieve diagnostic image quality in computed tomography images clinical practice requires scans to be repeated for reasons including incorrect protocol selection by the operator, patient positioning or motion, contrast timing, or the need to adjust scan parameters like tube current modulation and tube voltage (among others). Although repeated scans can be a sign of good clinical practice, high repeat rates indicate sub-optimal practices and contribute to unnecessary patient radiation exposure, contrast usage, time, and financial losses.
Qaelum released a new software (FOQAL-CT Repeat, Qaelum, Belgium) to automatically identify intra-study repeats, employing an algorithm developed and patented by the University of Wisconsin-Madison . The software uses DICOM metadata to identify standard protocol patterns and deviations from them; the flagged repeated scans are classified by repeat type including localizer, bolus tracking acquisition, helical or axial overlap, and extension scans. Imaging scenarios where the number of scans commonly varies may not be frequent enough to create an independent protocol pattern, for example contrast protocols with delayed contrast phases. For these cases a “whitelisting” function is utilized to exclude series descriptions attributed to these scans from the flagged repeats. Whitelisting may also be beneficial for protocols frequently repeated in clinical practice to avoid generating false positive repeats.
CT Repeat identifies normal repeat rates, highlights areas to target for optimization e.g. specific protocols, operators, exam types, and allows to track any improvement trajectories implemented to minimize unnecessary repetitions. Moreover, the impact of repeated scans on department resources like contrast agent volume, device time, tube lifetime, image storage, and financial costs is quantified automatically based on user-selected configurations.
The software was previously validated and applied for retrospective analysis of 61k studies by Dr. Tim Szczykutowicz and his team . Sensitivity and specificity of the algorithm were evaluated and reported as 92 and >99%, respectively, when excluding rare protocols that are not classified into common protocol patterns. Repeat rates were reported as 1.1, 1.8, 3.4, and 4.3% for helical overlap, helical extension, localizer, and bolus tracking scans, respectively, in the academic institution.
In a collaboration between Qaelum and the Maastricht University Medical Center (MUMC), repeat rates in pulmonary embolism (PE) exams were investigated. The hospital team determined repeat rates in helical scans via manual analysis. They quantified the number of helical scans with series descriptions occurring more than once in exams normally containing one scan with a given series description. These repeated series were then evaluated by a radiologist to exclude cases that were clinically indicated, for example scans corresponding to different contrast phases. Qaelum’s novel FOQAL-CT repeat software was employed to perform an automatic analysis on the same dataset of 4481 exams and compare the determined repeat rates to the manual results.
Figure 1. Repeat analysis by protocol pattern, one of the analysis sections of Qaelum’s FOQAL-CT repeat software showing repeat rates separated by type for each identified protocol pattern, prior to whitelisting.
Both manual and automatic analyses reported a repeat rate 3% for PE scans, including helical overlap and extension scans. The automatic algorithm also quantified repeat rates for localizers and bolus tracking scans as 2 and 5%, respectively.
These results were presented at ECR 2023 in Vienna  and are currently being applied to determine the impact of repeated scans on additional resource usage and time.
 Rose S, Viggiano B, Bour R, Bartels C, Szczykutowicz T. “A Multiinstitutional Study on Wasted CT Scans for Over 60,000 Patients.” American Journal of Roentgenology, 2020, 215:5, 1123-1129. DOI: 10.1016/j.jacr.2021.02.014.
 Romanyukha, A., Fitousi, N., Jeukens, C. (2023, February 28-March 3). Identification of repeat rates in pulmonary embolism exams: comparison of manual and automatic analyses. [Conference presentation]. ECR 2023, Vienna, Austria.
Anna Romanyukha received her Ph.D. degree in medical physics from the Centre of Medical Radiation Physics (UOW, Australia) and her M.Sc. degree in health physics from Georgetown University (Washington DC, USA). She worked as a post baccalaureate and pre doctoral fellow at the National Cancer Institute (NIH, Washington DC) on various projects including radiation dose estimation from diagnostic exposures. She now works in Qaelum NV, focusing on advanced software tools in patient radiation dose management and quality.
Niki Fitousi, PhD, is a certified medical physicist with professional experience in all fields of Medical Physics (Radiation Therapy, Diagnostic Radiology, Nuclear Medicine, Radiation Protection). She is currently the Head of Research and Applications in Qaelum, focusing mostly in the fields of radiation dose management, quality and efficiency in medical imaging. She is also a member of the Medical Physics World Board of the International Organization for Medical Physics.