A modular architecture for organizing, processing and sharing neurophysiology data.

We describe an architecture for organizing, integrating and sharing neurophysiology data within a single laboratory or across a group of collaborators. It comprises a database linking data files to metadata and electronic laboratory notes; a module collecting data from multiple laboratories into one location; a protocol for searching and sharing data and a module for automatic analyses that populates a website. These modules can be used together or individually, by single laboratories or worldwide collaborations.

Interactive data exploration websites for large-scale electrophysiology.

Methodological advances in neuroscience have enabled the collection of massive datasets which demand innovative approaches for scientific communication. Existing platforms for data storage lack intuitive tools for data exploration, limiting our ability to interact effectively with these brain-wide datasets. We introduce two public websites: (Data and Atlas) developed for the International Brain Laboratory which provide access to millions of behavioral trials and hundreds of thousands of individual neurons. These interfaces allow users to discover both the raw and processed brain-wide data released by the IBL at the scale of the whole brain, individual sessions, trials, and neurons. By hosting these data interfaces as websites they are available cross-platform with no installation. By releasing each site's code as a modular open-source framework, other researchers can easily develop their own web interfaces and explore their own data. As neuroscience datasets continue to expand, customizable web interfaces offer a glimpse into a future of streamlined data exploration and act as blueprints for future tools.

Compression strategies for large-scale electrophysiology data.

Objective.With the rapid adoption of high-density electrode arrays for recording neural activity, electrophysiology data volumes within labs and across the field are growing at unprecedented rates. For example, a one-hour recording with a 384-channel Neuropixels probe generates over 80 GB of raw data. These large data volumes carry a high cost, especially if researchers plan to store and analyze their data in the cloud. Thus, there is a pressing need for strategies that can reduce the data footprint of each experiment.Approach.Here, we establish a set of benchmarks for comparing the performance of various compression algorithms on experimental and simulated recordings from Neuropixels 1.0 (NP1) and 2.0 (NP2) probes.Main results.For lossless compression, audio codecs (FLACandWavPack) achieve compression ratios (CRs) 6% higher for NP1 and 10% higher for NP2 than the best general-purpose codecs, at the expense of decompression speed. For lossy compression, theWavPackalgorithm in 'hybrid mode' increases the CR from 3.59 to 7.08 for NP1 and from 2.27 to 7.04 for NP2 (compressed file size of ∼14% for both types of probes), without adverse effects on spike sorting accuracy or spike waveforms.Significance.Along with the tools we have developed to make compression easier to deploy, these results should encourage all electrophysiologists to apply compression as part of their standard analysis workflows.

Holmium-166 Transarterial Radioembolization for the Treatment of Intrahepatic Cholangiocarcinoma: A Case Series.

BACKGROUND: Transarterial radioembolization (TARE) is used to treat primary and secondary malignancies in the liver that are not amenable to curative resection. Accumulating evidence demonstrates the efficacy and safety of TARE with yttrium-90 (90Y), which is the most widely used radionuclide for TARE, and later with holmium-166 (166Ho) for various indications. However, the safety and efficacy of 166Ho TARE in patients with intrahepatic cholangiocarcinoma (ICC) remains to be studied. METHODS: This was a retrospective case series study of seven consecutive patients with ICC who were treated with 166-Ho-TARE in our center. We recorded the clinical parameters and outcomes of the TARE procedures, the tumor response according to mRECIST, subsequent treatments, and adverse events. RESULTS: Three out of the seven patients had a partial or complete response. Two patients had stable disease after the first TARE procedure, and two of the patients (one with a complete response, and one with stable disease) were alive at the time of analysis. No serious adverse events related to the procedure were recorded. CONCLUSIONS: This is the first case series reporting the safety and tumor response outcomes of 166Ho-TARE for ICC. The treatment demonstrated its versatility, allowing for reaching a high tumor dose, which is important for improving tumor response and treating patients in a palliative setting, where safety and the preservation of quality of life are paramount.

Bypassing spike sorting: Density-based decoding using spike localization from dense multielectrode probes.

Neural decoding and its applications to brain computer interfaces (BCI) are essential for understanding the association between neural activity and behavior. A prerequisite for many decoding approaches is spike sorting, the assignment of action potentials (spikes) to individual neurons. Current spike sorting algorithms, however, can be inaccurate and do not properly model uncertainty of spike assignments, therefore discarding information that could potentially improve decoding performance. Recent advances in high-density probes (e.g., Neuropixels) and computational methods now allow for extracting a rich set of spike features from unsorted data; these features can in turn be used to directly decode behavioral correlates. To this end, we propose a spike sorting-free decoding method that directly models the distribution of extracted spike features using a mixture of Gaussians (MoG) encoding the uncertainty of spike assignments, without aiming to solve the spike clustering problem explicitly. We allow the mixing proportion of the MoG to change over time in response to the behavior and develop variational inference methods to fit the resulting model and to perform decoding. We benchmark our method with an extensive suite of recordings from different animals and probe geometries, demonstrating that our proposed decoder can consistently outperform current methods based on thresholding (i.e. multi-unit activity) and spike sorting. Open source code is available at https://github.com/yzhang511/density_decoding.

Comparison of the prognostic value of dipyridamole and dobutamine myocardial perfusion scintigraphy in hemodialysis patients.

Screening for coronary artery disease (CAD) in hemodialysis patients is hampered by contraindications and/or limitations of the available techniques in this population. Myocardial perfusion scintigraphy (MPS) using dipyridamole has been considered inaccurate due to abnormally high basal levels of adenosine in uremia that could blunt the vasodilatory response. Since dobutamine may be more reliable, we directly compared the two in patients on hemodialysis. We performed MPS at rest and after separate dipyridamole or dobutamine stress in 121 chronic hemodialysis patients. More numerous, larger, and more intense reversible lesions were induced with dobutamine than with dipyridamole, mainly in the anteroseptal segments. Reversibility with dipyridamole but not dobutamine MPS was independently and strongly related with mortality associated with CAD and with fatal and non-fatal CAD. We hypothesize that the chronotropic action of dobutamine induced alterations of wall motion, leading to spurious perfusion defects, not unlike artifacts seen with left bundle branch block. Our study shows that even though dobutamine induced more pronounced myocardial ischemia than dipyridamole in chronic hemodialysis patients, dipyridamole MPS more accurately identifies patients at high risk for subsequent cardiac death or non-fatal CAD than dobutamine.

Maximum oxygen uptake at peak exercise in elderly patients with coronary artery disease and preserved left ventricular function: the role of inflammation on top of tissue Doppler-derived systolic and diastolic function.

BACKGROUND: Several studies have shown that longitudinal systolic function and left ventricular filling pressures, as assessed with tissue Doppler imaging, predict exercise capacity. AIM: The aim of this study was to evaluate whether natriuretic peptides and inflammatory parameters can independently predict maximum oxygen uptake at peak exercise (VO2max) on top of tissue Doppler imaging-derived markers. METHODS: We evaluated 142 patients (age 70 +/- 6 years, 77% men) with known or suspected coronary artery disease and a preserved left ventricular ejection fraction (> or = 50%). All patients underwent bicycle spiroergometry, and N-terminal pro-B-type natriuretic peptide levels were determined. Cytokines (IL-6 and soluble tumor necrosis factor receptors 1 and 2) and high-sensitivity C-reactive protein were measured as inflammatory markers. Tissue Doppler imaging was applied to evaluate peak long axis systolic velocities (Sm) and early mitral annulus velocities (E'). Ratio of early transmitral flow (E) to E' was assessed as marker of left ventricular filling. Analysis of variance, comparing VO2max quartiles, was used to determine univariate predictors and linear regression to determine multivariate VO2max predictors. RESULTS: Average VO2max was 18.5 +/- 5.7 mL/kg per minute (range 6-36.6). Compared with the highest quartile, patients with low VO2max were more frequently women (P < .0001). N-terminal pro-B-type natriuretic peptide and cytokine levels were significantly higher in the lower VO2max categories. Longitudinal myocardial velocities increased, and E/E' decreased along with increasing VO2max. In multivariate linear regression analysis, VO2max was independently predicted by sex, glucose, Sm, E/E', and cytokine levels. CONCLUSION: Maximum oxygen uptake at peak exercise in patients with known or suspected coronary artery disease and preserved systolic function was independently predicted by inflammatory makers on top of tissue Doppler-derived systolic and diastolic function.

Fasting blood glucose levels are related to exercise capacity in patients with coronary artery disease.

BACKGROUND AND AIM: Previous studies have demonstrated reduced exercise capacity in patients with diabetes mellitus. This study evaluated the relationship between fasting blood glucose (FBG) levels and exercise capacity in patients with coronary artery disease (CAD). METHODS: We evaluated 986 consecutive patients with CAD referred for bicycle spiroergometry combined with gated myocardial perfusion imaging. Maximum oxygen consumption (VO2max) and maximal watts were measured. Patients were divided into 4 FBG categories: < 100 mg/dL (n = 611), 100 to 109 mg/dL (n = 144), 110 to 125 mg/dL (n = 102), and > or = 126 mg/dL (n = 129). Differences in clinical characteristics, exercise hemodynamics, perfusion imaging, and univariate as well as multivariate predictors of exercise capacity were determined. RESULTS: Maximal watts and VO2max were significantly lower (P < .0001) in patients with higher FBG levels and were related to FBG values in univariate and multivariate analyses. Left ventricular volumes and ejection fractions did not differ between the FBG categories. Myocardial perfusion imaging showed a comparable degree of ischemia in the 4 FBG groups. However, patients with higher FBG levels had higher heart rate and blood pressure values at rest resulting in a higher rate-pressure product (values in the 4 FBG groups 8299 +/- 2051, 8733 +/- 2008, 9558 +/- 2583, and 9588 +/- 2468 beat/min x mm Hg, P < .0001), suggesting increased myocardial oxygen consumption per unit time at rest. CONCLUSION: Exercise capacity in patients with CAD is related to FBG levels. Patients with impaired fasting glucose or an FBG level > or = 126 mg/dL reached lower peak watts and lower VO2max values. This could be attributed to a higher myocardial oxygen consumption per unit time at rest and the inability to adapt their coronary flow adequately to higher metabolic demands during maximal exercise.

Relationship between QRS duration, left ventricular volumes and prevalence of nonviability in patients with coronary artery disease and severe left ventricular dysfunction.

BACKGROUND: Patients with coronary artery disease (CAD), a QRS duration >or=120 ms and left ventricular ejection fraction (LVEF) or=120 ms was present in 91 patients (69%). Although there were no differences in LVEF, patients with longer QRS durations had significant larger end-diastolic and end-systolic volumes (p<0.01). Substantial nonviable tissue in the inferior or lateral wall was present in 29% of patients with a QRS duration >or=120 ms versus 7% of those with a QRS duration <120 ms (p<0.01). CONCLUSIONS: An increased QRS duration is associated with more advanced remodeling in patients with CAD and poor LV function. Almost one third of these patients with a prolonged QRS duration have no viable tissue in the inferolateral wall, an area that is usually stimulated with CRT.

Comparison among tomographic radionuclide ventriculography algorithms for computing left and right ventricular normal limits.

BACKGROUND: Various algorithms have been developed to compute right ventricular (RV) and left ventricular (LV) end-diastolic volumes, end-systolic volumes, and ejection fractions (EF) from tomographic radionuclide ventriculography (TRV). The aims of this investigation were to establish sex-specific normal limits, to determine whether different algorithms produce the same normal values, and to compare TRV normal limits vs for magnetic resonance imaging values in the literature. METHODS: Fifty-one healthy volunteers (29 men, 22 women) were studied prospectively. All subjects had normal electrocardiograms and echocardiographic examinations, and underwent both planar radionuclide ventriculography and TRV. Four algorithms were used to process TRV data. RESULTS: Normal limits for most functional parameters differed significantly from one algorithm to another. Volumes were greater in men, but no statistically significant differences were found between men and women for LV EF or RV EF values for any method. Normal LV and RV EF and volumes were largely consistent with the literature for cardiac magnetic resonance imaging. CONCLUSIONS: Ventricular measurements differ significantly among TRV algorithms. Therefore, it is important to apply sex-specific normal limits that are specific to a given TRV algorithm in interpreting LV and RV EF and volume measurements for each patient.

Myocardial perfusion imaging in the elderly: a review.

Coronary artery disease is a major cause of morbidity and mortality in the elderly population. As a result of ageing of the population and better medical, interventional and surgical treatment of patients with coronary artery disease, more and more elderly patients are referred to the cardiology department for diagnostic work-up. Stress testing, in combination with myocardial perfusion imaging, is routinely used in elderly patients, a population in which the diagnosis of significant coronary artery disease is often challenging because of atypical symptomatology. Since the introduction of technetium-99m ligands for myocardial perfusion imaging, it is possible to perform electrocardiogram-gated perfusion imaging. This not only improves the specificity of the test for coronary artery disease detection, but also enables the simultaneous assessment of left ventricular functional parameters. This article briefly overviews the possible stress modalities, diagnostic accuracy and prognostic value of myocardial perfusion imaging in elderly patients.

Diabetes and impaired fasting glucose as predictors of morbidity and mortality in male coronary artery disease patients with reduced left ventricular function.

OBJECTIVES: To evaluate the prognostic value of impaired fasting glucose and diabetes mellitus in male patients with coronary artery disease and poor left ventricular function. METHODS AND RESULTS: From a prospective database on patients referred for gated myocardial perfusion imaging between 1998 and 2002 all male patients with a history of coronary artery disease and poor left ventricular function were selected. Poor function was defined as left ventricular ejection fraction < or = 40%. Subjects were classified as non-diabetics with fasting blood glucose levels < 110 mg/dL, non-diabetics with impaired fasting glucose (fasting blood glucose between 110 and 125 mg/dL) and diabetics. Median follow-up was 2.7years. End points were all-cause mortality, cardiac death and hospitalization for heart failure. One hundred and sixty patients were selected (age 65 +/- 9 years and left ventricular ejection fraction 29 +/- 8%). In univariate analysis atrial fibrillation, NYHA class, glycaemia and diabetes mellitus discriminated between survivors and non-survivors. In Cox multivariate regression analysis for all-cause mortality only NYHA class and diabetes mellitus remained significant. Kaplan Meier analysis showed that diabetics had the worst survival and non-diabetics with glucose < 110 mg/dL had the best survival. Non-diabetics with impaired fasting glucose had intermediate survival. Analysis for cardiac death/hospitalization for heart failure showed similar results. CONCLUSION: In male patients with coronary artery disease and impaired left ventricular function diabetes mellitus and fasting glucose are strongly predictive of poor outcome. Diabetics have the worst prognosis but non-diabetics with impaired fasting glucose also are at higher risk compared to nondiabetics with low fasting blood glucose.

Accuracy of 4 different algorithms for the analysis of tomographic radionuclide ventriculography using a physical, dynamic 4-chamber cardiac phantom.

UNLABELLED: Various automatic algorithms are now being developed to calculate left ventricular (LV) and right ventricular (RV) ejection fraction from tomographic radionuclide ventriculography. We tested the performance of 4 of these algorithms in estimating LV and RV volume and ejection fraction using a dynamic 4-chamber cardiac phantom. METHODS: We developed a realistic physical, dynamic 4-chamber cardiac phantom and acquired 25 tomographic radionuclide ventriculography images within a wide range of end-diastolic volumes, end-systolic volumes, and stroke volumes. We assessed the ability of 4 algorithms (QBS, QUBE, 4D-MSPECT, and BP-SPECT) to calculate LV and RV volume and ejection fraction. RESULTS: For the left ventricle, the correlations between reference and estimated volumes (0.93, 0.93, 0.96, and 0.93 for QBS, QUBE, 4D-MSPECT, and BP-SPECT, respectively; all with P < 0.001) and ejection fractions (0.90, 0.93, 0.88, and 0.92, respectively; all with P < 0.001) were good, although all algorithms underestimated the volumes (mean difference [+/-2 SDs] from Bland-Altman analysis: -39.83 +/- 43.12 mL, -33.39 +/- 38.12 mL, -33.29 +/- 40.70 mL, and -16.61 +/- 39.64 mL, respectively). The underestimation by QBS, QUBE, and 4D-MSPECT was greater for higher volumes. QBS, QUBE, and BP-SPECT could also be tested for the right ventricle. Correlations were good for the volumes (0.93, 0.95, and 0.97 for QBS, QUBE, and BP-SPECT, respectively; all with P < 0.001). In terms of absolute volume estimation, the mean differences (+/-2 SDs) from Bland-Altman analysis were -41.28 +/- 43.66 mL, 11.13 +/- 49.26 mL, and -13.11 +/- 28.20 mL, respectively. Calculation of RV ejection fraction correlated well with true values (0.84, 0.92, and 0.94, respectively; all with P < 0.001), although an overestimation was seen for higher ejection fractions. CONCLUSION: Calculation of LV and RV ejection fraction based on tomographic radionuclide ventriculography was accurate for all tested algorithms. All algorithms underestimated LV volume; estimation of RV volume seemed more difficult, with different results for each algorithm. The more irregular shape and inclusion of a relatively hypokinetic RV outflow tract in the right ventricle seemed to cause the greater difficulty with delineation of the right ventricle, compared with the left ventricle.

Agreement between four available algorithms to evaluate global systolic left and right ventricular function from tomographic radionuclide ventriculography and comparison with planar imaging.

BACKGROUND AND AIM: Left and right ventricular ejection fractions (LVEF and RVEF) and end-diastolic and end-systolic volumes (LVEDV, RVEDV, LVESV and RVESV) can be calculated from tomographic radionuclide ventriculography (TRV). The aim of this study was to validate and compare these parameters obtained using four different TRV software programs (QBS, QUBE, 4D-MSPECT and BP-SPECT). METHODS: LVEF obtained from planar radionuclide ventriculography (PRV) was compared with LVEF obtained from TRV using the four different software programs in 166 patients. Furthermore, ventricular volumes obtained using TRV (QBS, QUBE and 4D-MSPECT) were compared with those obtained using BP-SPECT, the latter being the only method with the validation of ventricular volumes in the literature. RESULTS: The correlation of LVEF between PRV and TRV was good for all methods: 0.81 for QBS, 0.79 for QUBE, 0.71 for 4D-MSPECT and 0.79 for BP-SPECT. The mean differences+/-standard deviation (SD) were 3.16+/-9.88, 10.72+/-10.92, 3.43+/-11.79 and 2.91+/-10.39, respectively. The correlation of RVEF between BP-SPECT and QUBE and QBS was poor: 0.33 and 0.38, respectively. LV volumes calculated using QBS, QUBE and 4D-MSPECT correlated well with those obtained using BP-SPECT (0.98, 0.90 and 0.98, respectively), with mean differences+/-SD of 7.31+/-42.94, -22.09+/-36.07 and -40.55+/-39.36, respectively. RV volumes showed poorer correlation between QBS and BP-SPECT and between QUBE and BP-SPECT (0.82 and 0.57, respectively). CONCLUSION: LVEF calculated using TRV correlates well with that calculated using PRV, but is not interchangeable with the value obtained using PRV. Volume calculations (for left and right ventricle) and RVEF require further validation before they can be used in clinical practice.

The role of nuclear medicine in the prediction and detection of radiation-associated normal pulmonary and cardiac damage.

Dose-effect calculations used in the planning of modern radiotherapy (RT) involving normal lung or cardiac tissue rely on structural imaging techniques, such as CT, as the basis for measuring and predicting dose-response. However, more accurate methods for predicting dose-response may result if information on the locoregional functional status of the irradiated organ(s) is included in the computational model. For RT cases that involve delivering dose to the lung and heart, this may be achieved by the assessment of tomographic scintigraphies of lung perfusion (Q) and ventilation (V) and scintigraphic imaging of myocardial perfusion and function, respectively.

Validation of gated blood-pool SPECT cardiac measurements tested using a biventricular dynamic physical phantom.

UNLABELLED: We have developed a biventricular dynamic physical cardiac phantom to test gated blood-pool (GBP) SPECT image-processing algorithms. Such phantoms provide absolute values against which to assess accuracy of both right and left computed ventricular volume and ejection fraction (EF) measurements. METHODS: Two silicon-rubber chambers driven by 2 piston pumps simulated crescent-shaped right ventricles wrapped partway around ellopsoid left ventricles. Twenty experiments were performed at Ghent University, for which right and left ventricular true volume and EF ranges were 65-275 mL and 55-165 mL and 7%-49% and 12%-69%, respectively. Resulting 64 x 64 simulated GBP SPECT images acquired at 16 frames per R-R interval were sent to Columbia University, where 2 observers analyzed images independently of each other, without knowledge of true values. Algorithms automatically segmented right ventricular activity volumetrically from left ventricular activity. Automated valve planes, midventricular planes, and segmentation regions were presented to observers, who accepted these choices or modified them as necessary. One observer repeated measurements >1 mo later without reference to previous determinations. RESULTS: Linear correlation coefficients (r) of the mean of the 3 GBP SPECT observations versus true values for right and left ventricles were 0.80 and 0.94 for EF and 0.94 and 0.95 for volumes, respectively. Correlations for right and left ventricles were 0.97 and 0.97 for EF and 0.96 and 0.89 for volumes, respectively, for interobserver agreement and 0.97 and 0.98 for EF and 0.96 and 0.90 for volumes, respectively, for intraobserver agreement. No trends were detected, though volumes and right ventricular EFs were significantly higher than true values. CONCLUSION: Overall, GBP SPECT measurements correlated strongly with true values. The phantom evaluated shows considerable promise for helping to guide algorithm developments for improved GBP SPECT accuracy.

Accuracy of commercially available processing algorithms for planar radionuclide ventriculography using data for a dynamic left ventricular phantom.

BACKGROUND: Automatic and semi-automatic algorithms to calculate ejection fraction (EF) from planar radionuclide ventriculography (PRV) have been used for many years in nuclear medicine. Validation of these algorithms is scarce and often performed on outdated versions of the software. Nevertheless, clinical trials where PRV is being used as the 'gold standard' for EF are numerous. Because of the importance attributed to the EF calculated by these programs, the accuracy of the resulting EF was assessed with a dynamic left ventricular physical phantom. METHODS: A dynamic left ventricular phantom was used to simulate 21 combinations of various ejection fractions (7-66%) and end diastolic volumes (27-290 ml). For each combination, a planar radionuclide ventriculograph was acquired, converted to an interfile format and transferred into processing stations with 10 different contemporaneously available commercial algorithms. The gold standard was the 'real' EF of the phantom, derived from the exact volume of the ventricle in end diastolic and end systolic position. Correlation and Bland-Altman analysis was performed between the real EF and the calculated EF. RESULTS: The correlation for all data was excellent (r=0.98), the mean difference was very acceptable (0.98%). Nevertheless, Bland-Altman analysis showed a significant trend in the difference between real and calculated EF, with a growing underestimation for higher ranges of EF, due to an overestimation of background in larger volumes compared to smaller ones. CONCLUSION: The determination of EF from PRV, calculated with commercially available algorithms, correlates closely to the real EF of a dynamic left ventricular phantom. This phantom can be used in the development and validation of algorithms for PRV studies, in software audits and in quality assurance procedures.

Fractional flow reserve and myocardial perfusion imaging in patients with angiographic multivessel coronary artery disease.

OBJECTIVES: The aim of this study was to investigate the correlation between myocardial ischemia detected by myocardial perfusion imaging (MPI) with single-photon emission computed tomography with intracoronary pressure-derived fractional flow reserve (FFR) in patients with multivessel coronary disease at angiography. BACKGROUND: Myocardial perfusion imaging can underestimate the number of ischemic territories in patients with multivessel disease. However, there are limited data comparing MPI and FFR, a highly accurate functional index of myocardial ischemia, in multivessel coronary disease. METHODS: Sixty-seven patients (201 vascular territories) with angiographic 2- or 3-vessel coronary disease were prospectively scheduled to undergo within 2 weeks MPI (rest/stress adenosine) and FFR in each vessel. RESULTS: In 42% of patients, MPI and FFR detected identical ischemic territories (mean number of territories 0.9 +/- 0.8 for both; p = 1.00). In the remaining 36% MPI underestimated (mean number of territories; MPI: 0.46 +/- 0.6, FFR: 2.0 +/- 0.6; p < 0.001) and in 22% overestimated (mean number of territories; MPI: 1.9 +/- 0.8, FFR: 0.5 +/- 0.8; p < 0.001) the number of ischemic territories in comparison with FFR. There was poor concordance between the ability of the 2 methods to detect myocardial ischemia on both a per-patient (kappa = 0.14 [95% confidence interval: -0.10 to 0.39]) and per-vessel (kappa = 0.28 [95% confidence interval: 0.15 to 0.42]) basis. CONCLUSIONS: Myocardial perfusion imaging with single-photon emission computed tomography has poor concordance with FFR and tends to underestimate or overestimate the functional importance of coronary stenosis seen at angiography in comparison with FFR in patients with multivessel disease. These findings might have important consequences in using MPI to determine the optimal revascularization strategy in patients with multivessel coronary disease.