- Email: [email protected]
Myocardial perfusion defects in contrast echocardiography: Spatial and temporal localisation
Ultrasound in Med. & Biol. Vol. 12, No. 7, pp. 581-586, 1986 Printed in the U.S.A.
0301-5629/86 $3.00 + .00 © 1986 Pergamon Journals Ltd.
OOriginal Contribution MYOCARDIAL ECHOCARDIOGRAPHY:
PERFUSION DEFECTS IN CONTRAST SPATIAL AND TEMPORAL LOCALISATION
VOLKMAR MISZALOK,~" THOMAS FRITZSCH~ a n d MICHAEL SCHARTL§ Universit~its-Augen-Klinik, Spandauer Datum 130, 1000 Berlin 19, West Germany Abstract--The position and size of experimental myocardial perfusion defects can be detected by using the echocardiographic contrast-medium Echoson ® together with digital image processing techniques. Video tapes showing the important sequences before, during and after cardial inflow of echo-enhancing agents are fed in total length into a digital image processing unit. The myocardial circumferences are then chosen interactively. All the frames are analysed and a synthetic picture is compiled, and shows the myocardial echo densities in a single, highly condensed form. Digital subtraction of blank values (condensed sequences before injection of contrast medium) greatly improves readability. Further concentration of the remaining information yields the following diagrams: flow/circumferential position (time summation); flow/time (position summation), which contain the desired information on homogeneity or defects of echo density against time and myocardial position.
Key Words: Contrast echocardiography, Myocardial perfusion, Flow measurement, Digital image processing. 1. INTRODUCTION
2. MATERIAL AND M E T H O D
Injection of ultrasonic contrast agents into the aortic root or directly into the coronary leads to a clear increase of echo intensity in the perfused myocardium. Perfusion defects (such as after the complete closure of a coronary artery) are evinced by the failure of this increase in intensity. Good agreement between the echographically and morphologically determined magnitudes of the defect has been demonstrated by a planimetric method (Biamino et al., 1983; Kemper et al., 1983; Tei et al., 1983; Schartl et al., 1984). In addition to this static-planimetric method, it is possible to determine the course per unit time of the contrast effect by a video densitometric technique. Conclusions about the myocardial perfusions can be drawn from the form of the intensity curves (Bommer et al., 1981, Maurer et al., 1984, Tei et al., 1984, Armstrong et al., 1982). So far, however, there is no reliable method which permits a precise spatial and temporal quantification of the contrast-medium effect. A suitable method by which the major parameters of the contrast effect are concentrated on clear diagrams by means of digital picture processing techniques, is described.
The echocardiographic examination was carried out with a Hewlett-Packard Model 77020 A electronic sector scanner. The frequency of the sound-head was 3.5 MHz. The findings were recorded on video tape. The examination was carried out in 4 Beagle dogs, one of which with a body-weight of 10.2 kg :[Combelen®/ Polamived® (1 + 9) s.c. and Rompun®/Ketavet® (1 + 3) i.m.] is used as our example in this report. With breathing spontaneously the animal was given Ethrane ® inhalation for the maintenance of uniform and general anaesthesia. A 7F catheter was introduced into the left femoral artery and advanced to the root of the aorta. This catheter was used for injection of the contrast medium. A second 7F-catheter was positioned in the root of the aorta through the left carotid artery. This catheter was advanced into the left coronary ostium and a 2F Fogarty catheter introduced through it into the ramus circumflexus for occlusion of the vessel. When the catheters were in place, the animal was put in a stable right-side position and the transducer of the ultrasonic instrument placed against the right wall of the thorax in the 4th intercostal space. The short axis of the left ventricle at the level of the papillary muscle was located and the transducer then fixed to the examination table. This position was maintained until the end of the examination. On completion of the preparations for the examination, two CM injections were administered under
~"University Eye Clinic. :~ Schering AG, Berlin and Bergkamen, Dep. Contrast Media Pharmacology. § University Medical Clinic, Dept. of Cardiology. 581
582
Ultrasound in Medicineand Biology
normal perfusion conditions at an interval of 5 min. At this point the Fogarty catheter was advanced to the ramus circumflexus and the balloon cautiously inflated. Complete closure of the vessel was verified by an injection of X-ray contrast medium via the coronary catheter.
Contrast medium The Echoson ® ultrasonic contrast medium consists of minute air bubbles adhering to saccharide microparticles. The air bubbles have an average size of 3.0 #m, 97% being smaller than 7.0 #g (Parthoscope measurement).
Computer An IPS II Image Processing System (KONTRON Bildanalyse G m b H Munich) with 8 MByte Image RAM was used. It is thus possible to store 128 frames of an image sequence. An input rate of 25 images/s allows storage of a 5 s sequence, at half the rate a 10 s sequence and so on. It has been shown that a frequency of 5 images/s is optimal for echographic analysis. It permits, on the one hand, sufficient time resolution during the rapid medium inflow and complete observation (25 s) of the wash-out phase on the other. The digital pictures have a spatial resolution of 256 × 256 points with 256 grey levels. 3. SOFTWARE PACKAGE
Myocardial contrast agent flow: MCAF 3.1 Preprocessing. To reduce the high-frequency noise of the echo raw picture, the total image sequence is first subjected to low-pass filtering. An acceptable compromise between noise-suppression and loss of contour can be accomplished by a 5 X 5 averaging lowpass. The limiting factor for this low-pass filtering is the minimal thickness of the myocardium. The filter matrix must never be chosen so large that the intensive reflex zones of the endocardium and pericardium meet. Filter matrices larger than 5 × 5 are appropriate when the myocardium is relatively thick. 3.2 Interactive input of myocardial geometry. The most reliable method of myocardial identification is that of manual input, even though it involves the most work. With the aid of a digitizer, a cursor is drawn along the centre of the myocardium in a continuous line in every image. It is important that this procedure is always begun and ended at a specific point which can be identified in all the images. This manual input procedure is assisted by the MCAF software package.
July 1986, Volume 12, Number 7 Nevertheless, this is quite a time-consuming and tedious procedure. With a certain degree of practice, 120 pictures can be processed in 6 min. 3.3 Linearisation and display of circular myocardial contour. The aim is to make the inflow and outflow of the contrast medium in the whole circumference of the myocardium visible in a single picture. We solved this information compression problem by constructing a new synthetic picture where one horizontal line contains the pictorial information of myocardial brightness extracted from one picture of the raw sequence. (A) Each frame of this sequence is presented to the operator and processed as follows: 1. The myocardium has to be traced manually (using a x/y pen). The computer checks coordinates, length and coherence of this trace and constructs one horizontal line of the new artificial picture. 2. Traces taken from frames during systolic contraction are shorter than during diastole. In order to obtain lines of constant length, all traces are mathematically normalised to the length of 512 pixels. This gives the impression of cutting open the myocardium at a fixed point and unrolling it over a fixed distance. (B) Combined presentation of all myocardial lines 1. The procedure described in A is repeated for each frame. 2. The resulting myocardial lines are assembled in the correct chronological order to form a new synthetic picture, the Condensed Position-Time Image: CPTI (see Fig. I). This image is rectangular with 512 pixels horizontally and f X c pixels vertically, f being the number of measured frames and c an integer constant representing the factor of repetition. Each pixel has a defined position: horizontal: relative position in the myocardium vertical: position in time (the top line represents the first frame) 3.4 Artefact suppression on CPTI. The CPTI can be processed like any other natural image. 3.4.1 Vertical smoothing. Since the lines are taken from different phases of cardiac activity, the CPTI has undesirable horizontal stripes. These have no informative value and are therefore superfluous. The CPTI can be smoothed by a rectangular low-pass filter (horizontally 3 pixels and vertically 2 X c pixels) (see Fig. 2). The limiting factor for this vertical filtering is the desired time resolution for the CM in-flow phase. The horizontal striping must not be completely suppressed. 3.4.2 Empty reference definition. The intensities of the
Contrast echocardiography• V. MISZALOKel
583
al.
S
B-dill--s1~s2
~s3--11~.ql.---d
E ====================================
1.inj.
f
2. inj. E
Fig. 1. Raw condensed position-time image CPTI: The dotted myocardial circumference has been drawn 120 times anticlockwise from B to E. The CPTI (photo) therefore contains 120 horizontal lines representing l0 s with 2 injection phases. Vertical stripes sl and s3 are brighter than s2 because they are perpendicular to the transducer. B: start point of unrolling; E: end point of unrolling.
echoes vary according to the position of the myocardium in relation to the transducer. Those parts of the myocardium which are perpendicular to the sound plane (normally anterior and posterior parts) give stronger echoes while parallel parts (normally left and right wall) give weak ones. With the unrolling of the myocardium in the CPTI, broad light stripes running
from top to bottom consequently appear even without any contrast medium. These stripes are n o t inhomogeneities of the myocardium but are artefacts. They can be suppressed in the following way. 1. The CPTI is first divided into two sections: (a) time before inflow of contrast-medium (upper part); (b) time from first appear-
584
Ultrasound in Medicine and Biology
July 1986, Volume 12, Number 7 tensity, OTID is then approximated in terms of the formula: I(t) = ae -(t/h) + b, where h is the fade-out half-life of the contrast effect.
Fig. 2. Noise suppression by vertical filtering makes 14 heart pulsations visible. Empty reference is computed from the upper 10%. ance of contrast medium to end of sequence (lower part). 2. A blank value line is formed from a part (a) by horizontal averaging. 3. The empty reference image (ERI) is obtained by extending this line to the size of a full image (identical repetition in the vertical direction). Consequently, it only contains the accumulated information from part (a) of the CPTI. 3.4.3 Empty reference subtraction. When the ERI is subtracted from the CPTI, a considerable number of the artefacts which result only from the recording technique, are removed. The myocardium now appears more homogeneous and the broad vertical stripes in the CPTI are reduced, making the CPTI darker. Normally, part (a) becomes completely black (apart from any traces of noise remaining). The areas of part (b) inaccessible to contrast-medium are likewise completely black. It is essentially only those parts of (b) where the contrast-medium has become echographically effective, which remain visible as bright areas. 3.5 Overall time-intensity diagram. If the brightness values of the CPTI are added and scaled line by line (horizontal averaging), the vector of the total echo activity of the myocardium in the time course is obtained. If the values of this vector are joined up, the curve of the course of the echo reflexes in the total myocardium: overall time-intensity diagram (OTID) is obtained. This curve shows the efficacy of the contrast medium, the speed of inflow and the time-course o f the outflow (see Fig. 3) superimposed on the right of the CPTI. The intensity axis (ordinate) is shown from right (intensity 0) to the left and the time axis from top (frame 1) to bottom. Starting from maximum in-
3.6 Overall position-intensity diagram. If the brightness values of the CPTI are added column by column and scaled (vertical averaging), the vector of the total echo activity along the myocardium (time summation) is obtained. When the values of this vector are joined up, the curve of the mean echo activity along the myocardium is obtained: overall position-intensity diagram (OPID) (see Fig. 3). This curve shows the homogeneity of the myocardium. In the ideal case, a homogeneously perfused myocardium would show a straight horizontal line. The absolute height of this straight line does not have any informative value since it is dependent on the number of empty frames [part (a)] of the CPTI and on the overall intensity. If the myocardium contains a perfusion defect, the OPID drops to zero at this point. The edge-sharpness of this depression and thus the accuracy of the method itself is dependent to a very much greater degree on the anatomical and technical recording factors than on the evaluation. The OPID naturally cannot contain more information than the original image-sequence itself but one can be sure that no information is lost.
Fig. 3. Same as Fig. 2 after empty reference subtraction. On the right, vertical overall time-intensity diagram (OTID); at the bottom, horizontal overall position-intensity diagram (OPID).
Contrast echocardiographyQ V. MISZALOKel The OPID (to remain in the logic of the CPTI) is superimposed on the lower edge of the CPTI. The intensity axis (ordinate) is shown as usual from the bottom (intensity 0) upwards, the position axis (relative distance from the start of the scaled circumference curve of the myocardium) from left to right. 4. EXAMPLE Figure 4 shows two phases of contrast echocardiography: A: blank phase, B: high-point of the contrast-medium effect. Closure of the Ramus circumflexus produced a large myocardial perfusion defect in B. In the CPTI raw image (Fig. 1), the time axis runs from the top downwards: 120 frames (12,5 per s = approx, l0 s) are analysed. The myocardial section (left edge of CPTI) in Fig. 4 A is at 2 o'clock, i.e. the point of attachment of the papillary muscle. 5 phases are recognizable in the CPTI reading from the top downwards: 1. blank phase (upper 10% of image); 2. main injection; 3. phase of constant intensity; 4. follow-up injection in the centre of the image; 5. slow fade-out phase. In order to suppress noise, filtering was carried out with a low-pass (3 horizontal, 10 vertical pixels [see Fig. 2)]. The two longitudinal stripes beginning at the upper edge are the expression of the strong reflex zones of the anterior myocardial wall (left) and the posterior myocardial wall (centre right). These artefacts are greatly reduced by subtraction of the empty reference
A sl
585
al.
image (ERI) formed from the average of the blank phase (see Fig. 3). The image now produced contains essentially only the effect of the contrast-medium. The overall position-intensity diagram (OPID) superimposed at the lower edge of the picture clearly shows the demarcations between the perfused and the nonperfused areas and the attainable echo-homogeneity in the perfused area. The deviations from the postulated plateau shape are residual artefacts. The nonperfused part is, however, homogeneous at the zero line, the demarcations are sharply defined. Perfusions defects extending over 10-15 degrees (360 degrees total circumference) should be identifiable. The overall time-intensity diagram (OTID) superimposed at the righthand edge of the image in Fig. 3 concentrates the phases of the process. A point of interest is the constriction artefact at the right edge of the perfused area in Phase 4 (follow-up injection): strong intensities in the vicinity of the transducer obviously erase echos and can simulate perfusion defects. 5. DISCUSSION 1. The advantages of digital image processing are clear: The strict evaluation objectivity yields reliable time-course curves. The overall effects and the regional changes in contrast (with appropriate input the endocardial and epicardial effects also) can be documented if desired. Homogeneity of perfusion can be judged. A comparison can be made of the abilities of various contrast media to detect underperfused areas. It is also to be expected that relative reductions in perfusion in the case of coronary stenoses can also be
Q
$2
Fig. 4. Two phases of contrast echocardiography:A: before contrast-mediuminflow,B: maximum contrast revealing large experimental perfusion defect. B/E: start and end point of interactive tracing; s l: artefact echo zone 1; s2: artefact echo zone 2.
586
Ultrasound in Medicine and Biology
d e m o n s t r a t e d b y c o m p a r i s o n o f the local curves before a n d after challenging ( S a k a m a k i et al., 1984). T h e v i d e o d e n s i t o m e t r y techniques used so far (De M a r i a et al., 1980; M e l t z e r et al., 1982) a p p e a r less useful d u e to their t e c h n i c a l limits, a n d the lack o f a facility for filtering a n d artefact suppression. T h e d i s a d v a n t a g e s o f digital i m a g e processing are the cost o f the c o m p u t e r s a n d the difficult p r o b l e m o f a u t o m a t i c c o n t o u r detection. Interactive d r a w i n g o f the m y o c a r d i a l c i r c u m f e r e n c e is still laborious. It takes a b o u t 3 s p e r frame, i.e. 6 m i n for 120 frames. T h e r e are several s o l u t i o n s for speeding u p the e v a l u a t i o n process: • E C G - t r i g g e r e d f r a m e input; • a p p r o x i m a t i v e m y o c a r d i u m definition by geom e t r i c a l elements: circle, ellipse; • a u t o m a t i c m y o c a r d i a l c o n t o u r detection. N o n e o f these is free o f severe artefacts. A t the present time, a n a u t o m a t i c m y o c a r d i u m definition a l g o r i t h m w o r k i n g with sufficient speed a n d a c c u r a c y o n sonographic video i n p u t with c o n t r a s t m e d i u m flow does n o t seem to be a r e a s o n a b l e possibility. 2. T h e effect o f the ultrasonic c o n t r a s t m e d i u m is based o n the n u m b e r a n d size o f t i n y a i r - b u b b l e s (micro-bubbles). T h e substance used in this s t u d y has p r o v e d to be s u p e r i o r to the substances n o r m a l l y used as far as r e p r o d u c i b i l i t y a n d h o m o g e n e i t y o f the contrast effect are concerned. (Smith et al., 1984; Feinstein et al., 1983). However, the flow b e h a v i o u r o f the bubbles within the i n t r a c a r d i a l m i c r o c i r c u l a t i o n , which is o f great i m p o r t a n c e for q u a n t i t a t i v e analysis o f the m y o c a r d i a l bloodflow, has n o t yet been clarified. So far, this q u e s t i o n has o n l y been clarified for large vessels (Levine et al., 1984), where the s a m e velocities were o b s e r v e d for m i c r o - b u b b l e s as for the r e d b l o o d cells. 3. T h e r e l a t i o n s h i p b e t w e e n reflected s o u n d energy a n d visible intensity is n o t linear. T h e r e are m a n y t e c h n i q u e s for c o n v e r t i n g m e a s u r e d values into grey tones. In a d d i t i o n , the c o n t r a s t m e d i u m echoes o f the close field w e a k e n those o f the d e e p e r structures. This is w h y no m e t h o d o f e v a l u a t i o n can s u p p l y a b s o l u t e m e a s u r e m e n t values. Despite the l i m i t a t i o n s m e n t i o n e d , the m e t h o d n o w p r o p o s e d p e r m i t s the systematic e x a m i n a t i o n o f the usefulness o f c o n t r a s t - m e d i u m e c h o c a r d i o g r a p h y for the assessment o f m y o c a r d i a l perfusion: • it concentrates pictorial i n f o r m a t i o n from video films into a single s t a t i o n a r y display, • it further c o n d e n s e s this d i s p l a y into d i a g r a m s c o n t a i n i n g all the i m p o r t a n t i n f o r m a t i o n o n general a n d local flow, • it greatly reduces noise a n d echo artefacts.
July 1986, Volume 12, Number 7 Acknowledgment--Supported in part by the Deutsche Forschungsgemeinschaft and the Dr Helmut und Margarete Meyer-Schwarting Foundation.
REFERENCES Armstrong W. F., Mueller T. M., Kinney E. L., Tickner E. G., Dillon J. C. and Feigenbaum H. (1982) Assessment of myocardial perfusion abnormalities with contrast-enhanced two-dimensional echocardiography. Circulation 66, 166-173. Biamino G., Henrichs K. J., Kruck I., Matsuka H., Schwietzer U., Winkler B,, Schroeder R. and Schaper W. (1983) Kontrast-echokardiographische Visualisation der Myokardperfusion: Korrelation zur Mikrosph/irentechnik (Abstr.) Z. Kardiol. (Supp. 2) 72, 20. Bommer W. J., Rasor J., Tickner G., Takeda P., Miller L., Lee G., Mason D. T. and De Mafia A. N. (1981) Quantitative regional myocardial perfusion scanning with contrast echocardiography (Abstr.) Am. J. Cardiol. 47, 403. De Mafia A. N., Bommer W. J., Riggs K., Dajee A., Keown M., Kwan O. L. and Mason D. T. (1980) Echocardiographic visualisation of myocardial perfusion by left heart and intracoronary injections of echo contrast agents. Circulation 62(Supp. III), 111143. Feinstein S. B., Ten Cate F. J., Zwehl W., Ong K., Maurer G., Tei C., Shah P. M., Meerbaum S. and Corday E. (1983) Two-dimensional contrast echocardiography. I. In vitro development and quantitative analysis of echo contrast agents. JACC 3, 14-20. Kemper A. J., O'Boyle J. E., Sharma S., Cohen C. A., Kloner R. A., Khufi S. F. and Parisi A. F. (1983) Hydrogen peroxide contrastenhanced two-dimensional echocardiography: Real time in vivo delineation of regional myocardial perfusion. Circulation 68, 603611. Levine R. A., Teichholz L. E., Goldman M. E., Steinmetz M. Y., Baker M. and Meltzer R. S. (1984) Microbubbles have intercardiac velocities similar to those of red blood cells. JACC 3, 28-33. Maurer G., Ong K., Haenchen R., Torres M., Tei C., Wood F., Meerbaum S., Shah P. and Corday E. (1984) Myocardial contrast two-dimensional echocardiography: comparison of contrast disappearance rates in normal and underperfused myocardium. Circulation 69, 418-429. Meltzer R. S., Roelandt J., Bastiaans O. L., Pierard L., Serruys P. W. and Lancee C. T. (1982) Videodensitometric processingof contrast two-dimensional echocardiographic data. Ultrasound in Med. & Biol. 8, 509-514. Sakamaki T., Tei C., Meerbaum S., Shimuora K., Kondo S., Fishbein M. C., Y-Rit J., Shah P. and Corday E. (1984) Verification of myocardial contrast two-dimensional echocardiographic assessment of perfusion defects in ischemic myocardium. JACC3, 3438. Schartl M., Ffitzsch T., Ffiedmann W. and Lange L. (1984a) Quantitative myocardial perfusion studies with a new safe echo contrast agent (Abstr.) JACC 3, 563. Schartl M., Ffitzsch T., Ffiedmann W. and Lange L. (1984b) Quantifizierung myokardialer Perfusiondefekte mittels zwei-dimensionaler Kontrastechokardiographie. Z. Kardiol. 73, 560-567. Smith M. D., Kwan O. L., Reiser H. J. and De Mafia A. N. (1984) Superior intensity and reproducibility of SHU-454, a new fight heart contrast agent, JACC 3, 992-998. Tei C., Sakamaki T., Shah P., Meerbaum S., Shimoura K., Kondo S. and Corday E. (1983) Myocardial contrast echo-cardiography: a reproducible technique of myocardial opacification for identifying regional perfusion deficits. Circulation 67, 585-593. Tei C., Kondo S., Meerbaum S., Ong K., Maurer G., Wood F., Sakamaki T., Shimoura K., Corday E. and Shah P. M. (1984) Correlation of myocardial echo contrast disappearance rate ("washout") and severity of experimental coronary stenosis. JACC 3, 39-46.
0301-5629/86 $3.00 + .00 © 1986 Pergamon Journals Ltd.
OOriginal Contribution MYOCARDIAL ECHOCARDIOGRAPHY:
PERFUSION DEFECTS IN CONTRAST SPATIAL AND TEMPORAL LOCALISATION
VOLKMAR MISZALOK,~" THOMAS FRITZSCH~ a n d MICHAEL SCHARTL§ Universit~its-Augen-Klinik, Spandauer Datum 130, 1000 Berlin 19, West Germany Abstract--The position and size of experimental myocardial perfusion defects can be detected by using the echocardiographic contrast-medium Echoson ® together with digital image processing techniques. Video tapes showing the important sequences before, during and after cardial inflow of echo-enhancing agents are fed in total length into a digital image processing unit. The myocardial circumferences are then chosen interactively. All the frames are analysed and a synthetic picture is compiled, and shows the myocardial echo densities in a single, highly condensed form. Digital subtraction of blank values (condensed sequences before injection of contrast medium) greatly improves readability. Further concentration of the remaining information yields the following diagrams: flow/circumferential position (time summation); flow/time (position summation), which contain the desired information on homogeneity or defects of echo density against time and myocardial position.
Key Words: Contrast echocardiography, Myocardial perfusion, Flow measurement, Digital image processing. 1. INTRODUCTION
2. MATERIAL AND M E T H O D
Injection of ultrasonic contrast agents into the aortic root or directly into the coronary leads to a clear increase of echo intensity in the perfused myocardium. Perfusion defects (such as after the complete closure of a coronary artery) are evinced by the failure of this increase in intensity. Good agreement between the echographically and morphologically determined magnitudes of the defect has been demonstrated by a planimetric method (Biamino et al., 1983; Kemper et al., 1983; Tei et al., 1983; Schartl et al., 1984). In addition to this static-planimetric method, it is possible to determine the course per unit time of the contrast effect by a video densitometric technique. Conclusions about the myocardial perfusions can be drawn from the form of the intensity curves (Bommer et al., 1981, Maurer et al., 1984, Tei et al., 1984, Armstrong et al., 1982). So far, however, there is no reliable method which permits a precise spatial and temporal quantification of the contrast-medium effect. A suitable method by which the major parameters of the contrast effect are concentrated on clear diagrams by means of digital picture processing techniques, is described.
The echocardiographic examination was carried out with a Hewlett-Packard Model 77020 A electronic sector scanner. The frequency of the sound-head was 3.5 MHz. The findings were recorded on video tape. The examination was carried out in 4 Beagle dogs, one of which with a body-weight of 10.2 kg :[Combelen®/ Polamived® (1 + 9) s.c. and Rompun®/Ketavet® (1 + 3) i.m.] is used as our example in this report. With breathing spontaneously the animal was given Ethrane ® inhalation for the maintenance of uniform and general anaesthesia. A 7F catheter was introduced into the left femoral artery and advanced to the root of the aorta. This catheter was used for injection of the contrast medium. A second 7F-catheter was positioned in the root of the aorta through the left carotid artery. This catheter was advanced into the left coronary ostium and a 2F Fogarty catheter introduced through it into the ramus circumflexus for occlusion of the vessel. When the catheters were in place, the animal was put in a stable right-side position and the transducer of the ultrasonic instrument placed against the right wall of the thorax in the 4th intercostal space. The short axis of the left ventricle at the level of the papillary muscle was located and the transducer then fixed to the examination table. This position was maintained until the end of the examination. On completion of the preparations for the examination, two CM injections were administered under
~"University Eye Clinic. :~ Schering AG, Berlin and Bergkamen, Dep. Contrast Media Pharmacology. § University Medical Clinic, Dept. of Cardiology. 581
582
Ultrasound in Medicineand Biology
normal perfusion conditions at an interval of 5 min. At this point the Fogarty catheter was advanced to the ramus circumflexus and the balloon cautiously inflated. Complete closure of the vessel was verified by an injection of X-ray contrast medium via the coronary catheter.
Contrast medium The Echoson ® ultrasonic contrast medium consists of minute air bubbles adhering to saccharide microparticles. The air bubbles have an average size of 3.0 #m, 97% being smaller than 7.0 #g (Parthoscope measurement).
Computer An IPS II Image Processing System (KONTRON Bildanalyse G m b H Munich) with 8 MByte Image RAM was used. It is thus possible to store 128 frames of an image sequence. An input rate of 25 images/s allows storage of a 5 s sequence, at half the rate a 10 s sequence and so on. It has been shown that a frequency of 5 images/s is optimal for echographic analysis. It permits, on the one hand, sufficient time resolution during the rapid medium inflow and complete observation (25 s) of the wash-out phase on the other. The digital pictures have a spatial resolution of 256 × 256 points with 256 grey levels. 3. SOFTWARE PACKAGE
Myocardial contrast agent flow: MCAF 3.1 Preprocessing. To reduce the high-frequency noise of the echo raw picture, the total image sequence is first subjected to low-pass filtering. An acceptable compromise between noise-suppression and loss of contour can be accomplished by a 5 X 5 averaging lowpass. The limiting factor for this low-pass filtering is the minimal thickness of the myocardium. The filter matrix must never be chosen so large that the intensive reflex zones of the endocardium and pericardium meet. Filter matrices larger than 5 × 5 are appropriate when the myocardium is relatively thick. 3.2 Interactive input of myocardial geometry. The most reliable method of myocardial identification is that of manual input, even though it involves the most work. With the aid of a digitizer, a cursor is drawn along the centre of the myocardium in a continuous line in every image. It is important that this procedure is always begun and ended at a specific point which can be identified in all the images. This manual input procedure is assisted by the MCAF software package.
July 1986, Volume 12, Number 7 Nevertheless, this is quite a time-consuming and tedious procedure. With a certain degree of practice, 120 pictures can be processed in 6 min. 3.3 Linearisation and display of circular myocardial contour. The aim is to make the inflow and outflow of the contrast medium in the whole circumference of the myocardium visible in a single picture. We solved this information compression problem by constructing a new synthetic picture where one horizontal line contains the pictorial information of myocardial brightness extracted from one picture of the raw sequence. (A) Each frame of this sequence is presented to the operator and processed as follows: 1. The myocardium has to be traced manually (using a x/y pen). The computer checks coordinates, length and coherence of this trace and constructs one horizontal line of the new artificial picture. 2. Traces taken from frames during systolic contraction are shorter than during diastole. In order to obtain lines of constant length, all traces are mathematically normalised to the length of 512 pixels. This gives the impression of cutting open the myocardium at a fixed point and unrolling it over a fixed distance. (B) Combined presentation of all myocardial lines 1. The procedure described in A is repeated for each frame. 2. The resulting myocardial lines are assembled in the correct chronological order to form a new synthetic picture, the Condensed Position-Time Image: CPTI (see Fig. I). This image is rectangular with 512 pixels horizontally and f X c pixels vertically, f being the number of measured frames and c an integer constant representing the factor of repetition. Each pixel has a defined position: horizontal: relative position in the myocardium vertical: position in time (the top line represents the first frame) 3.4 Artefact suppression on CPTI. The CPTI can be processed like any other natural image. 3.4.1 Vertical smoothing. Since the lines are taken from different phases of cardiac activity, the CPTI has undesirable horizontal stripes. These have no informative value and are therefore superfluous. The CPTI can be smoothed by a rectangular low-pass filter (horizontally 3 pixels and vertically 2 X c pixels) (see Fig. 2). The limiting factor for this vertical filtering is the desired time resolution for the CM in-flow phase. The horizontal striping must not be completely suppressed. 3.4.2 Empty reference definition. The intensities of the
Contrast echocardiography• V. MISZALOKel
583
al.
S
B-dill--s1~s2
~s3--11~.ql.---d
E ====================================
1.inj.
f
2. inj. E
Fig. 1. Raw condensed position-time image CPTI: The dotted myocardial circumference has been drawn 120 times anticlockwise from B to E. The CPTI (photo) therefore contains 120 horizontal lines representing l0 s with 2 injection phases. Vertical stripes sl and s3 are brighter than s2 because they are perpendicular to the transducer. B: start point of unrolling; E: end point of unrolling.
echoes vary according to the position of the myocardium in relation to the transducer. Those parts of the myocardium which are perpendicular to the sound plane (normally anterior and posterior parts) give stronger echoes while parallel parts (normally left and right wall) give weak ones. With the unrolling of the myocardium in the CPTI, broad light stripes running
from top to bottom consequently appear even without any contrast medium. These stripes are n o t inhomogeneities of the myocardium but are artefacts. They can be suppressed in the following way. 1. The CPTI is first divided into two sections: (a) time before inflow of contrast-medium (upper part); (b) time from first appear-
584
Ultrasound in Medicine and Biology
July 1986, Volume 12, Number 7 tensity, OTID is then approximated in terms of the formula: I(t) = ae -(t/h) + b, where h is the fade-out half-life of the contrast effect.
Fig. 2. Noise suppression by vertical filtering makes 14 heart pulsations visible. Empty reference is computed from the upper 10%. ance of contrast medium to end of sequence (lower part). 2. A blank value line is formed from a part (a) by horizontal averaging. 3. The empty reference image (ERI) is obtained by extending this line to the size of a full image (identical repetition in the vertical direction). Consequently, it only contains the accumulated information from part (a) of the CPTI. 3.4.3 Empty reference subtraction. When the ERI is subtracted from the CPTI, a considerable number of the artefacts which result only from the recording technique, are removed. The myocardium now appears more homogeneous and the broad vertical stripes in the CPTI are reduced, making the CPTI darker. Normally, part (a) becomes completely black (apart from any traces of noise remaining). The areas of part (b) inaccessible to contrast-medium are likewise completely black. It is essentially only those parts of (b) where the contrast-medium has become echographically effective, which remain visible as bright areas. 3.5 Overall time-intensity diagram. If the brightness values of the CPTI are added and scaled line by line (horizontal averaging), the vector of the total echo activity of the myocardium in the time course is obtained. If the values of this vector are joined up, the curve of the course of the echo reflexes in the total myocardium: overall time-intensity diagram (OTID) is obtained. This curve shows the efficacy of the contrast medium, the speed of inflow and the time-course o f the outflow (see Fig. 3) superimposed on the right of the CPTI. The intensity axis (ordinate) is shown from right (intensity 0) to the left and the time axis from top (frame 1) to bottom. Starting from maximum in-
3.6 Overall position-intensity diagram. If the brightness values of the CPTI are added column by column and scaled (vertical averaging), the vector of the total echo activity along the myocardium (time summation) is obtained. When the values of this vector are joined up, the curve of the mean echo activity along the myocardium is obtained: overall position-intensity diagram (OPID) (see Fig. 3). This curve shows the homogeneity of the myocardium. In the ideal case, a homogeneously perfused myocardium would show a straight horizontal line. The absolute height of this straight line does not have any informative value since it is dependent on the number of empty frames [part (a)] of the CPTI and on the overall intensity. If the myocardium contains a perfusion defect, the OPID drops to zero at this point. The edge-sharpness of this depression and thus the accuracy of the method itself is dependent to a very much greater degree on the anatomical and technical recording factors than on the evaluation. The OPID naturally cannot contain more information than the original image-sequence itself but one can be sure that no information is lost.
Fig. 3. Same as Fig. 2 after empty reference subtraction. On the right, vertical overall time-intensity diagram (OTID); at the bottom, horizontal overall position-intensity diagram (OPID).
Contrast echocardiographyQ V. MISZALOKel The OPID (to remain in the logic of the CPTI) is superimposed on the lower edge of the CPTI. The intensity axis (ordinate) is shown as usual from the bottom (intensity 0) upwards, the position axis (relative distance from the start of the scaled circumference curve of the myocardium) from left to right. 4. EXAMPLE Figure 4 shows two phases of contrast echocardiography: A: blank phase, B: high-point of the contrast-medium effect. Closure of the Ramus circumflexus produced a large myocardial perfusion defect in B. In the CPTI raw image (Fig. 1), the time axis runs from the top downwards: 120 frames (12,5 per s = approx, l0 s) are analysed. The myocardial section (left edge of CPTI) in Fig. 4 A is at 2 o'clock, i.e. the point of attachment of the papillary muscle. 5 phases are recognizable in the CPTI reading from the top downwards: 1. blank phase (upper 10% of image); 2. main injection; 3. phase of constant intensity; 4. follow-up injection in the centre of the image; 5. slow fade-out phase. In order to suppress noise, filtering was carried out with a low-pass (3 horizontal, 10 vertical pixels [see Fig. 2)]. The two longitudinal stripes beginning at the upper edge are the expression of the strong reflex zones of the anterior myocardial wall (left) and the posterior myocardial wall (centre right). These artefacts are greatly reduced by subtraction of the empty reference
A sl
585
al.
image (ERI) formed from the average of the blank phase (see Fig. 3). The image now produced contains essentially only the effect of the contrast-medium. The overall position-intensity diagram (OPID) superimposed at the lower edge of the picture clearly shows the demarcations between the perfused and the nonperfused areas and the attainable echo-homogeneity in the perfused area. The deviations from the postulated plateau shape are residual artefacts. The nonperfused part is, however, homogeneous at the zero line, the demarcations are sharply defined. Perfusions defects extending over 10-15 degrees (360 degrees total circumference) should be identifiable. The overall time-intensity diagram (OTID) superimposed at the righthand edge of the image in Fig. 3 concentrates the phases of the process. A point of interest is the constriction artefact at the right edge of the perfused area in Phase 4 (follow-up injection): strong intensities in the vicinity of the transducer obviously erase echos and can simulate perfusion defects. 5. DISCUSSION 1. The advantages of digital image processing are clear: The strict evaluation objectivity yields reliable time-course curves. The overall effects and the regional changes in contrast (with appropriate input the endocardial and epicardial effects also) can be documented if desired. Homogeneity of perfusion can be judged. A comparison can be made of the abilities of various contrast media to detect underperfused areas. It is also to be expected that relative reductions in perfusion in the case of coronary stenoses can also be
Q
$2
Fig. 4. Two phases of contrast echocardiography:A: before contrast-mediuminflow,B: maximum contrast revealing large experimental perfusion defect. B/E: start and end point of interactive tracing; s l: artefact echo zone 1; s2: artefact echo zone 2.
586
Ultrasound in Medicine and Biology
d e m o n s t r a t e d b y c o m p a r i s o n o f the local curves before a n d after challenging ( S a k a m a k i et al., 1984). T h e v i d e o d e n s i t o m e t r y techniques used so far (De M a r i a et al., 1980; M e l t z e r et al., 1982) a p p e a r less useful d u e to their t e c h n i c a l limits, a n d the lack o f a facility for filtering a n d artefact suppression. T h e d i s a d v a n t a g e s o f digital i m a g e processing are the cost o f the c o m p u t e r s a n d the difficult p r o b l e m o f a u t o m a t i c c o n t o u r detection. Interactive d r a w i n g o f the m y o c a r d i a l c i r c u m f e r e n c e is still laborious. It takes a b o u t 3 s p e r frame, i.e. 6 m i n for 120 frames. T h e r e are several s o l u t i o n s for speeding u p the e v a l u a t i o n process: • E C G - t r i g g e r e d f r a m e input; • a p p r o x i m a t i v e m y o c a r d i u m definition by geom e t r i c a l elements: circle, ellipse; • a u t o m a t i c m y o c a r d i a l c o n t o u r detection. N o n e o f these is free o f severe artefacts. A t the present time, a n a u t o m a t i c m y o c a r d i u m definition a l g o r i t h m w o r k i n g with sufficient speed a n d a c c u r a c y o n sonographic video i n p u t with c o n t r a s t m e d i u m flow does n o t seem to be a r e a s o n a b l e possibility. 2. T h e effect o f the ultrasonic c o n t r a s t m e d i u m is based o n the n u m b e r a n d size o f t i n y a i r - b u b b l e s (micro-bubbles). T h e substance used in this s t u d y has p r o v e d to be s u p e r i o r to the substances n o r m a l l y used as far as r e p r o d u c i b i l i t y a n d h o m o g e n e i t y o f the contrast effect are concerned. (Smith et al., 1984; Feinstein et al., 1983). However, the flow b e h a v i o u r o f the bubbles within the i n t r a c a r d i a l m i c r o c i r c u l a t i o n , which is o f great i m p o r t a n c e for q u a n t i t a t i v e analysis o f the m y o c a r d i a l bloodflow, has n o t yet been clarified. So far, this q u e s t i o n has o n l y been clarified for large vessels (Levine et al., 1984), where the s a m e velocities were o b s e r v e d for m i c r o - b u b b l e s as for the r e d b l o o d cells. 3. T h e r e l a t i o n s h i p b e t w e e n reflected s o u n d energy a n d visible intensity is n o t linear. T h e r e are m a n y t e c h n i q u e s for c o n v e r t i n g m e a s u r e d values into grey tones. In a d d i t i o n , the c o n t r a s t m e d i u m echoes o f the close field w e a k e n those o f the d e e p e r structures. This is w h y no m e t h o d o f e v a l u a t i o n can s u p p l y a b s o l u t e m e a s u r e m e n t values. Despite the l i m i t a t i o n s m e n t i o n e d , the m e t h o d n o w p r o p o s e d p e r m i t s the systematic e x a m i n a t i o n o f the usefulness o f c o n t r a s t - m e d i u m e c h o c a r d i o g r a p h y for the assessment o f m y o c a r d i a l perfusion: • it concentrates pictorial i n f o r m a t i o n from video films into a single s t a t i o n a r y display, • it further c o n d e n s e s this d i s p l a y into d i a g r a m s c o n t a i n i n g all the i m p o r t a n t i n f o r m a t i o n o n general a n d local flow, • it greatly reduces noise a n d echo artefacts.
July 1986, Volume 12, Number 7 Acknowledgment--Supported in part by the Deutsche Forschungsgemeinschaft and the Dr Helmut und Margarete Meyer-Schwarting Foundation.
REFERENCES Armstrong W. F., Mueller T. M., Kinney E. L., Tickner E. G., Dillon J. C. and Feigenbaum H. (1982) Assessment of myocardial perfusion abnormalities with contrast-enhanced two-dimensional echocardiography. Circulation 66, 166-173. Biamino G., Henrichs K. J., Kruck I., Matsuka H., Schwietzer U., Winkler B,, Schroeder R. and Schaper W. (1983) Kontrast-echokardiographische Visualisation der Myokardperfusion: Korrelation zur Mikrosph/irentechnik (Abstr.) Z. Kardiol. (Supp. 2) 72, 20. Bommer W. J., Rasor J., Tickner G., Takeda P., Miller L., Lee G., Mason D. T. and De Mafia A. N. (1981) Quantitative regional myocardial perfusion scanning with contrast echocardiography (Abstr.) Am. J. Cardiol. 47, 403. De Mafia A. N., Bommer W. J., Riggs K., Dajee A., Keown M., Kwan O. L. and Mason D. T. (1980) Echocardiographic visualisation of myocardial perfusion by left heart and intracoronary injections of echo contrast agents. Circulation 62(Supp. III), 111143. Feinstein S. B., Ten Cate F. J., Zwehl W., Ong K., Maurer G., Tei C., Shah P. M., Meerbaum S. and Corday E. (1983) Two-dimensional contrast echocardiography. I. In vitro development and quantitative analysis of echo contrast agents. JACC 3, 14-20. Kemper A. J., O'Boyle J. E., Sharma S., Cohen C. A., Kloner R. A., Khufi S. F. and Parisi A. F. (1983) Hydrogen peroxide contrastenhanced two-dimensional echocardiography: Real time in vivo delineation of regional myocardial perfusion. Circulation 68, 603611. Levine R. A., Teichholz L. E., Goldman M. E., Steinmetz M. Y., Baker M. and Meltzer R. S. (1984) Microbubbles have intercardiac velocities similar to those of red blood cells. JACC 3, 28-33. Maurer G., Ong K., Haenchen R., Torres M., Tei C., Wood F., Meerbaum S., Shah P. and Corday E. (1984) Myocardial contrast two-dimensional echocardiography: comparison of contrast disappearance rates in normal and underperfused myocardium. Circulation 69, 418-429. Meltzer R. S., Roelandt J., Bastiaans O. L., Pierard L., Serruys P. W. and Lancee C. T. (1982) Videodensitometric processingof contrast two-dimensional echocardiographic data. Ultrasound in Med. & Biol. 8, 509-514. Sakamaki T., Tei C., Meerbaum S., Shimuora K., Kondo S., Fishbein M. C., Y-Rit J., Shah P. and Corday E. (1984) Verification of myocardial contrast two-dimensional echocardiographic assessment of perfusion defects in ischemic myocardium. JACC3, 3438. Schartl M., Ffitzsch T., Ffiedmann W. and Lange L. (1984a) Quantitative myocardial perfusion studies with a new safe echo contrast agent (Abstr.) JACC 3, 563. Schartl M., Ffitzsch T., Ffiedmann W. and Lange L. (1984b) Quantifizierung myokardialer Perfusiondefekte mittels zwei-dimensionaler Kontrastechokardiographie. Z. Kardiol. 73, 560-567. Smith M. D., Kwan O. L., Reiser H. J. and De Mafia A. N. (1984) Superior intensity and reproducibility of SHU-454, a new fight heart contrast agent, JACC 3, 992-998. Tei C., Sakamaki T., Shah P., Meerbaum S., Shimoura K., Kondo S. and Corday E. (1983) Myocardial contrast echo-cardiography: a reproducible technique of myocardial opacification for identifying regional perfusion deficits. Circulation 67, 585-593. Tei C., Kondo S., Meerbaum S., Ong K., Maurer G., Wood F., Sakamaki T., Shimoura K., Corday E. and Shah P. M. (1984) Correlation of myocardial echo contrast disappearance rate ("washout") and severity of experimental coronary stenosis. JACC 3, 39-46.