[Study on identification of expiratory phase X-ray image]

Area of Science:

• Medical Imaging
• Radiology
• Image Analysis

Background:

• Accurate lung segmentation is crucial for medical diagnosis from X-ray images.
• Existing methods for lung segmentation face challenges due to variations in X-ray exposure and individual differences.

Purpose of the Study:

• To develop and evaluate an improved method for lung segmentation in X-ray images.
• To enhance the accuracy of lung image identification by addressing limitations of previous techniques.

Main Methods:

• Lung segmentation using maximum square error for thresholding.
• Thorax identification via edge-detection and tracking-bug algorithms.
• Lung isolation from the thorax using thresholding segmentation.
• Comparison of two eigenvalue methods: lung gray mean vs. lung-to-bone gray ratio.

Main Results:

• The initial method using the mean gray value in the lung showed limited effectiveness for image identification.
• The novel method, utilizing the ratio of gray values in lungs to bones, demonstrated superior performance.
• Comparative analysis using inter-class distance confirmed the enhanced efficacy of the new method.

Conclusions:

• The proposed method of using the lung-to-bone gray ratio is a more effective eigenvalue for identifying lung images.
• This technique offers improved accuracy in lung segmentation, considering factors like X-ray dosage and individual variations.