Computer-aided detection of pulmonary nodules: a comparative study using the public LIDC/IDRI database

Area of Science:

• Medical Imaging
• Artificial Intelligence in Radiology
• Pulmonary Nodule Detection

Background:

• The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI) provides a large annotated dataset for pulmonary nodule research.
• Accurate detection of pulmonary nodules is crucial for early lung cancer diagnosis.
• Computer-aided detection (CAD) systems aim to improve nodule detection rates and consistency.

Purpose of the Study:

• To benchmark the performance of state-of-the-art computer-aided detection (CAD) systems for pulmonary nodules.
• To evaluate CAD performance using the largest publicly available annotated CT database (LIDC/IDRI).
• To demonstrate CAD's ability to identify nodules missed by expert human readers.

Main Methods:

• Utilized the LIDC/IDRI database comprising 888 thoracic CT scans.
• Assessed the performance of two commercial and one academic CAD system.
• Investigated the impact of contrast, section thickness, and reconstruction kernel on CAD performance.
• Four radiologists independently reviewed false positive marks from the best-performing CAD system.

Main Results:

• The updated commercial CAD system achieved the highest performance with 82% sensitivity.
• This system produced an average of 3.1 false positive detections per scan.
• Forty-five false positive marks were identified as nodules by all four reviewing radiologists.

Conclusions:

• The LIDC/IDRI database is suitable for benchmarking pulmonary nodule CAD systems.
• An updated commercial CAD system effectively detects most pulmonary nodules with a low false positive rate.
• CAD systems demonstrate potential by identifying nodules overlooked during extensive human annotation processes.