Development and validation of a new sampling method to eliminate septal artifacts to improve the accuracy of quantitative analysis

The reliable diagnosis in the treatment of coronary heart disease (CAD) remains a challenging problem. Our purpose is to develop and validate a new sampling method to eliminate septal artifacts to improve the accuracy of the quantitative analysis results of radionuclide myocardial perfusion imaging (MPI), and aid physicians in the diagnosis of CAD. The proposed sampling model simulated the left ventricle as a three-dimensional mixture of hemispheres, cylinders, and irregular shapes to more closely resemble the real shape of the heart. Sampling results were displayed in the form of a bull's-eye map, and the doctor's diagnosis results were used as the gold standard to verify the accuracy of the proposed sampling model in identifying the location and severity of cardiomyopathy. Experimental results showed that the mean accuracy of the proposed method in identifying the location of myocardial lesions and septal segment ischemia was
$86.29\pm 7.41\%$
and 91.15%, respectively, and the accuracy in assessing the severity of myocardial ischemia was 77.5%. Compared with ECToolbox and MPI Quantification Toolbox, the above three indicators had significant differences (
$\mathrm{P} < 0.05$
). In conclusion, the proposed sampling method can improve the accuracy of quantitative analysis, effectively eliminate septal artifacts, and has reliable clinical application value.