Myocardial assessment in ischemic heart disease encompasses both the anatomical assessment of the cardiac dimensions and structure as well as indirectly assessing coronary artery stenosis severity and CAD chronicity. There is a complex interaction between observed coronary anatomy (i.e., luminal stenosis) and the presence of ischemia. Published data demonstrates that a luminal stenosis ≥ 50% by CCTA correlates poorly with myocardial ischemia by either single-photon emission computed tomography (SPECT) or positron emission tomography (PET) with positive predictive value (PPV) ranging from 29 to 58% [16]. Conversely, ischemia is still present in up to 12% of patients with ≥ 50% stenosis [16]. The same is true for invasive coronary angiography (ICA). Furthermore, revascularization based on ICA stenosis alone does not reduce death or nonfatal MI compared with medical therapy [17]. Physiologic assessment with invasive fractional flow reserve (iFFR) demonstrated that an intervention guided by vessel-specific ischemia for patients with indeterminate stenosis resulted in 33% less percutaneous coronary interventions and 30% improvement in composite cardiovascular outcomes [18, 19]. Given these robust data, many suggest that iFFR is the gold standard for ischemia assessment. The ongoing ISCHEMIA trial (NCT01471522) will inform the discussion regarding outcomes with revascularization based solely on ischemia. In the meantime, CCT with CCTA is positioned as the single modality capable of simultaneously evaluating coronary artery anatomy and CAD burden and assessment of physiologic myocardial blood flow.

SPECT is a static imaging modality that leverages differential distribution and uptake of modest energy (70–120 keV) radiotracers within the myocardium based on differences in coronary blood flow and myocardial viability. SPECT imaging, compared to iFFR, has a sensitivity of 74% and specificity of 79% for the diagnosis of significant obstructive CAD [20]. Important limitations of SPECT imaging include difficulty in diagnosing high-risk CAD in the setting of balanced ischemia (i.e., global low, but homogenous blood flow), poor spatial resolution and image quality in obese patients, and effective radiation doses that average 12–15 mSv for stress-rest protocols [21, 22]. Obesity-related artifacts can be mitigated with attenuation correction or prone imaging, though these techniques can lead to artefactual perfusion defects that require the reader to synthesize data from multiple acquisitions and can increase imaging time [23, 24]. Additionally, several academic centers have implemented protocols to reduce radiation dose to include routine use of half-dose acquisitions resulting in 5–6 mSv doses [25]. The advantages of SPECT imaging are the ability to perform testing in patients that can or cannot exercise, in virtually all heart rhythms, and in known CAD and prior coronary revascularization. Additionally, there is data demonstrating the ability of SPECT to assess viability, albeit with significantly reduced sensitivity when compared to PET or CMR [26•, 27]. Finally, dynamic SPECT techniques currently being validated offer the promise of quantifying myocardial blood flow utilizing SPECT tracers [28].

PET is a versatile nuclear imaging modality that detects high-energy (512 keV) photons that result from an annihilation interaction between a positron and a valence electron. In addition to static perfusion data, the radiotracers Rb-82 and 13N-ammonia can be used to quantify absolute coronary blood flow and coronary flow reserve [29, 30]. Viability assessment can also be performed utilizing the glucose analog fluorodeoxyglucose (FDG) by leveraging the difference in metabolic properties between infarcted and hibernating tissues. When combined with anatomic CCT imaging (CAC and/or CCTA), the diagnostic performance of PET imaging for the diagnosis of CAD is greatly increased with a reported sensitivity of 90% and specificity of 95% [31]. The radiation cost of PET is modest at 2–4 mSv with the primary limitation to more widespread use of this technology limited primarily by the cost, limited scanner locations, limited available readers, and unavailability or expense of stress radiotracers.

CMR is the gold standard for the assessment of cardiac structure and function. Additionally, with emerging applications such as T1 mapping, CMR is the best validated noninvasive modality for tissue characterization. The addition of intravenous gadolinium allows for both first-pass stress imaging, utilizing gradient echo sequences, for the assessment of myocardial ischemia [32, 33]. Compared to SPECT and ICA, stress CMR assessment of ischemia was found to have a sensitivity of 89% for both and specificity of 76 and 87%, respectively [21, 34‐37]. Performance of late gadolinium enhancement (LGE) sequences provides information on the presence and location of myocardial infarction, as well as robust prognostic information. Additionally, the transmural extent of LGE uptake serves as a powerful tool in the evaluation of viability. Beyond the evaluation of ischemic heart disease, mid-myocardial and/or epicardial uptake of LGE can also signal the presence of other infiltrative and inflammatory cardiomyopathies, such as sarcoidosis or idiopathic myocarditis. CMR with or without stress has its limitations. Notably, it is an expensive, time-consuming exam (often requiring 30–60 min), is poorly tolerated in patients with severe claustrophobia, and requires multiple (sometimes prolonged) breath holds, and gadolinium should not be used in patients with renal dysfunction (GFR < 30). Additionally, the presence of ferrometallic materials within the myocardium can create signal voids and limit the diagnostic utility of CMR even in those with MR conditional devices.

Rest first, followed by stress image acquisition protocol, is the most widely used in clinical practice and is best suited for low- to intermediate-risk patients without known CAD (Fig. 1a). This protocol involves an initial rest acquisition similar to simple CCTA in which an initial CAC followed by a prospective, ECG-triggered, contrast-enhanced CCTA is obtained first. Inherent in this is the fact that patients are prepped in a standard fashion with nodal blocking agents and sublingual nitroglycerin. If an indeterminate stenosis is detected, a vasodilator stress dataset is subsequently obtained. Depending on the scanner platform being used, this will either entail a retrospective, ECG-gated acquisition or, on wide-detector scanner platforms, a full R-R interval acquisition. This allows for assessment of any stress-induced wall motion changes. Finally, a delayed, noncontrast-enhanced dataset can be added approximately 10 min after the stress acquisition to evaluate for evidence of infarction. The advantage to this approach is the deferral of the stress acquisition when rest images either show nonobstructive CAD (no stenosis ≥ 50%) or a high-grade stenosis (≥ 70%). If stress imaging is pursued, a delay of 10–20 min following rest imaging should be implemented to ensure adequate contrast washout. ECG-based tube current modulation is recommended to reduce radiation dose [68]. In addition to the evaluation of stable chest pain in the outpatient setting, rest-stress CTP protocols may be ideal for the evaluation of acute chest pain in the emergency department, leveraging both the quality data and high NPV of CCTA in the ED with the ability to further evaluate indeterminate lesions and incrementally increase appropriate disposition [69]. The main limitations to rest-stress CTP protocols are the need to pretreatment with nitroglycerin and nodal blocking agents prior to rest acquisitions, which can mask ischemia, similar to data seen in SPECT imaging [70]. Additionally, residual circulating contrast from rest imaging can contaminate the stress acquisition and hinder the diagnostic performance.

Less commonly used when compared to rest-stress, stress-first CTP is best suited for patients with intermediate to high pretest risk known intermediate/indeterminate stenosis, or prior revascularization where the assessment of ischemia in a particular vascular territory is favored over coronary anatomy (Fig. 1b). When performing stress-first CTP, the pharmacokinetics of the vasodilatory agents being used must be taken into account. Dipyridamole, adenosine, or regadenoson can all be used and achieve hyperemia at various time periods following administration and sustain hyperemia for variable durations. Adenosine, owing to its rapid metabolism and thus rapid offset with cessation of infusion, was used in a majority of the validation studies. Regadenoson is also a viable option and is the preferred agent in SPECT and CMR due to ease of administration and a low side effect profile. The limitation of regadenoson stress-first CTP is to the persistence of heart rate elevation (30–40 min following regadenoson administration), making motion-free imaging of the coronaries challenging. Newer CT scanners can overcome the heart rate elevation associated with regadenoson with the use of motion correction software and faster gantry rotation speeds allowing for stress-only CTP and high-resolution coronary anatomy in a single, stress acquisition, mitigating the need for rest acquisition and thus conserving radiation dose.

Static imaging techniques, with or without stress acquisitions, are limited to single snapshots in time and do not provide comprehensive blood flow analysis. Historically, limitations in scanner technology made static CTP the only viable method. However, the latest generation 256- and 320-row detector platforms allow for imaging of the entire cardiac volume with a stationary table and a single gantry rotation. Additionally, second-generation dual-source CT (DSCT) can cover this same volume utilizing a table shuttle method. The third-generation DSCT has increased z-axis coverage up to 105 mm and, thus, can image the cardiac volume without the need for table shuttling [50, 51, 55, 71]. This technology allows for the performance of first-pass perfusion owing to the ability of these newer generation scanners to acquire full cardiac datasets in short succession, termed dynamic CTP. Dynamic CTP allows for comparison of time-attenuation profiles within myocardial segments, which facilitates direct quantification of myocardial blood flow (MBF) [72]. MBF calculation by dynamic CTP involves mathematic modeling derived from the deconvolution methods used in CMR [52, 73]. In semiquantitative analysis, the time-attenuation curve for a myocardial ROI is derived and a time-to-peak attenuation, attenuation upslope, and area under the curve are calculated. This is the most commonly used semiquantitative method as only the upslope time to peak attenuation is sampled, thus lowering effective radiation dose. Dynamic CTP validation studies, utilizing 320-row MDCT and second-generation DSCT, have shown varying, but mostly positive results in detection of hemodynamically significant CAD when compared against ICA, CMR, and SPECT. Dynamic CTP (Table 1) has demonstrated sensitivities ranging from 58 to 100%, specificities from 74 to 100%, NPV 82–100%, and PPV 43–100% [51, 56, 57, 60]. The biggest limitation of dynamic CTP is the relatively high radiation dose required (8.2 to 18.8 mSv in validation studies) [62, 73]. Dynamic CTP represents an emerging CCT application and further research is needed before more widespread implementation is pursued.

DECT was first introduced in 2008 and has undergone several advancements and innovations in the last decade that have significantly increased its diagnostic utility [74, 75]. DECT is based on the principles of the photoelectric effect and the energy-related attenuation difference of tissues observed with exposing the same sample volume to both a low (typically 80 kV) and high (140 kV) tube voltage. Utilizing monochromatic reconstructions at these differing energy levels, subtle differences in tissue contrast uptake can be more readily detected. Specific to CTP, DECT facilitates creating of an iodine map that serves as a surrogate for blood flow [76]. This is accomplished by utilizing one of four vendor-specific technologies (Fig. 3): two X-ray sources offset by 90° operating at different energy levels, rapid switching utilizing a single source where the X-ray tube cycles rapidly between low and high tube voltage during a single gantry rotation, a dual layer detector model where a single X-ray source provides a spectrum of energy levels in the presence of a double-layered detector configuration that registers only high- and low-energy photons, and gantry rotation kilovolt switching where a single X-ray source scans a full gantry rotation at high- and a full gantry rotation at low-energy settings of the same tissue volume (thus each volume is scanned twice) [77]. With these specialized acquisitions, a virtual monochromatic image (VMI) is generated that is less susceptible to beam hardening and other artifacts while maximizing the superior contrast seen with iodinated agents and soft tissue at low kilovolt settings [78‐80]. DECT can readily delineate the iodinated contrast in the blood pool within the ventricle and within the vessels and absorbed by the myocardium and can then be used to make color-coded maps, similar to SPECT images, that detail myocardial perfusion [76, 81]. Compared with SPECT and single-energy CTP, DECT protocols (Table 2) are observed to have a sensitivity of 82–94%, specificity of 71–94%, PPV of 53–91%, and NPV of 81–97% [84, 85]. Historically, one of the main limitations to DECT was the high required radiation dose and high contrast volume [86]. However, subsequent advancements have shown that the use of ultralow-energy levels (40–50 kV) enhances iodine contrast differences and improves the accuracy of delayed enhancement imaging, particularly for the detection of scar [87]. Several studies of DECT have achieved radiation doses of 0.5 to 4.4 mSv, significantly reduced when compared to early DECT or SPECT [88, 89]. Additionally, no reduction in image quality was observed despite reductions in contrast volume approaching 50% [90, 91]. Currently, DECT for myocardial perfusion is not routinely utilized in clinical practice as further study is ongoing to determine the optimal energy settings and to further investigate the various vendor-specific DECT solutions more thoroughly for cardiac imaging [92‐94].

Post-processing of CTP datasets relies on the visual assessment of the ischemic myocardial segments in comparison to normally perfused myocardium (Fig. 4). Multiplaner reformatted images allow for evaluation in the classic 17 segment model view. Image display settings should be adjusted to thick MPR slabs (3–8 mm) and minimum intensity projection (MinIP) or average HU attenuation projection as opposed to maximum intensity projection (MIP). This allows for more ready identification of ischemic segments. Finally, appropriate window width and level settings (200–300 and 100–150, respectively) should be utilized [39, 95]. These settings optimize the displayed grayscale centering around the normal HU attenuation of the myocardium (average HU of 90–100) and the narrow width accentuates ischemic or infarcted myocardium ranging from subzero HU to 30 HU [96, 97]. TPR (Fig. 2), as discussed above, is a semiquantitative assessment of perfusion that measures the ratio of the average HU of the subendocardial to subepicardial tissue where a normal TPR has been defined as above 1 and a ratio of 0.75 or less suggests ischemia [42]. The combination of DE-CCT with TPR compared to SPECT demonstrates a sensitivity of 86%, specificity of 92%, positive predictive value of 92%, and negative predictive value of 85% for diagnosing clinically significant perfusion defects.