Solid, part-solid, or non-solid?: classification of pulmonary nodules in low-dose chest computed tomography by a...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Pulmonary nodules require accurate classification for appropriate patient management.
• Distinguishing between solid, part-solid, and non-solid nodules is crucial.
• Current classification relies on expert human interpretation.

Purpose of the Study:

• To develop and validate a computer-aided diagnosis (CAD) tool.
• To automatically classify pulmonary nodules into solid, part-solid, and non-solid types.
• To assess the performance of the CAD tool against human experts.

Main Methods:

• A 3D segmentation-based CAD system was developed.
• Features including intensity, texture, and segmentation were extracted.
• The system classified nodules from low-dose computed tomography scans.
• Performance was evaluated using Cohen κ statistics against radiologists.

Main Results:

• The CAD tool achieved good agreement with human experts (κ range: 0.54-0.72).
• Interobserver agreement among radiologists was comparable (κ range: 0.56-0.81).
• The automated classification demonstrated reliable performance.

Conclusions:

• A novel automated classification tool for pulmonary nodules was successfully developed.
• The CAD tool showed good agreement with expert radiologists.
• Computer-aided diagnosis can assist radiologists in pulmonary nodule workup.