The study explores a new lung segmentation method aimed at improving the detection of juxta-pleural nodules in chest CT scans, which is a common challenge in the field.
It combines the Chan-Vese model with a Bayesian approach to enhance the accuracy of lung contour predictions and effectively reduce false positives through advanced detection techniques.
The proposed method demonstrated exceptional statistical performance with high sensitivity, specificity, and accuracy, significantly outperforming previous models and providing valuable benefits for computer-aided diagnosis systems.