Measurement of Non Perfused Size (Area) of Spect Images Using ImageJ Software

Abstract

Computer-aided diagnosis (CAD) using Image J software has been used to quantify myocardial infarct (MI) area. The study was designed to develop an algorithm for thresholding of myocardial perfusion SPECT image, and estimate an area with possible myocardial infarction from region with normal perfusion. The objective of this paper is to determine the accuracy of ImageJ algorithm for quantification of myocardial infarct area using SPECT images.

A phantom consists of a large container to mimic the human body. With a small cylindrical container inserted in the large container to simulate the heart, inside the small cylindrical container, a silicon rubber and some defects (40 mm² each) were placed to mimic the heart Myocardium and perfusion defects. A ^99mTc – tagged (0.6 ml) radioisotope was mixed with water in the large container for background solution, the small cylindrical container was also filled with water and mixed with ^99mTc – tagged (0.25 ml) radioisotopes, SPECT images were acquired and reconstructed with FBP technique using the Butterworth filter cut off frequency 0.4cm^-1 and order of 10. MI was quantified before and after thresholding using Image J software.

Overestimation of defect area was observed from the image without thresholding, with estimated mean area and standard deviation of about 273.93mm² 28.088, in comparison with the thresholded image with 38.09mm² 0. 657.The thresholded image almost approximate the actual infarct size. At 0 cm position, the mean area and standard deviation were 33.51 mm² 0.436, and at the 10 cm position is about 38.09mm² 0.657. At 10 cm position, the defects areas almost approximate the actual infarcts size. We compare the mean area of the two positions using the paired t-test, a p-value of 0.001 was estimated, which indicates a strong statistical significance difference. The results indicate that, thresholding using ImageJ software not only aid in accurate infarcts quantitation, it also increases the signal to noise ratio, which enhances the contrast of the region of interest.