Computer-aided detection of lung nodules using outer surface features

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Lung nodules are critical indicators of lung cancer.
• Accurate detection of lung nodules in computed tomography (CT) images is essential for early diagnosis and treatment.
• Existing computer-aided detection (CAD) systems face challenges in achieving high sensitivity and specificity.

Purpose of the Study:

• To develop and evaluate a novel CAD system for detecting lung nodules in CT images.
• To investigate the impact of incorporating outer surface texture features on CAD system performance.
• To optimize the classification algorithm for improved nodule detection accuracy.

Main Methods:

• A four-phase CAD system was developed, including 2D and 3D preprocessing.
• Feature extraction involved morphological, statistical, histogram, and outer surface texture features.
• A support vector machine (SVM) classifier was optimized using particle swarm optimization (PSO).

Main Results:

• The CAD system achieved 97.37% sensitivity, 86.38% selectivity, and 88.97% accuracy with three feature groups.
• Inclusion of outer surface texture features improved performance to 98.03% sensitivity, 87.71% selectivity, and 90.12% accuracy.
• The false positive rate decreased from 2.7 to 2.45 per scan with the enhanced feature set.

Conclusions:

• Outer surface texture features significantly enhance the performance of CAD systems for lung nodule detection.
• The developed CAD system demonstrates high potential for accurate and efficient lung nodule identification in CT imaging.
• This approach offers a promising tool for radiologists in improving diagnostic outcomes for lung cancer.