Perfusion Analysis

Myocardial perfusion is a measure of blood flow (e.g., mL/ min) per unit mass of myocardial tissue. Myocardial perfusion should ideally match the demand for oxygen in the myocardium. Perfusion is commonly assessed at both rest and with induced stress to evaluate the capacity of the coronary circulation to increase blood flow above its baseline level and thus match increases in oxygen demand. A ratio of the perfusion parameters, measured at stress and divided by the value for rest, will give a so-called perfusion reserve. In healthy individuals, myocardial blood flow increases approximately three- to fourfold above its baseline level with maximal vasodilation; with disease, the perfusion reserve decreases, and a flow reserve on the order of 2.5:1 is often used as the cutoff for deciding whether disease is present.

The analysis of myocardial perfusion can be carried to different levels, depending on the diagnostic needs and resources available. One type of qualitative analysis associated with nuclear imaging is performed by visual comparison of the contrast enhancement in different myocardial sectors. The images are often viewed for this purpose in cine mode; delays in con-

Fig. 21. Volume-time graph with the end-diastolic and end-systolic phases identified by ES and ED symbols, respectively. The upper graph shows the variation of left ventricular (LV) volume over the cardiac cycle for a patient with congestive heart failure (CHF), and the lower graph is the same type of graph for a healthy volunteer. The patient with CHF had an enlarged ventricle (i.e., large volume) and a very low ejection fraction. Because of the low ejection fraction, the curve in the upper graph is relatively flat. Ventricular volumes were calculated by Simpson's rule from a set of short-axis images. The endocardial border had been traced on each cine frame to obtain a complete curve of left ventricular volume vs time.

Fig. 21. Volume-time graph with the end-diastolic and end-systolic phases identified by ES and ED symbols, respectively. The upper graph shows the variation of left ventricular (LV) volume over the cardiac cycle for a patient with congestive heart failure (CHF), and the lower graph is the same type of graph for a healthy volunteer. The patient with CHF had an enlarged ventricle (i.e., large volume) and a very low ejection fraction. Because of the low ejection fraction, the curve in the upper graph is relatively flat. Ventricular volumes were calculated by Simpson's rule from a set of short-axis images. The endocardial border had been traced on each cine frame to obtain a complete curve of left ventricular volume vs time.

trast enhancement or a reduced peak contrast enhancement relative to other myocardial sectors are interpreted as signatures of locally reduced myocardial blood flow. However, to do so, the absence of image artifacts is important if the analysis is purely qualitative and visual; no image postprocessing is necessary. Nevertheless, a qualitative analysis might have limited capability to detect global reductions of myocardial perfusion, especially in patients with multivessel coronary artery disease.

A quantitative analysis of MRI perfusion studies commonly starts with image segmentation, similar to the procedure for analysis of cine studies. A user segments one image with good contrast enhancement along the endocardial and epicardial

Fig. 22. Graphical user interface of software for analysis of perfusion studies. Segmentation contours are drawn by the user to define the endocardial and epicardial borders. Similar to the approach used for cine analysis, the analysis is carried out on a sector basis; in this case, 16 sectors have been defined. The drawn contours can be copied to other images for the same slice position in the perfusion study. After adjustment of the contours in each image, the software calculated the mean signal intensity in each myocardial sector. As a result, graphs are obtained depicting the change in signal intensity in each myocardial sector as a function of image number or time (see inset panel).

Fig. 22. Graphical user interface of software for analysis of perfusion studies. Segmentation contours are drawn by the user to define the endocardial and epicardial borders. Similar to the approach used for cine analysis, the analysis is carried out on a sector basis; in this case, 16 sectors have been defined. The drawn contours can be copied to other images for the same slice position in the perfusion study. After adjustment of the contours in each image, the software calculated the mean signal intensity in each myocardial sector. As a result, graphs are obtained depicting the change in signal intensity in each myocardial sector as a function of image number or time (see inset panel).

borders of the left ventricle. These contours are then either copied to the remaining images in the data set or an automatic algorithm is employed to identify the borders of the myocardium and adjust the contour positions.

The latter option is extremely useful because the number of images in a perfusion data set can be very large compared to a cine data set. The task of simply copying the contours to all other images would require extensive manual editing of the contour by the user. Unlike cine images, the myocardial boundaries can be slightly blurred in perfusion images because of the reduced spatial resolution and cardiac motion. Segmentation of myocardial perfusion images is therefore considered more challenging than for cine MR studies.

Once the myocardium is extracted by image segmentation, it is divided into smaller segments or sectors similar to those defined in cine wall motion analyses. Specifically, signal intensity averages are calculated in each myocardial segment. Typically, the myocardium is divided into four segments at the apex and up to eight sectors at the base; recently, the use of six sectors has been proposed for standardization of this task (54). The signal intensity averages can be plotted vs the image number or vs the time from the beginning of the perfusion scan. Various parameters that characterize the contrast enhancement kinetics are computed from these curves for assessing perfusion. The interface of a software tool that is used for analysis of MR perfusion studies is shown in Fig. 22.

7.1. Curve Fitting

As the perfusion images are acquired quite rapidly (<250 ms per image), there is often significant noise in the images. Thus, to extract perfusion parameters, it is useful to perform some curve fitting to smooth the signal intensity curves. One widely used choice for this purpose is the gamma variate function (55), which approximates the first-pass portion of the measured curves quite well. A gamma variate curve fitted to a signal intensity curve obtained from an MR perfusion study in a patient study is shown in Fig. 23. Nevertheless, there are certain constraints for the gamma variate analyses; for instance, it is best optimized only when the first-pass portion of the curve is used (from the foot to the peak of the curve).

A number of parameters have been proposed for a semiquantitative assessment of perfusion. Commonly used parameters are the following:

• Percentage peak enhancement: the peak signal normalized by the derived average baseline signal (i.e., signal before arrival of contrast agent expressed as a percentage).

• Upslope: the slope of the first-pass segment primarily from the start of appearance of the contrast (foot) in the myocardium to the peak.

• Time to peak: the time from the foot to the peak of the curve.

• Mean transit time: the average time required for a unit volume of blood to transit through the region of interest. It can be determined as the ratio of blood volume in the region of interest to the blood flow through the region of interest. This value can be estimated from the gamma variate fit to the tissue curve.

• Dynamic distribution volume: the area under the signal intensity curve, often normalized by the area under the corresponding curve for the left ventricle.

The upslope parameter is increasingly becoming the most widely used parameter for a semiquantitative evaluation of myocardial perfusion. The upslopes of the tissue curves are generally normalized by the upslope of the signal intensity curves for a region of interest in the center of the left ventricle, with the latter considered as an arterial input in the analysis. A ratio, defined as the normalized upslopes of the tissue curve measured for maximal vasodilation, divided by the corresponding upslope value at rest has been proposed as a perfusion reserve index (56-59). The perfusion reserve derived from the upslopes generally underestimates the actual ratio of blood flows for maximal vasodilation and rest by approx 40% (60).

We have shown that accurate myocardial blood flow estimates can be obtained by MRI methodologies in comparison to invasive studies employing radio isotope-labeled micro-spheres (61-65); the latter are acknowledged as gold standards for the measurement of blood flow in tissues. MRI perfusion imaging may therefore play a pivotal future role in assessing novel therapeutic approaches for treating coronary artery disease, and automated quantitative analyses of MR perfusion measurements would play an essential role in this task.

Your Heart and Nutrition

Your Heart and Nutrition

Prevention is better than a cure. Learn how to cherish your heart by taking the necessary means to keep it pumping healthily and steadily through your life.

Get My Free Ebook


Post a comment