Introduction/Aim. Pulmonary nodules are frequent incidental findings and appropriate follow up should facilitate early diagnosis of malignancy while minimising negative sequalae of investigation. Radiologist assessment of computed tomography (CT) images of pulmonary nodules in conjunction with guidelines (e.g. Fleischner Society Guidelines; FSG) are used to determine management. This study assess the use of autonomous software to aid radiologist assessment of pulmonary nodules. Preliminary results are reported.

Methods. CT chest scans performed for pulmonary nodule surveillance between October and November 2018 were retrospectively identified from the Gold Coast University Hospital imaging database. Data from the initial radiologist report (RAD) was collected. All CT scans were analysed by the Philips Pulmonary Nodule Analysis® autonomous software. Radiologists reviewed both the CT and the software‐generated results and created a second report (AIRAD). FSG‐recommended follow up was calculated based on both reports.

Results. Scans were obtained from 20 patients who were 45% female (n = 9) and had a median age of 69 years (IQR 59‐75 years). The largest nodule's mean maximum diameter was 6.8 mm (SD 3.3 mm) as assessed by RAD and 7.8 mm (SD 3.3 mm) by AIRAD (P = 0.011). There was 5 (25%) disagreements regarding the lobar location of the largest nodule between RAD and AIRAD (P = 0.214) and 4 (20%) disagreements between RAD and AIRAD about the presence of spiculation (P < 0.001). FSG‐recommended follow up was different between RAD and AIRAD in 8 cases (40%; P = 0.047) and FSG‐recommend follow up based on AIRAD was earlier in 7 of these cases.

Conclusion. This study provides evidence that use of autonomous software may alter the opinion of radiologists performing pulmonary nodule analysis and lead to different follow up suggestions than conventional radiology assessment.