Back to the list
Congress: ECR25
Poster Number: C-21584
Type: Poster: EPOS Radiologist (scientific)
DOI: 10.26044/ecr2025/C-21584
Authorblock: W. Pettit1, M. Ryan2, A. Raginis-Zborowska2, E. Compton2, A. Kumar1; 1Berkshire and Surrey/UK, 2Sydney/AU
Disclosures:
William Pettit: Nothing to disclose
Melissa Ryan: Employee: Annalise AI
Alicja Raginis-Zborowska: Employee: Annalise AI
Emma Compton: Employee: Annalise AI
Amrita Kumar: Nothing to disclose
Keywords: Artificial Intelligence, Oncology, Respiratory system, Plain radiographic studies, Computer Applications-Detection, diagnosis, Cancer
Purpose The purpose of this retrospective evaluation was to assess the performance and generalisability of an artificial intelligence (AI) tool (Annalise Enterprise v3.8) for identification of “clinically remarkable” (study contained findings indicative of urgent suspected lung cancer and/or clinically acute findings) chest radiographs in a clinical setting prior to clinical use. The project is a part of the UK national AI Diagnostic Fund where it was deployed on a regional level as part of 5 hospitals.
Read more Methods and materials This is the first of a 3-phase AI implementation, to evaluate the performance for the specific population before further implementation is undertaken and understand whether there are any adjustments required. [fig 1] A cohort of adult (over 18 years) patients referred from General Practice and Outpatient for a frontal chest radiograph (CXRs) was collected from a single National Health Service Trust retrospectively from examinations conducted from November 2022 - January 2023. The reference index was established by consensus between at least two...
Read more Results A total of 538 patients were included alongside patient demographic data on age, sex and ethnicity to confirm generalisability. Of the 64 remarkable CXRs, the AI device correctly identified 40/64 (62.5%). Of the 474 unremarkable CXRs, the device correctly identified 409 (85.74%). The sensitivity, specificity, Positive Predictive Value and Negative Predictive Value were found to be 62.5%, 86.3%, 38.10%, and 94.5% respectively. [fig 2] ​
Read more Conclusion The results demonstrate the ability of an AI device to prioritise studies for clinical review based on AI-determined presence of findings. The highly configurable design of the AI device allows the customisation of findings that warrant the classification of urgent and tuning of thresholds for the subject population. In this instance the sensitivity was deliberately high as the aim was to ensure anything categorized as unremarkable is unremarkable, resulting in a high negative predictive value. ​Within the data set there...
Read more References No references 
Read more
GALLERY