Validation of prone myocardial perfusion SPECT with a variable-focus collimator versus supine myocardial perfusion SPECT with or without computed tomography-derived attenuation correction