Examining a hypothetical quantitative model for better approximation of culprit coronary artery and site of stenosis on 99mTc-sestamibi gated myocardial perfusion SPECT.

Abstract

A hypothetical quantitative model of analyzing gated myocardial perfusion SPECT is proposed and examined for the feasibility of its use as a predictor of diseased coronary artery and approximating the site of stenosis to determine whether it could serve as a useful noninvasive complement for coronary angiography. The extent and severity of perfusion defects on rest gated myocardial perfusion imaging SPECT-images were assessed on a five-point scale in a standard 17-segment model and total perfusion deficit was quantified by automated software. The first step was to locate the diseased coronary artery using a quantitative method: for this, the score of each segment belonging to a particular coronary artery was determined using a systematic presumptive approach. After determination of specific coronary artery segments, the scores of the contiguous segments in three short axis slices (apical, middle, and basal) were summed for six subdivisions (anterior, anterolateral, inferolateral, inferior, anteroseptal, and inferoseptal). The site of stenosis was determined from (a) the initial approximation of the involved segments with a defect score of 2-4 and (b) subsequent calculation of the defect score of each of the six subdivisions and allocating the site through a preassigned number for each coronary artery. For each coronary artery, only the subdivision with the highest defect score was considered. Proximal, middle, and distal segments of left anterior descending artery (LAD) were considered to be represented when the summed value of a subdivision within a particular arterial territory was more than or equal to 7, between 5 and 7, 5 and 3, respectively. For the left circumflex and right coronary artery, summed scores (of respective subdivisions) of more than or equal to 5 and between 3 and 5 were preassigned to proximal and distal stenosis, respectively. The results were then correlated with the coronary angiographic data. On coronary angiography, proximal LAD occlusion was considered when the site of occlusion occurred proximal to the first diagonal branch and a mid-LAD occlusion from the first diagonal branch to the next diagonal branch. Proximal left circumflex artery occlusion is considered as those located before the first relevant marginal branch and distal left circumflex artery occlusion is beyond the first marginal branch. From the ostium to the origin of the first acute marginal artery was considered the proximal segment of the right coronary artery and thereafter was designated as distal. The result obtained by the quantitative model was compared with the angiographic data along with statistical analysis. The unweighted κ-value was 0.803 and 95% confidence interval was between 0.716 and 0.890. Both the proposed quantitative myocardial perfusion imaging model and the angiographic results were in strong agreement; however, this model needs to be examined in a larger number of patients in a prospective setting to explore its pitfalls and define its practical applicability in future.