New research validates the precision of Head non-contrast CT scans, helping improve turn-around time for time-sensitive conditions. The results from a pioneering clinical study, which examined whether a comprehensive diagnostic support technology improves radiologists’ performance, have been published in European Radiology.
The multi-reader, multi-case study compared how accurate 32 radiologists were at detecting clinical findings on non-contrast CT brain scans with and without Annalise’s deep-learning system, Annalise Enterprise CTB. The results showed radiologists assisted by Annalise Enterprise CTB achieved significantly higher detection accuracy than unassisted radiologists. They also interpreted cases substantially faster than unassisted radiologists.
The findings suggest utilizing the Annalise model could have several benefits, including reduced error rates, enhanced efficiency, and improved triage, facilitating timely, effective patient care. The study demonstrated that Annalise’s AI solution improved radiologists’ accuracy by 32% and reduced their reading time by 11%.
Non-contrast computed tomography of the brain (NCCTB) is frequently used to detect intracranial pathology. However, the results of these complex scans are subject to interpretation errors. Annalise Enterprise CTB is designed by and for clinicians, acting like a second pair of eyes to assist with scan interpretation. It can detect up to 130 findings on unenhanced CT brain scans in under two minutes, including numerous conditions such as stroke or intracranial bleeds that require time-sensitive interventions.
“This important clinical study indicates that radiologist performance improves with diagnostic support from a comprehensive, multi-finding AI solution,’’ Annalise Chief Medical Officer Rick Abramson, MD, MHCDS, FACR, said. ‘’The results have important implications for patients and clinicians. We look forward to further measuring and validating them in the clinical trials we continue to conduct worldwide.”
In this study, radiologists first analyzed the cases without using the Annalise Enterprise CTB solution. After a “wash-out” period of at least four months, the same radiologists re-evaluated these cases with assistance from the Annalise CTB solution. Differences in area under the receiver operating characteristic curve (AUC) and Matthews correlation coefficient (MCC) were quantified using a ground-truth gold standard. The Annalise solution significantly improved radiologists’ accuracy for 91 findings and was comparable to the radiologists for the remaining findings.
This paper proves that the Annalise Enterprise CTB solution improves radiologist detection speed and accuracy.
The study affirms Annalise.ai’s commitment to scientific rigor. The Australian-based company continues to research around the globe, including independent, head-to-head comparisons with competitor products and retrospective and prospective studies.
Buchlak Q et al. Effects of a comprehensive brain computed tomography deep learning model on radiologist detection accuracy. Eur Radiol (2023). https://doi.org/10.1007/s00330-023-10074-8