Edge map analysis in chest X-rays for automatic pulmonary abnormality screening

Area of Science:

• Medical Imaging
• Radiology
• Computer Vision

Background:

• Tuberculosis (TB) screening is crucial for HIV+ populations, especially in resource-constrained regions.
• Current screening methods may be limited in accessibility and efficiency.

Purpose of the Study:

• To develop an automated method for detecting pulmonary abnormalities using chest radiographs (CXRs).
• To facilitate TB screening in HIV+ populations within resource-limited settings.

Main Methods:

• Analysis of thoracic edge map histograms from posteroanterior chest radiographs (CXRs).
• Utilized gradient orientations, varying bin numbers and pyramid levels.
• Employed five distinct regions-of-interest selections.

Main Results:

• Achieved a maximum abnormality detection accuracy (ACC) of 86.36%.
• Obtained an area under the ROC curve (AUC) of 0.93.
• Processed images at an average rate of 1 second per image.

Conclusions:

• An automated method for pulmonary abnormality screening using CXR thoracic edge maps has been developed.
• The proposed method demonstrates superior performance compared to existing state-of-the-art techniques.