Sunday, December 1 | 10:50 a.m.-11:00 a.m. | S2-SSRO01-3 | E451B
In this scientific presentation, researchers will describe AI's capability to spot an incidental pulmonary embolism on surveillance CT scans of cancer patients.
Venous thromboembolism affects about 3% of cancer patients, with a majority of incidental pulmonary embolism (iPE) overlooked on routine CT surveillance scans, French researchers including Nathalie Lassau, MD, PhD, of Paris-Saclay University in France, Ammari Samy, MD, PhD, from the Institut Gustave Roussy cancer research hospital in Europe, Astrid Orfali Camez, and others noted for the presentation.
In an example of using triage AI for reducing diagnostic delay, the researchers retrospectively analyzed the accuracy in detecting iPE and time to management of a commercial AI algorithm (CINA-IPE, Avicenna.ai) for a cohort of 3,047 oncology patients treated at Gustave Roussy Cancer Campus in France. From the group, the triage AI created 104 alerts for iPE detection (prevalence of 1.3%), and 2,942 negative findings. Thirty-six of the 104 patients had confirmed PE.
The sensitivity and specificity of the AI model were 97.3% and 97.74% while the positive predictive value (PPV) and negative predictive value (NPV) were 34.62% and 99.97%, respectively, the team reported.
The results obtained from the AI algorithm were compared with the attending radiologist's report and analyzed by both a radiology resident and a senior radiologist. What did the researchers conclude? You'll have to drop in on this Trainee Research Prize presentation to find out.
