Main

Morphogenesis of the cardiovascular system is likely linked to its functional development(1). This is especially true with respect to flow dynamics. Microangiographic techniques have revealed qualitative information on the flow in the cardiovascular system of the chick embryo(2–4). Quantitative techniques for the assessment of flow dynamics include blood pressure and Doppler flow measurements(1, 5). The purpose of the present study was to develop a videomicroangiographic system in which DSA and parametric imaging methods were used to simultaneously evaluate functional and anatomical data relating to the embryonic heart. Angiography combined with videodensitometry was developed as DSA in the 1980s. DSA techniques applied to each pixel of the image also have been developed as a parametric imaging method that unites anatomic information to functional information(6–10). In parametric imaging three steps are carried out to generate a new image from a series of images recorded over a time course. 1) At each pixel, a densogram is obtained which represents the dye density over a period of time. From this densogram, a parameter that reflects particular characteristics of the densogram is extracted. 2) The parameter is extracted at all the pixels in the image field. 3) The magnitude of the parameter is plotted at the pixel location in different gray scale levels according to the magnitude of the parameter. Although this requires extensive computation, with the recent development of videomicroscopy and microcomputer-based image analysis systems, application of parametric imaging in cardiac embryology to assess the anatomic and functional development of the embryonic heart has become practical.

METHODS

Angiography and acquisition of sequential images. White Leghorn eggs were obtained from a local poultry (McIntyre Farm, Lakeside, CA) and were incubated at 38 °C in a forced air incubator (Lyon Electric, Chula Vista, CA) until the embryos reached Hamburger-Hamilton(11) stage 20. The egg was laid in the incubator so that its long axis was horizontal. After transfer from the incubator to the examination table, the egg was oriented in the same fashion, and the highest portion was opened. The embryonic heart and aortic arches were studied in ovo under a fibroptic lighting source (Fiberoptic Specialties, Peabody, MA). At stage 20 the chick embryo usually lies on its left side(11). Because that was the case in all the embryos we evaluated, the right side of the embryo was studied. A glass-pulled needle (tip diameter = 7 μm) was placed into the sinus venosus using a micromanipulator MM510R(Micromanipulator-Microscope, Escondido, CA). Thereafter a bonus of 0.1 μL of 0.4% trypan blue containing 0.81% sodium chloride and 0.06% potassium phosphate at room temperature (T8154, Sigma Chemical Co., St. Louis, MO), which had been filtered through a filter whose pore size was 0.2 μm, was injected. The trypan blue solution was injected using a pressure-insensitive oil-filled tube attached to a Hamilton syringe and a microinjector (MicroLyte, SR-10, Micromanipulator-Microscope, Escondido, CA), whereas sequential images were recorded by a high speed videocamera (V-14-B)-recorder (HV-RV-200) system(Nac Inc., Tokyo, Japan) at 200 frames/s. Images were replayed frame by frame by a video cassette recorder JVC-BR-S601MV (Victor, Tokyo, Japan) and digitized by Macintosh IIci (Apple computer, Cupertino, CA) equipped with a frame-grabber board UM06869B (Data Translation, Marlbolo, MA) with a resolution of 640 × 480 pixels × 256 shades of gray. Each recording lasted less than 10 s. Planimetry and densitometric analysis was performed using the public domain National Institutes of Health Image program(written by Wayne Rasband at the U.S. National Institute of Health and available from zippy.nimh.nih.gov or on floppy disk from NTIS, 5285 Port Royal Rd., Springfield, VA 22161, part no. PB93-504868), and analytic routines for parametric imaging were coded by the authors.

Calibration of the densitometric system. Several volumes of trypan blue solution ranging in concentration from 10 to 100 μg/mL were put into containers yielding various depths of fluid layers (2.2 to 6.6 mm). Images of the containers were recorded and digitized as described above, and the gray scale value of the dye solution consisting of 20 combinations of dye concentrations and the depth of the solution were measured. The product of the dye concentration and the depth of the solution that gives the mass of the dye per unit area (dye density) being measured ranged from 0.022 to 0.66μg/mm2. It was shown that the relationship between the gray scale value of a pixel and the dye density could be fitted by the empiric mathematical function described by Rodbard: Eqn.1 The constants a,b,c, and d were determined using regression analysis. Conversion from the gray scale level to the dye density based on this function was performed pixel by pixel on all sequential images. A graticule of 2 mm on a glass slide also was recorded and used for spatial calibration purposes.

Extraction of three parameters from the densogram and its derivative at each pixel location. At each pixel location, the transit of the dye stream and the relative dye density can be expressed as a densogram (Fig. 1). The temporal derivative of the densogram at the i th frame was calculated as follows: Eqn.2 where Di denotes the pixel intensity of the ith frame (0< i < tmax). From the density curve and its derivative, three parameters were extracted as are shown in Fig. 1.

Figure 1
figure 1

Extraction of parameters from a densogram(top) and its derivative (bottom) at a single pixel location (at the dorsal aorta). The intensity level starts to raise at 25 ms and reaches its maximum at 40 ms. From these curves the following parameters were extracted: Dmax, maximal density value along the densogram (herein 0.12 μg/mm2); ΔDmax, peak derivative of the densogram (5.8 μg/mm2/s);tΔDmax (25 ms), time to peak derivative of the densogram.

Full size image

The first parameter Dmax, the value of the greatest dye density value along the densogram, was calculated as follows:Equation

The second parameter ΔDmax, the value of the greatest change in the dye density from one frame to the next independent of its time in the transit of the dye bolus, was calculated as follows:Equation where ΔDi is defined by Equation 2.

The third parameter tΔDmax was the time at which the change in dye density from one frame to the next independent of its time in the transit of the dye bolus was greatest.

Parametric imaging. Dmax,ΔDmax, and tΔDmax can be calculated at all the pixels in the image field. At each pixel location, the magnitude of each parameter was expressed as different gray scale levels. This procedure was done relative to the three parameters and resulted in the generation of three new images, namely parametric images. The first parametric image is the “MAX image,” representing the Dmax at each pixel location. The second image is the “peak derivative image,” representing ΔDmax at each pixel location. The third parametric image that we calculated was the “time to peak derivative image,” representing tΔDmax at each pixel location.

Estimation of the flow velocity. The wave front of the dye propagation is depicted by the boundary between pixels with a particular tΔDmax and neighboring pixels with incremented tΔDmax on the time to peak derivative image. This boundary represents an isochronal line. The flow velocity can be calculated by noting that propagation of the dye from one isochronal line to the next isochronal line takes exactly 5 ms. The distance between the isochronal lines along the center of the vessels (millimeters) was measured by computer-assisted planimetry and was divided by the sampling interval (5 ms) to determine the regional flow velocity.

Measurement of the vessel diameter. The diameter of the fourth aortic arch and the dorsal aorta were measured by computer-assisted planimetry using the MAX image. The edges of the vessels were determined by visual inspection. The diameter of the fourth arch was measured at the midportion between the aortic sac and the dorsal aorta, and the diameter of the dorsal aorta was measured at a point immediately distal to where the fourth arch meets the dorsal aorta.

Indirect estimation of the shear rate. The shear rate on a vessel wall is proportional to the intraluminal flow velocity and inversely proportional to the diameter of the vessel as is calculated as follows:Equation An estimation of the shear rate on the wall of the fourth aortic arch and the dorsal aorta was carried out.

Determination of heart rate, change in ventricular area, and ejection fraction. Heart rate and ejection fraction before and after injection were calculated from the real time image. The end-diastolic and end-systolic frames were determined by visually inspecting recorded real-time images frame by frame. The average frame interval between the end-diastolic phase and end-systolic phase of 10 cardiac cycles was used to determine the heart rate. As was accomplished successfully in previous studies(12–14), the border of the endocardial surface of the primitive ventricle was manually traced, and the area within the border both at end-diastolic phase and end-systolic phase was measured by computer-assisted planimetry. Change in ventricular area, which corresponds to the stroke volume, was calculated as follows: Equation

Inter- and intraobserver variability of the measurements. At two separate occasions images from 13 embryos were analyzed by two different experimenters using computer-assisted planimetry as indicated above. For each embryo, the following measurements were made: 1) the interval between the isochronal lines on the time of peak derivative image,2) the diameter of the aorta and the fourth aortic arch, and 3) end-systolic and end-diastolic area of the primitive ventricle. Results of each measurement were compared pairwise to obtain intraobserver variability and interobserver variability. Pearson's linear correlation coefficients were calculated to evaluate linear correlation.

Comparison of paired measurements. A Wilcoxon signed rank test was performed to compare the difference between the following measurements:1) flow velocity at the dorsal aorta and at the fourth aortic arch,2) shear rate on the wall of the dorsal aorta and the fourth aortic arch, 3) heart rate before and after injection of the dye,4) heart rate and ejection fraction before and after injection,5) change in ventricular area before and after injection, and 6) ejection fraction before and after injection. Statistical significance was arbitrarily defined as a p value of <0.05.

Density histogram profile across the outflow tract. Using the MAX image, histogram profiles of the outflow tract were obtained to determine the distribution of the dye across the diameter. A line perpendicular to the flow was set at the midportion of the outflow tract to construct the histogram.

RESULTS

Calibration of the densitometric system. The product of the dye concentration and the depth of the solution (dye density) was plotted on the vertical axis and the 8-bit gray scale value was plotted on the horizontal axis. The dye density increased monotonically with the value of the gray scale level. The curve fitting to Equation 1 according to Rodbard function gave: Equation The correlation coefficient was 0.9915.

Densogram and its derivative at different locations. Although 15 embryos were incubated, a successful injection was completed in 13. Using three densograms and their derivatives at three pixels along the center of the vasculature of the fourth aortic arch, proximal part of the dorsal aorta, and more distal part of the dorsal aorta propagation of the dye along the course of the flow was demonstrated (see Fig. 2). The onset of the increase in dye density was successively delayed along the blood flow.ΔDmax successively decreased along the flow, whereas tΔDmax increased successively along the flow, reflecting the progression of the wave front of the dye.

Figure 2
figure 2

Propagation of the dye along the course of the flow at three different locations along the course of the blood flow blood flow.(Left) a, fourth aortic arch; b, dorsal aorta, proximal portion; c, dorsal aorta, distal portion. The bar indicates 500 μm. On the right are shown the densogram (top) and its derivative (bottom) at the three location indicated above. Open circle, closed circle, and open box correspond to the measurements made at a, b, and c, respectively.

Full size image

Parametric images. Representative images of the MAX image, the peak derivative image, and the time to peak derivative are shown in Figures 3 and 4. The MAX image provides a good depiction of the chamber and vascular anatomy (Fig. 3). A representative image of the peak derivative image is depicted in Figure 4A, whereas the time of peak derivative image is shown in Figure 4B. In Figure 4B, the derivative of the densogram reached its maximum earlier at pixels that are of lighter gray scale level (i.e. fourth aortic arch) than the pixels of darker gray scale level (i.e. distal dorsal aorta).

Figure 3
figure 3

The MAX image depicting anatomy of the heart and the great vessels at stage 20 (right lateral view). At this stage blood flows from the sinus venosus through the atrium, primitive ventricle, the conotruncal portion, the aortic arches to the dorsal aorta. The bar indicates 500μm.

Full size image
Figure 4
figure 4

Parametric images. (A) The peak derivative image. The magnitude of the maximal value of the derivative of the densogram was plotted as a pixel value at every pixel locations. The fourth aortic arch and the proximal portion of the dorsal aorta had relatively large derivative values. (B) The time to peak derivative. The time during the cardiac cycle at which the derivative of the densogram reaches its maximum was plotted as a pixel value. The isochronal lines represent the progression of the wave front. The bar indicates 500 μm. Derivative of the densogram reached its maximum earlier at pixels which are of lighter gray scale level(i.e., fourth aortic arch) than the pixels of darker gray scale level (i.e. distal dorsal aorta).

Full size image

Inter- and intraobserver variability of the measurements. None of the following measurements revealed significant inter-/intraobserver variability: 1) measurement of the distance between the isochronal lines on the time of peak derivative image, 2) measurement of the diameter of the aorta and the fourth aortic arch, and 3) planimetry of the end-systolic and end-diastolic area of the primitive ventricle. (See Table 1 for details.)

Table 1 Observer variability
Full size table

Regional flow velocity and estimated shear rate on the wall at the fourth aortic arch and the dorsal aorta. There was a significant difference between the velocity at the dorsal aorta and at the fourth aortic arch(p = 0.02). Flow velocity at the dorsal aorta was 35.9 ± 1.9 mm/s (mean ± SEM), the flow velocity at the fourth aortic arch 45.0± 2.7 mm/s (mean ± SEM).

Similarly, there was a significant difference between the shear rate on the wall of the dorsal aorta and on that of fourth aortic arch (p = 0.0002). The diameters of the dorsal aorta and of the fourth aortic arch, measured from the MAX image, were 314 ± 12 μm (mean ± SEM) and 146 ± 6 μm (mean ± SEM), respectively. The shear rate on the wall of the dorsal aorta was 466 ± 30 s-1 (mean ± SEM), whereas that on the wall of the fourth aortic arch was 1257 ± 96 s-1 (mean ± SEM).

Comparison of heart rate, change in ventricular area, and ejection fraction before and after injection. Heart rate before and after injection was 152 ± 3/min (mean ± SEM) and 152 ± 4/min (mean± SEM), respectively (p = 0.78). Change in ventricular area before and after injection was 0.103 ± 0.01/mm2 (mean ± SEM) and 0.128 ± 0.01/mm2 (mean ± SEM), respectively(p = 0.04). Ejection fraction before and after injection was 66± 3% (mean ± SEM) and 62 ± 3% (mean ± SEM), respectively (p = 0.64).

Density histogram profile across the outflow tract. None of the dye density histogram profiles from the 13 embryos showed multimodal distribution. A representative histogram is shown in Figure 5 in the form of a surface plot.

Figure 5
figure 5

Three-dimensional projection of the MAX image. The pixel value of the MAX image, which is a two-dimensional image, is assigned as the z axis value. Histogram profile across the outflow tract is depicted as the white plane. A line perpendicular to the flow was set at the midportion of the outflow tract as indicated by the line a in the inset. Note the histogram profile across the outflow tract is unimodal and fairly symmetric.

Full size image

DISCUSSION

Parametric imaging represents a mathematical derivation of a single image from a chronologic sequence of images(6–10). The parametric imaging technique, when applied to the recording of conventional microangiography, can generate a “summary” image of the sequential events after dye injection. In this report, we presented three algorithms to generate three different parametric images of the chick embryonic cardiovascular system: the MAX image, the peak derivative image, and the time of peak derivative image. The MAX image allows a simultaneous visualization of the entire vasculature, different parts of which were maximally perfused at different times after dye injection. The peak derivative image represents the magnitude of the temporal change of the perfusion at different points of the vasculature. The rapidity of the temporal change at specific points is represented by the darkness of the peak derivative image. This allows semiquantitative comparison of the stress at different points of the vasculature. The time to peak derivative image visualizes the progression of the dye wave front which is represented by the isochronal lines. The density of the isochronal lines represents the magnitude of the regional flow in a semiquantitative way.

Subsequent computer-assisted planimetry of the intervals between the isochronal lines and planimetry of the vessel diameter enabled calculation of the regional peak flow velocity and indirect estimation of shear rate. Measurement of the peak flow velocity has not been possible using previously described angiography systems. As it relates to the embryonic vascular system, it can play a complementary role to the Doppler technique which is a reliable way of quantifying the mean flow at the dorsal aorta of the chick embryo. Determination of the mean flow is important when estimating such parameters as stroke volume and cardiac output(1, 5). In addition to measurement of the peak flow velocity, the parametric imaging system differs from the Doppler method in that it allows measurement of the flow velocity at the aortic arches which is not possible using the Doppler method because of the size of the ultrasound probe.

The validity of the parametric imaging system presented herein is based on several assumptions. First, we have assumed that the dye was mixed completely in the atrium before ejection into the outflow tract and the aortic arches. Hogers et al.(4) have recently shown that up to six independent streams originating from six vitelline veins run through the heart and selectively perfuse a subset of the aortic arches. We do not think these streaming phenomena had a significant affect on the results of the present study, because we have confirmed the following observations in each embryo. 1) The dye was injected into the sinus venosus cranial to the right cardinal vein, a point beyond which all six branches of the vitelline veins have already merged. 2) None of the dye density histogram profiles across the outflow tract from the 13 embryos showed multimodal distribution, and the overall shape of the histogram was symmetric. Based on these observations, selective perfusion of a subset of the aortic arches is most likely not a major confounding factor.

Second, we have assumed that positive/negative chronotropic and/or inotropic action was minimal in that the heart rate before and after injection did not change significantly. However, cardiac output, which is the product of the heart rate and stroke volume, is likely to have increased after infusion in that the change of ventricular area corresponds to the stroke volume. If cardiac output did increase, it is possible that the shape of the time-velocity curve might have been affected, because cardiac output also represents a product of the time integral of the time-velocity curve and the diameter of the vessels. Because this system detects the peak of the time velocity curve, the change of the shape of the time-velocity curve could exert an effect on the measurements made using the parametric imaging system. It is unlikely that the change of the time-velocity curve itself would invalidate the comparison of the calculated flow velocity at different points of the vasculature in a single embryo. However, caution needs to be raised when comparing the flow velocity values at particular points of the vasculature in different embryos.

Third, we assumed that use of an empirical equation for calibration, which is not based on any theoretical model, was valid. We chose to use a combination of light-absorbent dye with incident lighting. Under this experimental setting, we were not able to construct a theoretical equation based on any physical model. The combination of fluorescent dye and transmitted light in an explanted chick embryo(3) could allow adaptation of a calibration equation based on the Beer-Lambert law(15). However, slow response of fluorescent cameras in general would impose a serious limitation on the peak flow velocity measurement, because the interval between the isochronal lines in the time to peak derivative image is proportional to the sampling time interval if the flow velocity is kept constant. In other words, the spatial resolution of the isochronal lines is determined by the rapidity of the sampling rate.

Fourth, we have assumed that the size of the particle of the contrast material, i.e. trypan blue, is infinitely small. Particle aggregation in the trypan blue solution was not a confounding factor because of the following. 1) Particle aggregate formation was not observed at the tip of the micropipette, and 2) the size of the particle of trypan blue is unlikely to confound imaging processing because the pore size of the filter is smaller than the size of the pixel (9.2 × 9.2μm).

Relative to the implications of this study to embryonic vascular development, we have documented a 2.7-fold increase in shear rate noted on the wall of the fourth aortic arch compared with the dorsal aortic aorta. This finding is compatible with the embryologic observation that the aortic arch system undergoes extensive remodeling as opposed to the dorsal aorta, which retains its basic structure after the fusion of the paired primordium of the aortic arch. The increased shear rate on the walls of the aortic arches could mediate this process. Furthermore, the peak derivative image (Fig. 4A) revealed a strong intensity preferentially along the fourth aortic arch due to a more rapid velocity in that area. Although delivery of the dye mass is not equivalent to the delivery of the fluid mass which causes shear, it is possible that areas with large values of peak derivative tend to receive more fluid and may be exposed to higher shear rates during the cardiac cycle.

One potential application of this methodology is the functional assessment of the developing heart of transgenic mice embryos. Previously, this has been tried, measuring such parameters as ejection fraction(16, 17). The methods herein presented would allow additional measures of functional assessment of such transgenic embryos. Apart from embryonic application, the algorithms and computer programs herein presented are applicable to any videographic images with contrast enhancement,i.e. angiographic images from patients or contrast enhanced echocardiography, in pediatric research.