Contraction wave in axial direction in free wall of guinea pig left ventricle

Borut Kirn, Vito Starc


Mechanical activation of the normal left ventricle (LV) is not simultaneous; however, the potential consequences of the ejection function of the ventricle are not entirely known. We studied contraction of the LV free wall to determine whether it reveals a contraction wave in the axial direction during ejection. Seven guinea pig hearts in situ were studied via thoracotomy. In each heart, the ventricular and aortic pressures were measured by two microtipped manometers (2-Fr, Millar). Contraction of the LV free wall was assessed with a video system (Dalsa D6-0256 camera and EPIX PIXCI D32 frame grabber; acquisition rate, 500 frames/s), and 15–18 epicardial markers were used to divide the region into 20–25 triangular areas. The area sizes were studied during contraction to locate the position of the contraction wave. For each triangular area, two variables were determined as follows: the time (tc) from the end of diastole until the size of the area reached 80% of maximum size reduction (normalized with the duration of systole) and the normalized latitude (Lax) of the area (determined at the end of diastole). A relationship between these two variables was determined by regression analysis. We found that the tc at which the contraction wave reached a triangular area was in positive correlation with the Lax value for that triangular area with a slope of 0.25 ± 0.09 and a linear correlation coefficient of 0.41 ± 0.08. Thus contraction in the guinea pig LV free wall occurs progressively from apex to base with successive areas reaching 80% contraction.

  • latitude
  • triangular area
  • epicardial marker
  • apex
  • base

the mechanical activation of the normal left ventricle (LV) is not simultaneous (3, 4, 1315, 19, 20). It is known that within the circumferential plane, mechanical activation starts at the septum and progresses as a contraction wave toward the free wall (19). Regarding the axial direction, however, the contraction wave has not yet been measured. A potential contraction wave along the axial direction is important because it influences ejection efficiency. By spreading from apex to base during the ejection as well, the contraction wave would propel the blood toward the ventricular outflow tracts.

In the past, LV contraction was frequently assessed by mapping strains from different locations of the LV wall (2, 3, 69, 10, 12, 16, 17). In those studies, patterns of circumferential and axial strains were presented, but no data on contraction waves along the axial direction were provided.

Circumferential strains are more relevant than axial strains for studying the ejection function of the ventricle. However, only both strains together provide a full picture of ventricular function. Therefore, as an alternative to circumferential and axial strains, the sizes of small areas on the ventricular surface could be studied during contraction.

In this study, we mapped the contraction of a normally functioning guinea pig LV free wall using a video technique and then analyzed the contraction of small areas on the ventricular surface during ejection in search of a contraction wave along the LV axial direction from the apex to the base.


Animal Preparation

Seven male guinea pigs (422 ± 16 g body wt, 15 ± 1 wk old) were initially anesthetized with 20% urethane (1,350 mg/kg ip). The chest was opened, and the epicardial surface was exposed by thoracotomy. Ventilation was kept constant with a positive-pressure ventilator (Harvard, Rodent Ventilator). Two microtipped manometers (2-Fr, Millar) were used to measure the ventricular and aortic pressures. The first manometer was inserted into the LV chamber through the apex, and the second was positioned just behind the aortic valve, which was accessed via cannulation of the right carotid artery. The pressure signals and the electrocardiogram (using limb leads) data were fed into a computer using a National Instruments analog-to-digital card and LabView 4.1 software with an acquisition rate of 500 samples/s. After the experiment, the animal was killed by administration of cardioplegic solution. The experiments were approved by the Veterinary Administration of the Republic of Slovenia.

Mapping LV Free Wall Contractions

Contraction of the LV free wall was assessed using a high-speed digital video-acquisition system and surface markers. A similar technique of measurement was previously used on dogs (5, 16); however, we further improved it for higher heart rates and application on miniature samples.

The video system included one monochrome digital camera (Dalsa D6–0256) and a frame grabber (EPIX PIXCI D32). The camera resolution was 256 × 256 pixels with 0–255 gray levels. The camera was equipped with a 50-mm Canon lens. Images were continuously fed into the computer's memory at a rate of 500 frames/s. PXIPL Image Processing C Library version 2.5 (EPIX) was used for image recording and analysis.

After data collection, images were analyzed to detect all markers and to determine the trace of their centers throughout the image series. In the first image, each marker (∼6 × 6 pixels) was found using the PXIPL function pxip8_bloblist with an intensity threshold of 10. In each of the subsequent images, we searched for the marker position in the immediate neighborhood of its position in the previous image. Near the position from the previous image, an area with a square border (side length, 20 pixels) was determined. The new position of the marker was then calculated as a mass center of the square. The pixel brightness was used as an equivalent for the mass. To achieve higher precision, calculation of the mass center was repeated on the same image. The size of the second square area was reduced to 15 pixels, and its center overlapped the mass center of the first square area. The mass center position of the second square area was used as a marker position. Repeating the described process throughout the image series enabled us to reliably trace the position of each marker. The marker brightness on the images ranged between 10 and 40, and the pixel brightness in the background area was <8. The combination of markers, illumination technique, and camera were brought to the very limits of the available technical solutions. The intensity of the markers' glow coupled with the sensitivity of the camera and short shutter times did not enable us to use the entire range of pixel gray levels. To avoid artifacts of the background, the square area was treated with the contrast-enhance function pxip8_pixcontrastperc from the PXIPL library before the mass center was calculated. With this function, pixels within histogram percentile levels of 85–100% were stretched into the range of 0–255 gray levels. Owing to the application of specific markers and illumination, all markers within the image were found, and no backscatter was detected.

Markers were acrylate crystals, which glow under a low-intensity ultraviolet A light bulb (OSRAM L 18W/73; wavelength, 350–400 nm). Markers were approximate squares with side length of ∼0.15 mm or ∼1/50 of LV width and were obtained from thin sheets of dry fluo spray paint. The intensity of the incident light on the epicardial surface was ∼70 W/m2.

Markers were adhered to the epicardium with a two-part mixture of fibrin glue (Beriplast, Centeon). One part was thinly spread over the marker area on the epicardium, and markers were bathed in the other part before being positioned onto the epicardium. Therefore, two parts of the fibrin glue mixed where the markers touched the epicardium. By using this type of glue, we were able to attach very small markers and avoid constraining the myocardium with a “blob” of cyanoacrylate.

On the epicardium between the anterior and lateral regions of the LV free wall, 15–18 markers were adhered at mutual distances of ∼1 mm in random order. The area extended 3–4 mm in the circumferential direction and 4–5 mm in the axial direction or ∼1/2 of the total LV width and length. The center of the area was in the equatorial region. The camera was positioned above the center of the area of each marker at a distance of ∼30 cm.

The orientation of the LV principal axes was assessed with two markers that were positioned in the visually estimated axial direction.

Measurement Protocol

In each animal, several sessions of a run and a pause (60–120 s) were repeated. During each run, at least one complete cardiac cycle was recorded. By the beginning of the run, a trigger signal was generated. This signal enabled us to synchronize data from the computer that recorded the electrocardiogram and pressure values and another computer on which the video-acquisition system was installed. During the run, ventilation was held in expiration.

Data Processing


In the cardiac cycle, three events were determined: the end of the diastolic phase, the beginning of the ejection phase, and the end of the systolic phase. Moments when the ventricular and aortic pressure values exceeded 10 kPa/s were defined as the end of diastole and the beginning of ejection, respectively. The end of systole was defined as the peak negative derivative of ventricular pressure. Heart rate was determined from the electrocardiogram.

Determination of contraction wave.

The LV free wall was divided into small triangular areas with markers in all corners (Fig. 1). The size of the triangular area was determined and studied during contraction of the ventricle to locate the position of the wave front. For each triangle, two variables were determined: the time at which the wave front reached the triangular area (tc), which was represented by the time interval from the end of diastole until the size of the triangular area achieved 80% of maximum reduction (80% contraction); and the distance from the apex, which was represented by the latitude (Lax) determined at the end of diastole. Both variables were normalized: tc so that at tc = 0, it coincided with the end of diastole, and at tc = 1, it corresponded with the end of systole; and Lax so that at Lax = 0, it coincided with a marker closest to the apex, and at Lax = 1, it corresponded with the marker most distant from the apex (Fig. 1). The decision to use 80% contraction as a criterion for contraction wave was based on the speculation that by the end of the contraction period, the wave front, if it existed, would be entirely developed. At the same time, 80% contraction was not too close to the end of the contraction period (when it would be difficult to accurately determine wave front position due to flatness of the contraction curves).

Fig. 1.

Triangular areas with normalized latitudes (Lax) of their centers (left). Lax was determined at the end of diastole and normalized so that at Lax = 0 it coincided with a marker closest to the apex and at Lax = 1 it corresponded with the marker most distant from the apex. Shape and position of the ventricle are approximate.

Next, we searched for the dependence of tc on Lax. A linear regression was performed, and the slope and correlation coefficient (r2) were determined. In the case of simultaneous contraction of the LV free wall along the axial direction, all triangular areas should achieve 80% contraction simultaneously and the slope would be zero. However, in the case of a contraction wave in the direction from apex to base, the areas closer to the apex would achieve 80% contraction earlier than those closer to the base, which would yield a positive slope. Each slope of linear regression was tested with a null hypothesis that slope = 0 and an alternative hypothesis that slope >0. A P value <0.05 was considered significant. Values are presented as means ± SD.

Error Evaluation

In the past, ultrasound crystals were frequently used for continuous measurements of displacement on canine hearts. However, for small animal hearts, the video technique is more appropriate, because it has less impact on the quantities measured and greater precision. Nevertheless, ultrasound crystals are the only commercially available technique for continuous measurements of small displacements. For this purpose, we used them to calibrate and determine resolution of the displacement measurement of the video system. We measured the size of a phantom (soft biological tissue) simultaneously with both techniques. The phantom was equipped with a pair of ultrasound crystals (Sonometrics; resolution, 24 μm; crystal diameter, 1 mm), and one marker was attached to the top of each crystal. During measurement, the phantom was manually stretched and released to enable calibration and to test the system in dynamic conditions. The resolution of the video system was determined as a standard deviation of calibrated distance between markers at a constant phantom size. We measured it to be ±1.6 μm, which is >10 times more precise than ultrasound crystals.

Application of a single-plane video for assessing three-dimensional stretch of the epicardium introduces error in the determination of marker coordinates. This type of error influences the absolute-distance values between markers (16, 18). However, because we compared relative sizes of triangles, the absolute values of the marker coordinates were not as important as long as the angle between the triangle plane and the camera axis remained constant. Hence, a potential limitation was rolling of the heart. We therefore took care to keep the heart as still as possible. Our observation was that rolling of the heart was mostly along its long axis. This may have reduced the correlation coefficient (r2). In contrast, we observed little if any rolling along the short axis, which if present would have considerably influenced our results.


There were 4–18 runs repeated on each animal. The average heart rate of the animals during the experiments was 276 ± 38 beats/min and the average systolic pressure was 87.9 ± 10.2 mmHg.

In Fig. 2, triangle sizes during systole are presented for a typical cardiac cycle. For the same cardiac cycle, the relationship between tc and Lax is presented in Fig. 3. The results of linear regression analyses are presented in Table 1 including number of runs, systolic pressure values, and heart rates for each animal.

Fig. 2.

Time course of the size of triangular areas during one period of systole. Note prestretching of some areas before the beginning of ejection (BE) as a result of a nonsimultaneous beginning of contraction. Vertical lines 1–7 are explained in Fig. 4. ED, end of diastole; ES, end of systole.

Fig. 3.

Relationship between the delay of the wavefront (tc) and Lax for one cardiac cycle (black dots, dark line). Areas closer to the apex achieve 80% contraction earlier than areas closer to the base. Relationship between tc and Lax was obtained from repeated runs on the same subject (animal 7; light lines).

View this table:
Table 1.

Average parameter values for each animal

We found that tc and Lax were positively correlated with an average correlation coefficient of 0.41 ± 0.08 and an average slope of 0.25 ± 0.09. The contraction of the triangles occurred progressively from the apex toward the base with successive areas of LV free wall reaching 80% contraction. The triangles near the apex achieved 80% contraction at 33 ± 7% of systole and 25 ± 9% of systole later, 80% contraction was achieved by triangles closest to the base.

In Fig. 4, data from the same cardiac cycle as shown in Figs. 2 and 3 are presented as animation of the LV free-wall contraction. At seven consecutive moments during contraction (for timing, see Fig. 2), the triangle size is presented with respect to its Lax. For each moment, a quadratic polynomial was interpolated. From Figs. 2 and 4 (curve 2), we see that before the ejection, triangular areas closer to the base stretch, while triangular areas closer to the apex shrink. The areas closer to the apex are in the lead of the contraction compared with the rest of the LV free wall throughout the contraction.

Fig. 4.

Sizes of triangular areas presented with respect to Lax at seven consecutive moments (which correspond to vertical lines 1–7 in Fig. 2) during contraction. A quadratic polynomial was interpolated into the relationship at each moment. Note prestretching in the basal region (line 2).


We found that during ejection, the contraction spreads over the LV free wall as a contraction wave in the direction from apex to base. This type of contraction pattern does not interfere with any of the known patterns of ventricle mechanics such as radial symmetric contraction, ellipticalization, torsion, and contraction wave in circumferential plane (1, 11, 19).

The contraction wave in the axial direction could be beneficial for the LV, because it accelerates blood toward the exit from the ventricular chamber. In this way, a drift of blood toward the apex at any time during ejection is prevented. The apex region is always in the lead of the contraction over the equatorial region despite the potential regional differences in myocardial performance. This type of ventricle mechanics could also be interesting for clinical application; namely, this is a precisely synchronized action and therefore probably sensitive to any disorder in ventricular function. It could be expected that this type of contraction synchronization would be the first to fail. Besides, for an efficient resynchronization of the ventricle contraction by pacing, not only the pattern of mechanical activation is important but also formation of the axial contraction wave.

Some of the published results are consistent with our findings. In the work of Wyman et al. (20), in Fig. 4 (line A), four consecutive images of three-dimensional LV volume-strain renderings for a right atrium-paced canine heart are shown during contraction. In the subsequent figures, the apex regions shrink first, whereas regions close to the base first stretch and later shrink. The same can be seen in Figs. 2 and 4.

In the past, ventricular contraction was often studied with respect to circumferential and axial strains. Undoubtedly, analyses of patterns of these strains would be valuable for further understanding and evaluation of this phenomenon. However, the triangle-area size method that we used reduced the amount of information by considering both the circumferential and the axial strains united in area size, which helps to clarify the description of this phenomenon.

Observation of the contraction wave in the axial direction was limited in our study to the free wall of the LV. By the time the free wall contracts, the septum is already firm, because the septum contracts before the free wall (19). Therefore, the contraction wave influences the distribution of blood in the chamber regardless of whether it can be found in other sides of the LV wall.

The contraction wave in the axial direction of the LV free wall was found in hearts of guinea pigs. This type of contraction synchronization might have an important influence on the ventricular ejection function. It should be studied further to determine whether it is initiated by a nonsimultaneous mechanical activation or formed by regional contractility properties of myocardium.


This study was supported by Ministry of Education, Science and Sport, Slovenia, Grant PO-510-381.


The authors thank Vesna Metalan for excellent technical support.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract