Automatic Categorization and Scoring of Solid, Part-Solid and Non-Solid Pulmonary Nodules in CT Images with...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Pulmonary nodules require accurate categorization for effective diagnosis.
• Existing methods for nodule analysis can be prone to errors from image segmentation.
• Computer-aided diagnosis (CADx) systems aim to improve diagnostic accuracy and consistency.

Purpose of the Study:

• To develop and evaluate a Convolutional Neural Network (CNN)-based CADx system for automatic categorization of pulmonary nodules.
• To assess the system's performance in classifying solid, part-solid, and non-solid nodules in CT images.
• To compare the CNN-based system's effectiveness against a traditional histogram analysis (HIST) method.

Main Methods:

• A CNN model was developed to analyze two-dimensional regions of interest (ROIs) of pulmonary nodules from CT scans.
• The CNN model learned hierarchical features directly for classification and regression tasks.
• Two computerized texture analysis schemes (classification and regression) were implemented.
• A histogram analysis (HIST) method was used for comparative evaluation.

Main Results:

• The CNN-based CADx system demonstrated significant performance improvements over the HIST method in both classification and regression tasks.
• Nodule classification and rating performance achieved by the CNN model were concordant with those of practicing radiologists.
• The system effectively categorized solid, part-solid, and non-solid nodules without requiring explicit image segmentation.

Conclusions:

• CNN-based CADx systems offer a promising approach for accurate and automated pulmonary nodule categorization.
• The developed system can reduce inter-observer variation among screening radiologists.
• This AI tool provides a quantitative reference for further nodule analysis and diagnosis.