Fully automatic lung segmentation and rib suppression methods to improve nodule detection in chest radiographs

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Lung segmentation is crucial for lung cancer detection in chest radiographs.
• Overlaying bony structures like ribs can impede accurate diagnosis in CAD systems.
• Existing CAD schemes can benefit from enhanced lung field visualization.

Purpose of the Study:

• To develop an automated method for accurate lung segmentation in Posterior-Anterior (PA) chest radiographs.
• To implement an effective rib suppression technique to improve the conspicuity of lung nodules.
• To evaluate the performance and applicability of the proposed algorithms.

Main Methods:

• Image enhancement using adaptive contrast equalization and non-linear filtering.
• Lung area estimation via morphological operations and region growing for precise contouring.
• Rib suppression utilizing oriented spatial Gabor filters.

Main Results:

• Achieved 96.25% accuracy for lung segmentation on a database of 247 chest radiographs.
• Demonstrated improved conspicuity of lung nodules through rib suppression.
• The method is fully automatic, computationally efficient, and robust to noise.

Conclusions:

• The proposed automated method provides accurate lung segmentation and effective rib suppression for chest radiographs.
• This technique can enhance the performance of computer-aided diagnosis systems, particularly for lung nodule detection.
• Its applicability extends to bedside portable X-rays due to no requirement for additional radiation or specialized equipment.