Fully automated regional lung perfusion quantification in SPECT/CT images with open-source software

BACKGROUND: Nuclear medicine's lung perfusion scintigraphy is a valuable imaging technique for assessing many health conditions. Various methods have been described in the literature for segmenting and quantifying the lung perfusion in single-photon emission computed tomography/computed tomography (SPECT/CT) images, but they rely on commercially available software, require manual definition of regions/volumes of interest, or both. OBJECTIVE: This study proposes a never reported approach to segment and quantify SPECT (and SPECT/CT) lung perfusion images by developing a fully automated algorithm utilizing only free software. METHODS: Python programming language was used to write a completely automated algorithm for 3D Slicer to segment and quantify SPECT and SPECT/CT images. The algorithm was tested in 37 lung perfusion images, collected retrospectively from a public hospital database. RESULTS: The algorithm was able to perform fully automated lobar perfusion quantification. The mean relative perfusion found were: LUL - 23.5%, LLL - 22.3%, RUL - 24.6%, RML - 7.9%, and RLL - 21.7%. The algorithm also segmented and quantified the relative perfusion of the left (L) and right (R) lungs without the aid of CT: L - 44.6% and R - 55.3%; and found no statistical difference in the results obtained with or without CT ( P -value = 0.38 and 0.44, respectively). CONCLUSION: The algorithm created required no user interaction, presented good agreement with previously reported works, and was on average 10 times faster than the fastest algorithm reported on the literature, thus making it a free, efficient, and reliable tool for assisting diagnosis.