IJC Heart & Vasculature 32 (2021) 100721 Contents lists available at ScienceDirect IJC Heart & Vasculature journal homepage: www.journals.elsevier.com/ijc-heart-and-vasculature Diagnostic value of quantitative parameters for myocardial perfusion assessment in patients with suspected coronary artery disease by single- and dual-energy computed tomography myocardial perfusion imaging Aroa Ruiz-Muñoz a,b,c,d, Filipa Valente a,b,c,d, Lydia Dux-Santoy a,b,c,d, Andrea Guala a,b,c,d, Gisela Teixidó-Turà a,b,c,d, Laura Galián-Gay a,b,c,d, Laura Gutiérrez a,b,c,d, Rubén Fernández-Galera a,b,c,d, Guillem Casas a,b,c,d, Teresa González-Alujas a,b,c,d, Ignacio Ferreira-González a,c,d,e, Arturo Evangelista a,b,c,d, José Rodríguez-Palomares a,b,c,d,⇑ a Vall d’Hebron Institut de Recerca (VHIR), Barcelona, Spain b CIBER-CV, Instituto de Salud Carlos III, Madrid, Spain c Department of Cardiology, Hospital Universitari Vall d’Hebron, Barcelona, Spain d Universitat Autònoma de Barcelona, Barcelona, Spain e CIBERESP, Instituto de Salud Carlos III, Madrid, Spain a r t i c l e i n f o a b s t r a c t Article history: Purpose: To compare performance of visual and quantitative analyses for detecting myocardial ischaemia Received 3 December 2020 from single- and dual-energy computed tomography (CT) in patients with suspected coronary artery dis- Received in revised form 9 January 2021 ease (CAD). Accepted 14 January 2021 Methods: Eighty-four patients with suspected CAD were scheduled for dual-energy cardiac CT at rest (CTA) and pharmacological stress (CTP). Myocardial CT perfusion was analysed visually and using three parameters: mean attenuation density (MA), transmural perfusion ratio (TPR) and myocardial perfusion Keywords: reserve index (MPRI), on both single-energy CT and CT-based iodine images. Significant CAD was defined Coronary artery disease Myocardial ischaemia in AHA-segments by concomitant myocardial hypoperfusion identified visually or quantitatively CT myocardial perfusion imaging (parameter < threshold) and coronary stenosis detected by CTA. Single-photon emission CT and invasive Transmural perfusion ratio coronary angiography were used as reference. Perfusion-parameter cut-off values were calculated in a Dual-energy CT-based iodine imaging randomly-selected subgroup of 30 patients. Results: The best-performing thresholds for TPR, MPRI and MA were 0.96, 23 and 0.5 for single-energy CT and 0.97, 47 and 0.3 for iodine imaging. For both CT-imaging modalities, TPR yielded the highest area under receiver operating characteristic curve (AUC) (0.99 and 0.97 for single-energy CT and iodine imag- ing, respectively, in vessel-based analysis) compared to visual analysis, MA and MPRI. Visual interpreta- tion on iodine imaging resulted in higher AUC compared to that on single-energy CT in per-vessel (AUC: 0.93 vs 0.86, respectively) and per-patient (0.94 vs 0.93) analyses. Conclusion: Transmural perfusion ratio on both CT-imaging modalities is the best-performing parameter for detecting myocardial ischaemia compared to visual method and other perfusion parameters. Visual analysis on CT-based iodine imaging outperforms that on single-energy CT. Ó 2021 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction ⇑ Corresponding author at: Department of Cardiology, Hospital Universitari Vall d’Hebron, Paseo Vall d’Hebron 119-129, 08035 Barcelona, Spain. Coronary artery disease (CAD) management has improved over E-mail addresses: aroa.ruiz.munoz@gmail.com (A. Ruiz-Muñoz), filipaxavierva- time; however CAD remains the leading cause of morbidity and lente@gmail.com (F. Valente), lydia.dux@gmail.com (L. Dux-Santoy), andrea. mortality in the western world [1]. Computed tomography angiog- guala@yahoo.com (A. Guala), gteixido@vhebron.net (G. Teixidó-Turà), lauragalian- gay@gmail.com (L. Galián-Gay), lauraguga@gmail.com (L. Gutiérrez), rubenfdezga- raphy (CTA) is considered an accurate, non-invasive method for the lera@gmail.com (R. Fernández-Galera), gcasasmasnou@gmail.com (G. Casas), anatomical evaluation of luminal narrowing in coronary arteries. mtegonzalez@vhebron.net (T. González-Alujas), iferreir@vhebron.net (I. Ferreira- Nonetheless, in patients with known CAD, the addition of com- González), arturevangelistamasip@gmail.com (A. Evangelista), jfrodriguezpalomar- puted tomography (CT) myocardial perfusion imaging may es@gmail.com (J. Rodríguez-Palomares). https://doi.org/10.1016/j.ijcha.2021.100721 2352-9067/Ó 2021 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 improve the detection of haemodynamically-significant stenosis 2.2.2. CTA/CTP [2]. Patients underwent a rest coronary CT angiography (CTA) fol- CT myocardial perfusion imaging is usually based on visual lowed by a Dipyridamole stress-induced CT study (CTP). For CTA, interpretation of left ventricular contrast enhancement on both patients were given sublingual nitroglycerine if systolic blood rest and stress-induced CT images (CTA and CTP, respectively). pressure was > 100 mmHg, and intravenous beta-blockers (Esmo- However, a quantitative approach could avoid the limitations of lol, 1–3 mg/kg) if heart rate was > 60 bpm. Contrast (Iomeron 400) subjective visual estimation [3]. Several parameters such as trans- was administered via a cubital vein, with a flow of 5.5 mL/second mural perfusion ratio (TPR) [3–8], myocardial perfusion reserve and volume between 80 and 100 mL depending on the patient’s index (MPRI) [9] and mean attenuation (MA) have been proposed weight. CTP was acquired once Dipyridamole was injected for the quantitative analysis of myocardial perfusion. Nevertheless, (0.84 mg/Kg over 6 min [12]) into a cubital vein followed by comparative studies indicating which parameter is more accurate 80 mL of iodine contrast (Iomeron 400) at 4.0–5.0 mL/second for are lacking. each of the separate, axial, prospectively ECG-triggered acquisi- CT myocardial perfusion imaging can be performed using dual- tions [2]. Then, all patients received intravenous Aminophylline energy CT, allowing mapping of iodine distribution in the myocar- (250 mg/10 mL) to reverse the effect of Dipyridamole. dium that can be used as a quantitative perfusion marker [10]. The scans were acquired on a 64-slice CT scanner (Discovery CT Nonetheless, the benefits of using CT-based iodine images com- 750 High Definition; General Electrics (GE); WI; USA) that includes pared to native CT scans for quantitative assessment of myocardial an x-ray source which can switch energy between 80 and 140 kVp ischaemia have not been sufficiently studied. within 0.3–0.5 ms, and is complemented by a gemstone detector This study aimed (1) to compare and rank the performance of with fast primary speed and low afterglow. quantitative perfusion parameters and visual analysis in the evalu- The scanning protocol was adapted to individual body mass ation of myocardial ischaemia by both single- and dual-energy index (calculated as weight divided by square of heigh; kg/m2) CTA/CTP images in a cohort of patients with chest pain and (2) to using Gemstone Spectral Imaging (GSI) presets as previously ascertain whether the use of dual-energy CT-based iodine imaging reported in detail [13]. In brief, tube current ranged from 375 to has advantages over analysis with single-energy CT scans in the 640 mA, while GSI mode with fast tube voltage switching between identification of myocardial ischaemia. 80 and 140 kVp on adjacent views during a single rotation resulted in a mean tube voltage ranging from 105 to 112 kV. The following scanning parameters were used in addition: axial scan mode with 2. Materials and methods 64  0.625 mm, gantry rotation time of 350 ms and temporal res- olution of 175 ms. 2.1. Population and study design From the acquired data, conventional polychromatic images corresponding to the 140 kVp tube voltage as well as monochro- Eighty-four consecutive patients with chest pain, a prior matic image sets from the GSI data file (50, 60, 70, 80, 90, 100, myocardial perfusion single-photon emission computed tomogra- 120 a 140 keV) were reconstructed using Adaptive Statistical Iter- phy study, and referred for invasive coronary angiography (ICA) ative Reconstruction (ASiR) technology [14]. All these data were were enrolled in the study. Exclusion criteria were atrial fibrilla- transferred to a dedicated workstation (AW VolumeShare5 work- tion, supraventricular arrhythmias, high-grade atrioventricular station, General Electrics (GE), WI, USA) to obtain CT-based iodine block, renal insufficiency (estimated glomerular filtration images and the short-axis views of the left ventricle (LV) (basal, rate < 30 mL/min), chronic obstructive pulmonary disease, signifi- mid-cavity and apical) on both single-energy CT (Hounsfield Units) cant aortic stenosis, known CAD, coronary percutaneous or surgical and iodine images for the posterior analysis of myocardial revascularisation and contraindication to iodinated contrast. perfusion. Before ICA, patients were scheduled for cardiac CT, including dual-energy CTA (rest acquisition) and CTP (stress-induced acqui- 2.3. Myocardial perfusion assessment sition) (Fig. 1). Monochromatic images generated from the dual- energy CT scan were used as a surrogate of single-energy CT stud- Myocardial perfusion defect was defined as the concurrent ies. Myocardial perfusion was assessed by visual interpretation and presence of two conditions: (1) hypoperfusion in the myocar- using three quantitative parameters (TPR, MPRI, and MA) at per- dium identified through visual or quantitative analysis patient and per-vessel levels. Firstly, parameter cut-off values were (parameter < threshold), and (2) a corresponding significant calculated from a randomly-selected test group of 30 patients. coronary stenosis with > 50% or > 70% reduction in vessel lumen Qualitative and different quantitative analyses were compared in on CTA. Each myocardial segment was assigned to one major epi- a validation group of other 54 patients using SPECT + ICA as refer- cardial coronary artery - left anterior descending artery (LAD), left ence. The study protocol was approved by the institutional Ethics circumflex artery (LCX) and right coronary artery (RCA) - in accor- Committee and all participants provided written informed consent. dance with previous recommendations [15] and coronary dominancy. 2.2. Imaging protocols The hypoperfusion in the myocardium was assessed by both single-energy CT and CT-based iodine studies. The LV on CTA (rest 2.2.1. SPECT acquisition) and CTP (stress-induced acquisition) was divided into SPECT studies were carried out according to current recommen- equal thirds perpendicular to the long axis using anatomical land- dations [11], following a single-day stress-rest protocol using a marks and following the published recommendations [15] (Fig. 2). technetium-99 m radiotracer. Treadmill exercise was performed Then the short-axis images of the LV (basal, mid-cavity and apical according to the Bruce protocol. Image acquisition was conducted section) were divided into myocardial segments according to the with an Elscint SP4 nuclear gamma camera with a high-resolution 16-AHA segmentation model of the LV [15] and assessed for collimator and short- and long-axis view images were obtained ischaemia both visually and quantitatively by an experienced and analysed by two experts. By consensus, SPECT was considered observer (Figs. 1 and 2). More details about AHA-segmentation in positive if a perfusion defect was shown in at least two of the three Appendices A1. axes or in three consecutive views from the same axis with total or Coronary stenosis was evaluated on CTA based on the 17 coro- partial reversibility at rest. nary segment model according to SCCT recommendations [16]. 2 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Fig. 1. Algorithm illustrating Computed Tomography (CT) myocardial perfusion analysis. *Significant coronary stenosis defined with >50% or >70% reduction in vessel lumen on Invasive Coronary Angiography (ICA) or Computed Tomography Angiography (CTA). Abbreviations: AHA = American Heart Association; CTA = Computed Tomography Angiography; CTP = Computed Tomography Myocardial Perfusion study; end = endocardium; epi = epicardium; HU = Hounsfield Units; ICA = Invasive Coronary Angiography; MA = Mean Attenuation; MPRI = Myocardial Perfusion Reserve Index; myo = whole myocardium; R = rest; S = stress; SPECT = Single Photon Emission Computed Tomography; TPR = Transmural Perfusion Ratio. 3 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Fig. 2. Quantitative myocardial perfusion analysis. (A) Myocardial perfusion was assessed using three short-axis views of the left ventricle (basal, mid and apical) both at rest and under stress by an experienced observer. (B) For each of the six short-axis views (3 at rest and 3 at stress), three reference points were selected. (C) The left ventricle (LV) wall segmentation was done by Region Growing and using an in-house Matlab code (MathWorks, Massachusetts, US). (D) The mask and reference points were transformed from Cartesian into Polar Coordinate System and the angles of reference points were used to extract myocardial AHA segments. Mean attenuation (MA), transmural perfusion ratio (TPR) and myocardial perfusion reserve index (MPRI) were computed; once attenuation at the subendocardium, subepicardium and the whole myocardium were quantified automatically. (E) Ischaemic perfusion defects were defined as the concurrent presence of two conditions: myocardial hypoperfusion identified through quantitative analysis (parameter < reference threshold), and a corresponding significant coronary stenosis with > 50% or > 70% reduction in vessel lumen on CTA. 4 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Stenosis was considered significant if coronary vessel diameter (c) Myocardial perfusion reserve index (MPRI): the percentage was reduced by >50% or >70%, as recommended [2,17]. of MA at stress minus MA at rest divided by MA at rest [9]. Diagnostic accuracy of CTA/CTP was evaluated at both per- patient and per-vessel levels with the latter being based on a The quantitative approach which uses short-axis views of the three-vessel territory model including LAD, LCX and RCA. The per- left ventricle as input images was based on an in-house Matlab formance of CTA/CTP was compared with SPECT + ICA, considered code (MathWorks, Massachusetts, US) that allowed to: (1) select here as a reference. A perfusion defect was defined if SPECT iden- three anatomical reference points for automatic AHA segmenta- tified hypoperfusion in a region of the myocardium supplied by a tion, (2) segment the left ventricle wall semi-automatically, (3) stenotic main coronary artery evaluated by ICA (Fig. 1). define each AHA segment automatically, (4) measure LV myocar- dial perfusion in each AHA-segment in terms of MA, TPR and MPRI 2.3.1. Visual assessment and (5) consider an AHA-segment to be hypoperfused if the corre- Visual analysis of myocardial perfusion was made in the valida- sponding CTP parameter was below one established threshold in tion group. AHA-segments with hypoperfusion were identified as this region (Fig. 2). Parameter-specific cut-off values were obtained hypoattenuating (darker) areas containing reduced amounts of from the test group (Fig. 1). For each parameter, a large range of contrast material on stress (reversible ischaemia) or on both rest possible cut-off values were defined and their performances were and stress (necrosis) images [3]. assessed in terms of Youden index in a vessel-based analysis. The best threshold was defined as the one with the highest Youden 2.3.2. Semi-automatic, quantitative assessment Index. Further details on Appendices (A1, A2 and A3). Quantitative myocardial perfusion analysis was made using the following parameters: 2.4. Statistical analysis (a) Mean attenuation (MA): mean value of the attenuation of The Kolmogorov-Smirnov test was used to evaluate distribution pixels included in the subendocardium (MAendR/S), subepi- normality. Continuous demographic variables were expressed as cardium (MAepiR/S), and whole myocardium (MAmyoR/S). This mean  standard deviation (SD) if normally distributed and as parameter was evaluated in both rest (R) and stress (S) median [first-third] non-adjusted quartiles otherwise. The Mann- images. Whitney U test was used with non-normal continuous variables (b) Transmural perfusion ratio (TPR): the MA ratio in the whereas Fisher’s exact test or bivariate logistic regression analysis subendocardium relative to the MA of the entire subepi- was applied for non-normal categorical variables. cardium of any given short-axis slice [3]. This parameter Qualitative and quantitative approaches by single-energy CT or was evaluated in both rest (R) and stress (S) images. CT-based iodine imaging were compared in terms of accuracy, sen- Table 1 sitivity, specificity, positive and negative predictive values (PPV Baseline characteristics. and NPV, respectively), Youden index and the area under the recei- ver operating characteristic (ROC) curve (AUC). AUC(s) were com- Characteristic Test group Validation group p- Value pared using the approach of DeLong et al. [18]. All data are reported with 95% confidence intervals. Epidat (Xunta de Galicia, N1 Number of subjects 30 54 - N2 Number of vessels 90 162 - Spain; OPS-OMS; Universidad CES, Colombia) was used for the Sex (male) 17 (56.7%) 36 (66.7%) 0.145 analysis. Age (years) 71.0 [60.0;74.0] 71 [62.0;74.5] 0.421 Body mass index (kg/m2) 28.0 [25.0;30.75] 28.0 [26.0;30.0] 0.850 Hypertension 24 (80.0%) 39 (72.2%) 0.124 3. Results Diabetes mellitus 12 (40.0%) 17 (31.5%) 0.145 Dyslipidaemia 22 (73.3%) 37 (68.5%) 0.372 Smoking history 3.1. Population description Current smoker 4 (13.3%) 4 (7.4%) 0.999 Former smoker 5 (16.7%) 12 (22.2%) 0.453 Demographic and clinical characteristics of both test and vali- Non-smoker 19 (63.3%) 35(64.8%) 0.754 Family history of CAD 1 (1.9%) 1 (1.9%) 0.453 dation groups are presented in Table 1. No significant differences Angina at presentation were observed in patient characteristics and comorbidities. All No chest pain 4 (13.3%) 10 (18.5%) 0.643 patients completed the CTA/CTP protocol with no complications. Chest pain 9 (30.0%) 20 (37.0%) 0.488 Atypical chest pain 9 (30.0%) 13 (24.1%) 0.582 In the validation group, 8 (4.9%) LAD vessels showed significant Nonanginal chest pain 3 (10.0%) 3 (5.6%) 0.590 coronary stenosis (luminal reduction >70%) by ICA and 4 (2.5%) Fatigue or dyspnoea on exertion4 (13.3%) 7 (13.0%) 0.999 of these were also associated with at least one positive myocardial Positive SPECT findings 15 (50.0%) 22 (40.7%) 0.166 segment on SPECT. Regarding LCX vessels, 7 (4.3%) showed >70% Coronary stenosis > 50% by lumen narrowing by ICA and 5 (3%) of these were related to hypop- ICA: LAD 11 (36.7%) 17 (31.5%) 0.681 erfusion by SPECT. Significant stenosis in the RCA was found in 9 LCX 5 (16.7%) 7 (13.0%) 0.643 cases (5.6%), 8 of which (4.9%) were associated with a perfusion RCA 7 (23.3%) 13 (24.1%) 0.955 defect by SPECT (Table 1). Coronary stenosis > 70% by ICA: LAD 5 (16.7%) 8 (14.8%) 0.810 3.2. Cut-off values for quantitative perfusion parameters LCX 5 (16.7%) 7 (13.0%) 0.643 RCA 6 (20.0%) 9 (16.7%) 0.725 Radiation dose in CT (mSv) 6.5 [4.9;7.8] 6.4 [4.7;7.9] 0.671 The cut-off values for perfusion parameters to distinguish Total contrast dose (mL) 170.0 [160.0;180.0]170.0 [167.5;180.0]0.573 normally-perfused from hypoperfused myocardial tissue were Values are reported as N (%) or median [interquartile range]. determined in a randomly-selected test group of 30 patients Abbreviations: CAD = Coronary artery disease; CT = Computed Tomography; (Appendices A3). The best-performing thresholds for TPR, MPRI LAD = left anterior descending; LCX = left circumflex artery; RCA = right coronary and MA were 0.96, 23 and 0.5 for single-energy CT, and 0.97, 47 artery; SPECT = single photon emission computed tomography. and 0.3 for iodine images, respectively. 5 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 3.3. Myocardial perfusion analysis from single-energy CT These results were not affected by the choice of the 50% or the 70% coronary stenosis thresholds, the ones with the 50% are shown The diagnostic performance of CTA/CTP compared with in Table A3 (Appendices). SPECT + ICA was assessed in the 54 patients forming the validation group. The number of artefacts (motion artefacts, beam hardening, 3.4. Myocardial perfusion analysis from dual-energy CT-based iodine etc) identified through visual interpretation was minimal: only 3% imaging of myocardial segments, which were excluded from the analyses. Dual-energy CT was only performed in 40 patients out of the 54 3.3.1. Vessel-based analysis included in the validation group. The vessel-based diagnostic performance of CTA/CTP for myocardial perfusion assessment with >70% of lumen narrowing 3.4.1. Vessel-based analysis to determine significant coronary stenosis and the use of single- The vessel-based diagnostic performance of iodine-based CTA/ energy CT images is shown in Table 2. Visual analysis proved to CTP strategy for myocardial perfusion assessment with a coronary be an effective and appropriate approach for detecting perfusion stenosis threshold of 70% is presented in Table 2. The AUC obtained defects with high performance in terms of accuracy and AUC. with visual analysis of iodine images were higher than those TPR outperformed the visual method and the other quantitative obtained with single-energy CT images [0.93 (0.84–1.02) vs. 0.86 parameters, thereby providing the highest values of sensitivity, (0.76–0.97); p = 0.367]. PPV and Youden index (Table 2). TPR resulted in the highest AUC TPR provided the best performance [AUC: 0.97 (0.95–0.99)] [0.99 (0.98–1.00)] compared to that for the visual analysis [0.86 compared with visual method [AUC: 0.93 (0.84–1.02); p = 0.363], (0.76–0.97); p = 0.025 from Delong‘s test], MPRI [0.95 (0.89– MA [AUC: 0.90 (0.81–1.00); p = 0.167] and MPRI [AUC: 0.94 1.01); p = 0.291] and MA [0.98 (0.96–1.00); p = 0.521]. MPRI (0.87–1.01); p = 0.437] (Table 2). Although specificity and NPV showed the lowest sensitivity [94% (79–100)] and AUC [0.95 were similar among all approaches; visual method, MPRI and MA (0.89–1.01)] among the other quantitative parameters, and TPR resulted in lower sensitivity, Youden index and AUC compared to provided the highest PPV [81% (62–100)]. Specificity, NPV and TPR (Table 2). accuracy were similar in all cases, although TPR and MA resulted The 70% or the 50% of lumen narrowing to determine significant in the highest Youden index and AUC (Table 2). coronary stenosis did not affect the CTA/CTP performing results. The diagnostic performance of per-vessel CTA/CTP strategy with Visual analysis on CT-based iodine images resulted in higher AUC a lumen narrowing of 50% to define significant coronary stenosis is compared to that on single-energy CT images [0.89 (0.82–0.97) shown in Table A2 (Appendices). Similar results than those vs. 0.86 (0.77–0.95); p = 0.564] for the 50% coronary stenosis obtained with the 70% threshold were found: myocardial perfusion threshold. TPR yielded higher Youden index and AUC than visual assessment was feasible with visual analysis, although quantitative analysis and the other quantitative parameters (Table A2 analysis with TPR yielded better results than visual interpretation Appendices). [AUCs: 0.93 (0.86–0.98) vs. 0.86 (0.77–0.95), respectively; p = 0.264]. A case example showing myocardial perfusion assess- 3.4.2. Patient-based analysis ment with TPR, MA and MPRI is shown in Fig. 3 (see Fig. 4). The patient-based diagnostic performance of iodine-based CTA/ CTP for myocardial perfusion assessment with 70% of coronary 3.3.2. Patient-based analysis stenosis threshold are shown in Table 3. As in per-vessel analysis, The patient-based diagnostic performance of CTA/CTP for TPR resulted to be the best-performing method. MPRI and MA per- myocardial perfusion assessment with 70% coronary stenosis cut- formed the worst in terms of Youden index and AUC. off is shown in Table 3. Ischaemia assessment was feasible through When a 50% of lumen narrowing was used to define significant visual interpretation of single-energy CT with high specificity, PPV, coronary stenosis, TPR resulted in higher sensitivity, Youden index accuracy and AUC. TPR and MA proved to be the best-performing and AUC compared to MPRI and MA (Table A3 Appendices). parameters in terms of sensitivity, Youden index and AUC (Table 3). Two false positives were found: one may be a false-negative of 4. Discussion SPECT and the other one as a result of the inhomogeneous distribu- tion of contrast. Youden index and AUC of TPR and MA were sub- In patients with or without suspicion of CAD, a single CT exam- stantially higher compared to visual analysis and MPRI (Table 3). ination including CTA and CTP has proven to be effective for detect- Table 2 Performance of visual and quantitative myocardial perfusion vessel-based analysis by single- and dual-energy CT. Single-energy CT images CT-based iodine images Visual analysis TPR MPRI MA [HU] Visual analysis TPR MPRI MA [mg/mL] Threshold 0.96 23 0.5 0.97 47 0.3 Sensitivity 76 (53–100) 100 (97–100) 94 (79–100) 100 (97–100) 87 (66–100) 100 (97–100) 93 (77–100) 87 (66–100) Specificity 97 (93–100) 97 (94–100) 97 (93–100) 96 (92–99) 99 (97–100) 94 (89–99) 95 (91–100) 94 (89–99) PPV 72 (49–96) 81 (62–100) 76 (56–97) 74 (54–94) 93 (76–100) 71 (50–93) 74 (51–96) 68 (45–92) NPV 97 (94–100) 100 (99–100) 99 (98–100) 100 (99–100) 98 (95–100) 100 (99–100) 99 (97–100) 98 (95–100) Accuracy 94 (91–98) 98 (95–100) 96 (93–100) 96 (93–100) 97 (94–100) 95 (91–99) 95 (91–99) 93 (88–98) Youden Index 0.73 (0.53–0.93) 0.97 (0.95–1.00) 0.91 (0.79–1.02) 0.96 (0.93–0.99) 0.86 (0.68–1.03) 0.94 (0.90–0.99) 0.89 (0.75–1.02) 0.81 (0.63–0.99) AUC on ROC 0.86 (0.76–0.97) 0.99 (0.98–1.00) 0.95 (0.89–1.01) 0.98 (0.96–1.00) 0.93 (0.84–1.02) 0.97 (0.95–0.99) 0.94 (0.87–1.01) 0.90 (0.81–1.00) Values are % (95% confidence interval). Abbreviations: AUC = area under the curve; CTA = Computed tomography angiography; CT = Computed tomography; HU = Hounsfield Units; MA = mean attenuation; MPRI = myocardial perfusion reserve index; PPV and NPV = positive and negative predictive values, respectively; ROC = receiver-operating characteristic; TPR = transmural perfusion ratio. 6 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 on dual-energy CT-based iodine images performed better than visual analysis on single-energy CT; and (3) on both CT-imaging modalities TPR was the best-performing method among visual analysis and the other parameters for detecting myocardial ischaemia. In this study, rest CTA images were acquired before CTP as sug- gested in several studies such as the CORE320 [2] to facilitate translation to clinical care. Patients with normal or near-normal coronary vessels in CTA images will not generally need perfusion imaging, and patients with intermediate degrees of coronary stenosis would then proceed to having a stress CTP for clinical decision-making. Moreover, if CTA precedes CTP, the assessment of coronary stenosis on CTA should be more accurate and CTA image quality may not be affected by the high cardiac frequency as effect of Dipyridamole. 4.1. Diagnostic performance with single-energy CT Of note, the conclusions of the present study were not substan- tially affected by the choice of the 50% or 70% threshold. Visual analysis of CTA/CTP images for myocardial perfusion assessment in addition to coronary anatomy information by CTA was as useful as SPECT + ICA in CAD diagnosis. Our results in terms of specificity and NPV, which neared 95% for both coronary thresh- olds (>50% and >70%), are in line with those of two studies based on the CORE320 data. In those works, in which CTA/CTP + coronary CTA (>50%) was compared to SPECT + ICA, specificity values were close to 90% [2,19]. TPR outperformed visual analysis as well as MA and MPRI. This result, which held true for both 50% and 70% stenosis thresholds, supports the use of TPR compared to MA and MPRI for an ischae- mia CT-image assessment. The diagnostic performance of TPR (<0.85) + coronary CTA(>70%) was assessed by Cury et al. [5] using SPECT + ICA as the reference method. The results of the present study (Table 2) are in line with those of Cury et al, showing similar sensitivity [94% (71–100)] and PPV [89% (64–99)] in the per- patient analysis. However, they reported lower specificity [78% (40–97)] and AUC [0.85 (0.66–0.96)] compared to those obtained in this work. Several authors have reported the good performance of TPR on detecting ischaemia [4,5,20]. However, whether quantitative Fig. 3. Case example showing quantitative myocardial perfusion analyses with myocardial perfusion assessment is superior to visual analysis is transmural perfusion ratio (TPR), mean attenuation (MA) and myocardial perfusion not well established. Some authors found the potential of TPR reserve index (MPRI) from single-energy CT in a 49-year-old patient with typical chest pain. Static single-energy CT images at the mid-ventricular short-axis view at [4,5,20], whereas others did not find its incremental value [3,6]. Dipyridamole-induced stress (CTP) (left) and at rest (CTA) (right). 16-AHA Pontone et al. [6] found that the addition of stress-CTP with visual segmentation of the left ventricle wall was performed. (A) TPR was calculated evaluation to coronary CTA imaging had similar diagnostic perfor- automatically as the ratio of the segment-specific mean attenuation (MA) in the mance on detecting ischaemia when compared with TPR. These subendocardium to the MA of the entire subepicardial layer at stress (left) and rest discrepancies may be related to (i) the use of different thresholds (right). (B) MA was quantified at the subendocardium (red values) and the subepicardium (green values) at stress (left) and rest (right). (C) MPRI was for TPR, (ii) different reference methods for validation (fractional computed as the percentage of MA at stress minus MA at rest divided by MA at rest. flow reserve (FFR), SPECT, SPECT + ICA, CMR, etc.), (iii) differences AHA-segments associated with TPR < 0.96, MA < mean-0.5standard deviation (of all in the study population and (iv) different thresholds for the defini- subendocardial MA values at rest or stress) or MPRI < 26 are shown in purple, tion of significant coronary stenosis. yellow and blue, respectively. Significant coronary stenosis (>70% lumen narrow- ing) in the proximal left anterior descending artery (LAD) was confirmed by CTA. (For interpretation of the references to colour in this figure legend, the reader is 4.2. Diagnostic performance with CT-based iodine imaging referred to the web version of this article.) This study is the first to quantify TPR, MPRI and MA from CT- based iodine images and demonstrate that TPR resulted in higher ing both coronary stenosis and myocardial hypoperfusion [2]. To sensitivity, Youden index and AUC than visual interpretation. It the best of authors’ knowledge, this is the first study to compare was also observed that visual analysis on iodine images provided visual interpretation and three previously-proposed perfusion better results than the visual interpretation on single-energy CT parameters (TPR, MPRI, and MA) [3–5,9] and compare their perfor- images. mance in ischaemia assessment for both single-energy CT and CT- Myocardial ischaemia based on visual interpretation of iodine based iodine studies. distribution has not been extensively studied. Ko et al. [21] demon- The main findings of this study were that: (1) visual interpreta- strated the incremental diagnostic value of combined coronary CTA tion on single-energy CT was effective in detecting myocardial and iodine CTP compared to coronary CTA alone for the detection ischaemia compared with SPECT + ICA; (2) visual identification of significant coronary stenosis. Compared to present results, sim- 7 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Fig. 4. Stress single-energy (first column) and dual-energy (second column) CT perfusion (CTP), and SPECT (third column) for a 49-year-old patient without prior history of coronary artery disease and with chest pain symptoms. Both single- and dual-energy CTP studies as well as SPECT at the mid-ventricular short-axis view revealed ischemia in the mid anteroseptal and inferoseptal segments. Table 3 Performance of visual and quantitative myocardial perfusion patient-based analysis by single- and dual-energy CT. Single-energy CT images CT-based iodine images Visual analysis TPR MPRI MA [HU] Visual analysis TPR MPRI MA [mg/mL] Threshold 0.96 23 0.5 0.97 47 0.3 Sensitivity 90 (66–100) 100 (95–100) 90 (66–100) 100 (95–100) 88 (58–100) 100 (94–100) 89 (63–100) 89 (63–100) Specificity 95 (88–100) 95 (88–100) 93 (85–100) 95 (88–100) 100 (98–100) 94 (83–100) 94 (83–100) 94 (83–100) PPV 82 (54–100) 83 (58–100) 75 (46–100) 83 (58–100) 100 (93–100) 82 (54–100) 80 (50–100) 80 (50–100) NPV 98 (92–100) 100 (99–100) 98 (92–100) 100 (99–100) 97 (90–100) 100 (98–100) 97 (89–100) 97 (89–100) Accuracy 94 (87–100) 96 (90–100) 93 (85–100) 96 (90–100) 98 (91–100) 95 (87–100) 93 (83–100) 93 (83–100) Youden Index 0.85 (0.66–1.05) 0.95 (0.89–1.02) 0.83 (0.64–1.03) 0.95 (0.89–1.02) 0.88 (0.65–1.10) 0.94 (0.85–1.02) 0.82 (0.60–1.05) 0.82 (0.60–1.05) AUC on ROC 0.93 (0.82–1.03) 0.98 (0.95–1.00) 0.92 (0.81–1.02) 0.98 (0.95–1.01) 0.94 (0.81–1.06) 0.97 (0.92–1.01) 0.92 (0.79–1.03) 0.92 (0.79–1.03) Values are % (95% CI). Abbreviations: AUC = area under the curve; CI = confidence interval; CTA = Computed tomography angiography; CT = Computed tomography; HU = Hounsfield Units; MA = mean attenuation; MPRI = myocardial perfusion reserve index; PPV and NPV = positive and negative predictive values, respectively; ROC = receiver-operating characteristic; TPR = transmural perfusion ratio. ilar sensitivity [88% (81–93)] and NPV [91% (86–95)] but lower stenosis (lumen narrowing >50%). Thirteen and 19 of them (1.5% specificity [79% (73–85%)], PPV [73% (65–80)] and AUC [0.84 and 2.9%, respectively) were also identified as ischaemic segments (0.79–0.88)] were obtained. by visual analysis on single-energy CT and iodine CT-based images, Our analyses revealed that the use of iodine as a perfusion mea- respectively. The hypoperfused segments detected by SPECT and sure led to a reduction in the number of artefacts, better image not by CT or vice versa without significant coronary stenosis may quality, fewer false-negatives and a higher detection rate of perfu- be false-positive or false-negative results of one of these tech- sion defects which might be misdiagnosed by SPECT + ICA. niques. Conversely, ischaemic myocardial segments identified by both SPECT and CT without significant coronary stenosis may be 4.3. Study limitations due to microvascular ischaemia. This last statement may be inves- tigated with the use of another technique for detecting myocardial First, ICA + FFR would have been the best reference method. ischemia, such as magnetic resonance imaging, which is not avail- Nonetheless, SPECT + ICA was used here because the study popula- able in this study. tion had already had a SPECT to detect ischemia and to avoid coro- Finally, performance of both visual and quantitative analyses nary manipulation with wires in non-significant coronary stenosis. may appear unusually high. In this study, consecutive patients Second, the use of a 64-slice CT scanner may lead to a greater num- were referred for an ICA after SPECT based on the following clinical ber of artefacts compared to those obtained with more advanced criteria in order to minimize patient risk and reduce invasive pro- CT systems. However, the ASiR technology used here to reconstruct cedures: positive SPECT (37 patients with ischemia) or persistent CT images has proven to result in noise reduction and significantly symptomatology with negative SPECT (47 patients). The fact of impacted image quality, thereby improving the proportion of inter- excluding patients with known CAD or prior revascularization pretable myocardial segments compared with filtered backprojec- (without stent or by-pass) implies lower probability of having sig- tion (FBP) reconstruction [14]. Furthermore, several studies have nificant CAD in the population and probably better results in anal- proven the potential application of 64-slice CT scanners in the yses of myocardial ischaemia. Maybe the high threshold of lumen detection of myocardial ischaemia [7,17,22] and the number of narrowing (>70%) to determine significant coronary stenosis and artefacts identified through visual interpretation in this study the use of advanced ASiR reconstruction technique may have con- was minimal: only 3% of myocardial segments were excluded for tribute to these results. the analysis. Third, myocardial segments were associated with one of the three major coronary arteries based on the established 5. Conclusions clinical recommendations. Nonetheless, the variability in coronary circulation may result in different coupling between coronary CT myocardial perfusion analysis by visual interpretation arteries and LV segments. Fourth, a total of 33 (3.8%) myocardial proved to be effective in detecting myocardial ischaemia compared segments had ischaemia by SPECT without significant coronary to SPECT + ICA, although the use of quantitative data in terms of 8 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 TPR improves its detection. However, visual interpretation of As suggested, for each of the six short-axis views (3 at rest and 3 at myocardial perfusion on dual-energy CT-based iodine images out- stress), three reference points were selected (Fig. A1, top): the first performs visual analysis on single-energy CT, thereby being less one situated on the external contour of the ventricle wall that sep- necessary the use of quantitative methods in this case. arates the anterior and the anteroseptal (or septal, in the case of apical section) segments, the second one located on the center of the lumen and the third one situated on the external contour of Declaration of Competing Interest the ventricle wall that separates the inferior and the inferoseptal (or septal) segments. The authors declare that they have no known competing finan- The left ventricle wall segmentation was done by Region Grow- cial interests or personal relationships that could have appeared ing, an algorithm which allows the partitioning of an image into to influence the work reported in this paper. similar/homogenous areas through the application of homogene- ity/similarity criteria such as colour, intensity and/or texture, by Appendix A in-house Matlab code (MathWorks, Massachusetts, US). The mask and reference points were transformed from Cartesian into Polar A1. AHA segmentation of the left ventricle Coordinate System and the angles of reference points were used to extract myocardial AHA segments automatically (Fig. A1, The left ventricle on CTA and CTP was divided into equal thirds bottom). perpendicular to the long axis based on anatomical landmarks and following the published recommendations [15]. The basal A2. Evaluation of myocardial perfusion defects according to third was acquired and displayed from the area extending from quantitative CTP parameters: MA, TPR, and MPRI the mitral annulus to the tips of the papillary muscles, the mid- cavity view was selected from the region that includes the entire Mean attenuation density (MA) length of the papillary muscles, and the apical short-axis view For both rest- and stress CT images, mean attenuation values of was selected from the area beyond the papillary muscles to just the 16 AHA-segments were calculated at subendocardium, subepi- before the cavity ends. The slice thickness was of 0.8 cm and deter- cardium and whole-myocardium providing six mean and standard mined on the basis of modality-specific resolution. deviation (SD) values of attenuation (3 at rest and 3 at stress). The AHA-segmentation of the left ventricle was performed using subendocardium, the subepicardium and whole-myocardium of three short-axis views of the left ventricle (basal, mid and apical each AHA- segment which presented an attenuation value lower sections) both at rest and under stress by an experienced observer. than the mean–kSD, were classified as hypoperfused. Different Fig. A1. AHA segmentation of the apical (left) and basal (right) sections in a theoretically case in which the angles of the reference points 1, 2 and 3 were 45, 0 and 315 degrees, respectively, for the apical section; and 60, 0 and 300 degrees for the basal section. 9 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Fig. A2. Cut-off values determination for myocardial perfusion computed topography (CT) parameters. Youden index versus different thresholds for transmural perfusion ratio (TRP) (first row), myocardial perfusion reserve index (MPRI) (second row) and mean attenuation (MA) (third row) by conventional CT (left coloumn) and CT- based iodine imaging (right coloumn). thresholds (k) were analyzed (Fig. A2). Perfusion defects were Myocardial perfusion reserve index (MPRI) identified as the AHA-segments with hypoperfusion in at least An MPRI value was obtained per each segment, so hypoperfu- the subendocardium or in the whole myocardium at stress (is- sion was defined if that value was below the threshold. chemia) or at both rest and stress (infarction). Those segments with an MA value below the threshold at the subepicardium but A3. Cut-off values of CTP parameters not at subendocardium were categorized as artifacts. Single-energy CT and CT-based iodine imaging are quantified Transmural perfusion ratio (TPR) using different measuring units, Hounsfield Units and mg/mL of Two TPR values from both rest and stress images were obtained iodine, respectively; for this reason, two sets of thresholds were per segment. Hypoperfusion was defined if TPR value was below determined. For both native and iodine imaging, the 90 main coro- the threshold in the stress-phase (ischemia) or in both rest- and nary vessels (LAD, LCX or RCA) of the test group (30 patiens, stress-phases (infarct). randomly-selected among those having both images available) 10 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Table A1 Cut-off parameter determination in the test group. Performance results of quantitative vessel-based perfusion analysis with coronary CTA > 70%. Single-energy CT images CT-based iodine images TPR MPRI MA TPR MPRI MA Threshold 0.96 23 0.5 0.97 47 0.3 Sensitivity 100 (96–100) 100 (96–100) 100 (94–100) 100 (96–100) 92 (72–100) 100 (96–100) Specificity 99 (95–100) 97 (93–100) 97 (93–100) 92 (86–99) 94 (87–100) 92 (86–99) PPV 93 (77–100) 87 (68–100) 87 (68–100) 67 (42–91) 69 (43–95) 67 (42–91) NPV 100 (99–100) 100 (99–100) 100 (99–100) 100 (99–100) 99 (95–100) 100 (99–100) Accuracy 99 (96–100) 98 (94–100) 98 (94–100) 93 (88–99) 93 (88–99) 93 (88–99) Youden Index 0.99 (0.96–1.01) 0.97 (0.94–1.01) 0.97 (0.94–1.01) 0.92 (0.86–0.98) 0.85 (0.69–1.02) 0.92 (0.86–0.98) AUC on ROC 0.99 (0.98–1.00) 0.98 (0.97–1.01) 0.98 (0.97–1.01) 0.96 (0.93–0.99) 0.93 (0.84–1.01) 0.96 (0.93–0.99) Values are % (95% CI). Abbreviations: AUC = area under the curve; CI = confidence interval; CTA = Computed tomography angiography; CT = Computed tomography; MA = mean attenuation; MPRI = myocardial perfusion reserve index; PPV and NPV = positive and negative predictive values, respectively; ROC = receiver-operating characteristic; TPR = transmural perfusion ratio. Table A2 Per-vessel CT myocardial perfusion analysis with validation group. Performance results of visual and quantitative perfusion analysis with single-energy CT or dual-energy CT- based iodine images and coronary CTA > 50%. Single-energy CT images CT-based iodine images Visual analysis TPR MPRI MA Visual analysis TPR MPRI MA Threshold 0.96 23 0.5 0.97 47 0.3 Sensitivity 75 (56–94) 92 (79–100) 92 (79–100) 96 (86–100) 82 (66–100) 95 (84–100) 86 (70–100) 82 (63–100) Specificity 97 (94–100) 93 (88–97) 92 (87–97) 91 (86–96) 96 (91–100) 89 (82–96) 91 (85–97) 89 (82–96) PPV 82 (63–100) 69 (51–86) 67 (49–84) 66 (49–83) 82 (66–100) 66 (48–84) 68 (49–87) 62 (43–81) NPV 96 (92–99) 98 (96–100) 98 (96–100) 99 (97–100) 96 (91–100) 99 (96–100) 97 (93–100) 96 (91–100) Accuracy 92 (90–98) 93 (88–97) 92 (87–96) 92 (87–96) 93 (88–98) 90 (84–96) 90 (84–96) 87 (81–94) Youden Index 0.72 (0.55–0.90) 0.86 (0.73–0.96) 0.82 (0.72–0.96) 0.87 (0.78–0.96) 0.81 (0.65–0.96) 0.84 (0.74–0.95) 0.77 (0.62–0.93) 0.71 (0.53–0.88) AUC on ROC 0.86 (0.77–0.95) 0.93 (0.86–0.98) 0.90 (0.86–0.98) 0.94 (0.89–0.98) 0.89 (0.82–0.97) 0.92 (0.87–0.98) 0.89 (0.81–0.96) 0.85 (0.76–0.94) Values are % (95% CI). Abbreviations: AUC = area under the curve; CI = confidence interval; CTA = Computed tomography angiography; CT = Computed tomography; MA = mean attenuation; MPRI = myocardial perfusion reserve index; PPV and NPV = positive and negative predictive values, respectively; ROC = receiver-operating characteristic; TPR = transmural perfusion ratio. Table A3 Per-patient CT myocardial perfusion analysis with validation group. Performance results of visual and quantitative perfusion analysis with single-energy CT or dual-energy CT- based iodine images and coronary CTA > 50%. Single-energy CT images CT-based iodine images Visual analysis TPR MPRI MA Visual analysis TPR MPRI MA Threshold 0.96 23 0.5 0.97 47 0.3 Sensitivity 87 (66–100) 93 (77–100) 93 (77–100) 100 (97–100) 93 (75–100) 100 (96–100) 86 (64–100) 93 (76–100) Specificity 95 (86–100) 89 (78–100) 85 (72–97) 85 (72–97) 95 (79–100) 81 (64–98) 81 (64–98) 81 (64–98) PPV 87 (66–100) 78 (56–100) 70 (47–93) 71 (50–93) 87 (66–100) 74 (51–96) 71 (46–95) 72 (49–96) NPV 95 (86–100) 97 (90–100) 97 (90–100) 100 (98–100) 96 (85–100) 100 (98–100) 91 (78–100) 95 (84–100) Accuracy 93 (84–100) 91 (82–99) 87 (77–97) 89 (80–98) 92 (782–100) 87 (76–99) 82 (69–96) 85 (73–97) Youden Index 0.81 (0.63–1.00) 0.84 (0.67–0.99) 0.78 (0.61–0.95) 0.85 (0.73–0.96) 0.85 (0.67–1.02) 0.81 (0.66–0.96) 0.66 (0.43–0.90) 0.74 (0.53–0.94) AUC on ROC 0.90 (0.81–1.00) 0.92 (0.83–1.00) 0.89 (0.80–0.98) 0.92 (0.87–0.98) 0.92 (0.83–1.01) 0.90 (0.83–0.98) 0.83 (0.71–0.95) 0.87 (0.76–0.97) Values are % (95% CI). Abbreviations: AUC = area under the curve; CI = confidence interval; CTA = Computed tomography angiography; CT = Computed tomography; MA = mean attenuation; MPRI = myocardial perfusion reserve index; PPV and NPV = positive and negative predictive values, respectively; ROC = receiver-operating characteristic; TPR = transmural perfusion ratio. were classified as having coronary artery disease if they had both a iodine imaging were compared with SPECT + ICA as reference in lumen narrowing >70% by invasive coronary angiography (ICA) and terms of Youden index (Fig. A2). Finally, the best thresholds were at least one myocardial segment of the myocardial territory determined as those with the highest Youden Index (Table A1). ascribed to the narrowed coronary artery with a specific myocar- In the case of HU, the best-performing results of TPR, MPRI, dial perfusion CT parameter < threshold. For each parameter a and MA were raised with 0.96, 23 and 0.5, respectively, in accor- large range of possible cut-off values were defined taking into dance with Youden indexes of 0.99 for TPR and 0.97 for both account that minimum and maximum analyzed thresholds were MPRI and MA (Table A1). In the case of iodine, the best cut-off bounded by taking into account lowering both false negatives with values for TPR, MPRI and MA were 0.97, 47 and 0.3, respectively, too high cut-off values and false positives with too low thresholds. with Youden indexes of 0.96 for both TPR and MA, and 0.93 for The performance of each threshold on both single-energy CT and MPRI (Table A1). 11 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 Table A4 Overview of dissenting CTA/CTP results compared to the reference method (SPECT + ICA) and reasons for those errors in 4 cases: visual analysis and TPR with the use of single- energy CT or iodine images. Patient ID Vessel Type of Reason for false results on CT false result on CT Single-energy CT images CT-based iodine images Visual analysis TPR Visual analysis TPR A LAD FP Artifact Artifact FN of SPECT FN of SPECT B LAD FP Contrast distribution Contrast distribution Contrast distribution LCX FP Artifact FN of SPECT C LAD FP FN of SPECT FN of SPECT FN of SPECT RCA FP FN of SPECT FN of SPECT D LAD FP Contrast distribution Contrast distribution E RCA FN Image quality Impossible to detect visually F LAD FN Image quality Impossible to detect LCX FN Image quality RCA FN Image quality G LCX FN Abbreviations: ID = arbitrary identification; LAD = left anterior descending; LCX = left circumflex artery; RCA = right coronary artery; TPR = transmural perfusion ratio; FN = false negative; FP = false positive; SPECT = single-photon emission computed tomography. A4. Dissenting CTA/CTP results when compared to reference perfusion defects by single photon emission computed tomography: the CORE320 study, Eur. Heart J. 35 (2014) 1120–1130, https://doi.org/10.1093/ SPECT + ICA in myocardial perfusion analysis eurheartj/eht488. [3] B.S. Ko, J.D. Cameron, M. Leung, et al., Combined CT coronary angiography and As diagnostic inaccuracies between CTA/CTP and SPECT + ICA stress myocardial perfusion imaging for hemodynamically significant stenoses occurred in only seven patients, those with dissenting results are in patients with suspected coronary artery disease: a comparison with fractional flow reserve, JACC Cardiovasc Imag. 5 (2012) 1097–1111, https:// referred to as patients A to G in the following sections (Table A4 ). doi.org/10.1016/j.jcmg.2012.09.004. [4] R.T. George, A. Arbab-Zadeh, J.M. Miller, et al., Adenosine stress 64-and 256- Myocardial perfusion analysis from single-energy CT row detector computed tomography angiography and perfusion imaging a pilot study evaluating the transmural extent of perfusion abnormalities to Visual assessment in the vessel-based analysis resulted in 5 predict atherosclerosis causing myocardial ischemia, Circ. Cardiovasc. Imag. 2 false-positive and 4 false-negative vessels. The majority of false- (2009) 174–182, https://doi.org/10.1161/CIRCIMAGING.108.813766. positive results appeared in the LAD artery due to beam- [5] R.C. Cury, T.A. Magalhaes, A.T. Paladino, et al., Dipyridamol stress and rest transmural myocardial perfusion ratio evaluation by 64 detector-row hardening streaking artefacts (patient A) or to the inhomogeneous computed tomography, J. Cardiovasc. Comput. Tomog. 5 (2011) 443–448. distribution of contrast (patients B and D). On the other hand, CT [6] G. Pontone, D. Andreini, A.I. Guaricci, et al., Quantitative vs. qualitative images of patient C, who presented coronary stenosis > 70% in evaluation of static stress computed tomography perfusion to detect haemodynamically significant coronary artery disease, Eur. Heart J. the RCA and LAD, showed subendocardial hypoperfusion along Cardiovasc. Imag. 19 (2018) 1244–1252, https://doi.org/10.1093/ehjci/jey111. the basal- anteroseptal, inferoseptal and inferior segments which [7] K. Hosokawa, A. Kurata, T. Kido, et al., Transmural perfusion gradient in was not identified by SPECT. The correct association between these adenosine triphosphate stress myocardial perfusion computed tomography, Circulat. J. 75 (2011) 1905–1912, https://doi.org/10.1253/circj.CJ-10-1144. coronary stenoses and hypoperfused regions might be related to [8] B.S. Ko, J.J. Linde, A.R. Ihdayhid, et al., Non-invasive CT-derived fractional flow perfusion deficits undetected by SPECT. The 4 false-negative reserve and static rest and stress CT myocardial perfusion imaging for results were found in patients E and F owing to the limited quality detection of haemodynamically significant coronary stenosis, Int. J. of the images (Table A4 Appendices). Three patients were misdiag- Cardiovasc. Imag. 35 (2019) 2103–2112, https://doi.org/10.1007/s10554-019- 01658-x. nosed: two false positives (patients B and C) and one false negative [9] J.T. Kühl, R.T. George, V.C. Mehra, et al., Endocardial–epicardial distribution of (patient F) (Table A4 Appendices). myocardial perfusion reserve assessed by multidetector computed No false-negative vessels were detected with TPR although a tomography in symptomatic patients without significant coronary artery disease: insights from the CORE320 multicentre study, Eur. Hear J. – total of 4 false-positive vessels were identified in three patients Cardiovasc. Imag. 9 (2015), https://doi.org/10.1093/ehjci/jev206. (patient A: 1 in LAD, patient B: 1 in LAD and 1 in LCX, patient C: [10] A. So, J. Hsieh, S. Narayanan, et al., Dual-energy CT and its potential use for 1 in LAD). quantitative myocardial CT perfusion, J. Comput. Tomog. 6 (2012) 308–317, https://doi.org/10.1016/j.jcct.2012.07.002. [11] F.J. Klocke, M.G. Baird, B.H. Lorell, et al., ACC/AHA guideline ACC/AHA/ASNC Myocardial perfusion analysis from dual-energy CT-based iodine guidelines for the clinical use of cardiac radionuclide imaging — executive imaging summary, J. Am. College Cardiol. 42 (2003) 1318–1333, https://doi.org/ 10.1016/S0735-1097(03)01090-8. Visual iodine-based analysis yielded a false-positive vessel in [12] P. Lancellotti, J.L. Zamorano, G. Habib, L. Badano, The EACVI Textbook of the LAD of patient A, which may be a perfusion defect undetected Echocardiography, 2nd ed., Oxford University Press, Oxford, UK, 2017. by SPECT, as stated previously. On the other hand, 2 false-negative [13] T.A. Fuchs, J. Stehli, M. Fiechter, et al., First experience with monochromatic coronary computed tomography angiography from a 64-slice CT scanner with vessels were detected: one in the RCA of patient D, in whom a Gemstone Spectral Imaging (GSI), J. Cardiovas. Comput. Tomog. 7 (2013) 25– small subendocardial perfusion defect in the apical inferior seg- 31, https://doi.org/10.1016/j.jcct.2013.01.004. ment was detected with TPR but not by visual interpretation, and [14] J. Leipsic, T.M. Labounty, B. Heilbron, et al., Adaptive Statistical Iterative the other in the LAD of patient E (Table A4 Appendices). Reconstruction: Assessment of Image Noise and Image Quality in Coronary CT Angiography, 2010, pp. 649–654. https://doi.org/10.2214/AJR.10.4285. Using TPR, only patient B had a false positive whereas, using the [15] M.D. Cerqueira, N.J. Weissman, V. Dilsizian, A.K. Jacobs, S. Kaul, W.K. Laskey, D. visual approach, only patient D had a false negative. J. Pennell, J.A. Rumberger, T. Ryan, M.S. Verani, Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the cardiac imaging committee of References the council on clinical cardiology of the American heart association, Circulation 105 (2002) 539–542, https://doi.org/10.1067/mje.2002.123374. [16] G.L. Raff, Abidov A Chair, et al., SCCT guidelines for the interpretation and [1] V.L. Roger, A.S. Go, D.M. Lloyd-Jones, et al., Heart disease and stroke statistics- reporting of coronary computed tomographic angiography, J. Cardiovas. 2011 update: a report from the American Heart Association, Circulation 123 Comput. Tomog. 3 (2009) 122–136, https://doi.org/10.1016/j.jcct.2009.01.001. (2011), https://doi.org/10.1161/CIR.0b013e3182009701. [17] E.D. Nicol, J. Stirrup, E. Reyes, et al., Sixty-four-slice computed tomography [2] C.E. Rochitte, R.T. George, M.Y. Chen, et al., Computed tomography coronary angiography compared with myocardial perfusion scintigraphy for angiography and perfusion to assess coronary artery stenosis causing 12 A. Ruiz-Muñoz, F. Valente, L. Dux-Santoy et al. IJC Heart & Vasculature 32 (2021) 100721 the diagnosis of functionally significant coronary stenoses in patients with a [20] J.J. Linde, J.T. Kühl, J.D. Hove, et al., Transmural myocardial perfusion gradients low to intermediate likelihood of coronary artery disease, J. Nucl. Cardiol. 15 in relation to coronary artery stenoses severity assessed by cardiac (2008) 311–318, https://doi.org/10.1016/j.nuclcard.2008.02.017. multidetector computed tomography, Int. J. Cardiovasc. Imag. 31 (2014) [18] E.R. DeLong, D.M. DeLong, D.L. Clarke-Pearson, Comparing the areas under two 171–180, https://doi.org/10.1007/s10554-014-0530-9. or more correlated receiver operating characteristic curves: a nonparametric [21] S.M. Ko, M.G. Song, H.K. Chee, et al., Diagnostic performance of dual-energy CT approach, Biometrics 44 (1998) 837–845, https://doi.org/10.3847/0067-0049/ stress myocardial perfusion imaging: direct comparison with cardiovascular 227/2/20. MRI, Am. J. Roentgenol. 203 (2014) W605–W613, https://doi.org/10.2214/ [19] T.A. Magalhaes, S. Kishi, R.T. George, et al., Combining coronary angiography AJR.14.12644. and myocardial perfusion by computed tomography in the identification of [22] G.M. Feuchtner, F. Plank, C. Pena, et al., Evaluation of myocardial CT perfusion flow-limiting stenosis – the CORE320 study: a coronary CTA/myocardial CTP in patients presenting with acute chest pain to the emergency department: integrated analysis, Am. J. Cardiol. 116 (2016) 379–383, https://doi.org/ comparison with SPECT-myocardial perfusion imaging, Heart 98 (2012) 1510– 10.1016/j.amjcard.2015.04.051.Predictors. 1517, https://doi.org/10.1136/heartjnl-2012-302531. 13