A computerized analysis system in chest radiography: evaluation of interstitial lung abnormalities

Area of Science:

• Radiology
• Medical Imaging
• Computer-Aided Diagnosis

Background:

• Interstitial lung abnormalities (ILAs) present diagnostic challenges in chest radiography.
• Accurate detection of ILAs is crucial for timely patient management.
• Existing computerized analysis systems have limitations in ILA detection.

Purpose of the Study:

• To evaluate a novel computerized analysis system for detecting ILAs in digitized chest radiography.
• To assess the performance of a combined radiographic index derived from image processing techniques.
• To compare the system's detection performance against high-resolution computed radiography (HRCT) as the gold standard.

Main Methods:

• The system employs four-directional Laplacian-Gaussian filtering, linear opacity judgment, and subtraction.
• A combined radiographic index was calculated from normalized indices.
• Regions of interest (ROIs) from normal, mild ILA, and severe ILA groups were analyzed using HRCT as reference.

Main Results:

• The computerized system demonstrated high performance in detecting ILAs.
• The combined radiographic index achieved an area under the receiver operating characteristic curve (Az) of 0.94 +/- 0.02 for mild ILAs.
• Excellent performance was observed in the severe ILA group with Az = 0.98 +/- 0.01.

Conclusions:

• The developed computerized analysis system significantly improves the detection of interstitial lung abnormalities.
• The combined radiographic index enhances diagnostic performance compared to previous systems.
• This system shows promise for improved accuracy in diagnosing ILAs from chest radiography.