Quantitative evaluation of an artificial intelligence-assisted platform in CT acquisition workflow

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Healthcare
• Medical Workflow Optimization

Background:

• Artificial intelligence (AI) shows promise in simplifying computed tomography (CT) image acquisition.
• AI may reduce examination time and streamline the scanning process.
• This study quantitatively assesses AI-assisted CT acquisition platforms.

Purpose of the Study:

• To quantitatively and objectively assess the benefits of an AI-assisted CT acquisition platform.
• To compare the AI-assisted platform with a non-AI-assisted platform in terms of time and user interactions.
• To evaluate the impact on CT acquisition workflow efficiency.

Main Methods:

• Twelve radiographers scanned a phantom using four protocols on AI-assisted and non-AI-assisted CT systems.
• Video recordings were used to extract total examination time and user interactions.
• Statistical analysis included Mann-Whitney U test, Spearman correlation, and intra-class correlation coefficient (ICC).

Main Results:

• AI-assisted platform significantly reduced acquisition time (40.2-52.8%) and user interactions (35.6-45.1%) (P < 0.001).
• Radiographer experience did not significantly correlate with acquisition time or interactions (P ≥ 0.3).
• Substantial inter-reader agreement was observed for both platforms (ICC for time: 0.82-0.85; interactions: 0.76-0.81).

Conclusions:

• AI-assisted CT acquisition platforms can enhance CT acquisition workflow efficiency.
• Further multicenter studies with large patient datasets are needed for confirmation.
• Reduced time and interactions offer pragmatic implications for radiographer workload and throughput.