Application of computer-aided detection (CAD) software to automatically detect nodules under SDCT and LDCT scans with...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Computed tomography (CT) is crucial for detecting pulmonary nodules.
• Low-dose CT (LDCT) reduces radiation exposure but can impact image quality.
• Computer-aided detection (CAD) software aims to improve nodule identification accuracy.

Purpose of the Study:

• To evaluate the efficacy of CAD software in detecting pulmonary nodules.
• To assess the impact of different CT parameters on CAD performance, including standard-dose CT (SDCT) and LDCT.
• To investigate the influence of adaptive statistical iterative reconstruction (ASIR) blending levels and definition modes on CAD accuracy for optimizing clinical workflow.

Main Methods:

• 117 patients underwent both SDCT and LDCT scans.
• CAD performance was assessed using varying ASIR blending levels (0%, 60%, 80%) and high-definition (HD) or non-HD modes.
• Key metrics recorded included true positive (TP) rate, false positive (FP) rate, and sensitivity.

Main Results:

• CAD sensitivity was 78.03% for SDCT and 70.15% for LDCT.
• Non-HD mode yielded higher CAD sensitivity for nodule detection compared to HD mode.
• Reconstructions with 60% and 80% ASIR significantly improved nodule detectability over 0% ASIR (p < 0.001).
• On LDCT (HD mode), 60% ASIR showed higher sensitivity than 0% ASIR (p < 0.05) but lower than 80% ASIR.
• Under non-HD mode, CAD performance on LDCT with 60% ASIR was comparable to 80% ASIR.

Conclusions:

• CAD systems can maintain high diagnostic sensitivity for pulmonary nodules on LDCT images when appropriate ASIR levels are used.
• This approach allows for reduced radiation dose, contributing to an optimized radiology workflow.
• The findings support the clinical utility of CAD with optimized LDCT parameters for lung nodule detection.