A Lung Volume-Based Perfusion SPECT and CT Comparison Algorithm for Enhanced Pulmonary Embolism Diagnosis

Perfusion SPECT is a medical imaging modality often used alongside other techniques to diagnose pulmonary embolism. In this article, we developed a method that enables the calculation of the ratio between the lung volume derived from the SPECT examination and that obtained from a regular CT scan. This value allows for the quantification of lung regions with impaired perfusion. To compute this metric, it was necessary to segment the lung images from both examinations. For the CT images, we utilized an off-the-shelf algorithm based on U-Net architecture, whereas for the SPECT images, a fixed threshold determined empirically was applied. The subsequent step involved calculating the volume based on the associated metadata from the DICOM files. In a dataset comprising five patients (two healthy and three diseased), the developed algorithm flawlessly distinguished between the two groups. Thus, this method enables the determination of a precise numerical metric quantifying the severity of potential pulmonary embolism.