Perfusion-Ventilation Lung SPECT Image Analysis System Based on Minimum Cross-Entropy Threshold and Watershed Segmentation

A perfusion-ventilation lung SPECT image analysis system is developed in this paper. The system applies the minimum cross-entropy threshold selection method to recognizing the lung from perfusion-ventilation lung SPECT image, and utilizes the watershed segmentation method to separate the right and left pulmonary lobes. In this system, a contour tracing operator is used to detect the edge of pulmonary lobe. The system can automatically divide the pulmonary lobe into 9 regions and calculate the radiation counts and its proportion in the total counts. Finally, the system can manage a medical information database to store the perfusion and ventilation lung SPECT images and the corresponding diagnosis information.