Earlier discharge from pulmonary nodule follow-up using artificial intelligence based volume measurements in computed tomography

Abstract

Background: Lung cancer is the leading cause of cancer death worldwide. Effective screening and early detection are critical in reducing mortality. Artificial intelligence (AI) methods have been proved useful in the diagnosis of pulmonary nodules and early diagnosis of lung cancer. However, the implementation of lung cancer screening and frequent detection of incidental pulmonary nodules lead to more computed tomography scans resulting in increased costs. Therefore, determining the cost-effectiveness of AI is important for implementing these methods in routine clinical practice. Based on volume measurements of pulmonary nodules performed by AI, patients could potentially be discharged earlier from incidental lung nodule follow-up.

Objective: To determine whether using AI volume measurements of pulmonary nodules on CT scan results in shorter follow-up time of incidental lung nodule follow-up.

Methods: For this retrospective cohort study patients with follow-up chest computed tomography for incidental pulmonary nodules were included. The primary outcome was the proportion of patients that could have been discharged earlier from follow-up based on the current BTS guidelines using AI volume measurements.

Results: A total of 252 patients were included, of which 49 (19,4 %; 95 % confidence interval [CI], 14.7-24.9) patients could have been earlier discharged from follow-up using AI volume measurements.

Conclusion: Based on current BTS guidelines using AI volume measurements of pulmonary nodules leads to shorter follow-up time period for incidental lung nodule follow-up and therefore a reduction of unnecessary computed tomography imaging, appointments and cost reduction.

Keywords: Artificial intelligence; Computed tomography; Computer aided detection; Cost-effectiveness; Efficiency; Incidental pulmonary nodules; Lung cancer; Return on investment.