|
|
| | Clear Image Window | |
Changes in systolic and diastolic distribution of myocardial tissue in five adults with coronary heart disease, were determined by gated thallium scintigraphy (GTS). Due to the risk associated with this method, healthy volunteers were not considered.
Five patients (of ages between 54 and 64 years) with suspected coronary heart disease were examined by GTS, in conjunction with an ordinary thallium scintigraphy.
Monovalent thallium ion, T1+, can substitute for potassium ions in heart muscle. This property is used with the isotope T1-201 in thallium-201 scintigraphy (for a survey, [83]). The decay of T1-201 is recorded with a gamma camera, and yields a projection of T1-201 activity in volume elements, emanating in a rectangular grid pattern from the detector. The method is used for the detection of potassium uptake deficiency.
Mono- or multigated T1-scintigraphy is known, but has been used sparingly ([125] and lit. cited therein). With this technique we recorded images for specific portions of the heart cycle, and stored them in a data base. Superimposed images from a large number of cardiac cycles for each of the portions were replayed for analysis.
Examinations were carried out with a General Electric 400 T gamma camera and a PDP 11/34 computer, provided with an image analysis system with a Gamma 11 program (Picker Int.). In order to obtain a projection image of the heart in a plane approximately perpendicular to the long axis of the heart, the camera was provided with a slanthole collimator (slit angle 30o). The camera was set up for a modified frontal projection. Slit direction corresponded to the direction of the long heart axis. The cardiac cycle was gated on the ECG R-peak with 18 frames. A deviation of up to 15 % from the predetermined length of the heart cycle was accepted. Registration time was 35 min.
A motion sequence of the human heart obtained in this way gives the impression of a squeeze-like movement of the heart. The impression is in much the same way, as the traditional view of the pumping heart. This is for the following reason; when the heart muscle contracts (and thereby thickens) the isotope level in a dissecting plane perpendicular to the detector, will increase towards the lumen of the ventricles (cf. Fig 5-1). (The dissecting plane is approximately tangential to the ventricular walls of the heart).
A computer-matched grey scale, with a constant number of grey values are to be distributed over the range zero to max intensity in each of the frames; zero always corresponds to black, and max intensity always corresponds to white. It is therefore impossible to follow variations between images by comparison of grey tones. What will be observed is merely the redistribution of grey values within each frame.
If GTS images are normalised with respect to an absolute colour scale, count density can be displayed in a uniform manner.
Subtraction was done from all frames of the grey value which by visual inspection is judged to correspond to average background and demarcation. With the help of a computer program, that resulted in a motion sequence of areas with remaining activity.
The apex and the lower lateral part of the left ventricle are now projected upon each other both in systole and diastole. The establishment of isocontours at the diaphragmal and medial part of the heart, does not allow delimitation of the outer contour of the RV. That is due to (in spite of a sampling period of 35 min) interference of background radiation from the diaphragm and statistical variation between individual frames.
The scintillation image is normally obtained by exposure during a large number of heart beats, and thus contains superimposed information from both systole and diastole. The fact that the heart is not fixed in place over such a long time period, puts a severe limitation on the use of gated thallium scintigraphy.
Movement of the diaphragm also affects the position of the heart. T1-201 scintigraphy has a number of sources of error, the most prominent one being attenuation, which varies according to the absorptivity of the tissues in the radiation path.
Quantification is uncertain as one cannot discern between signals having been produced by strong radiation sources at depth, or by weak ones near the surface.
With the heart as the object of investigation, there may be attenuation variations because of the lungs. In this case it is not even possible to ascertain, whether or not identical radiation sources at the same distance from the detector will give identical signals.
Isocontour setting and image normalisation problems could be overcome by storing image sequences tentatively identified as "pure" systolic and diastolic images. This is a way to make exposure time identical for both groups (Fig. 5-2A , Fig. 5-2B).
Fig. 5-2A
and Fig. 5-2B put together as a small
animation
This was followed by subtraction of both images from each other, without any other manipulation of data. When the diastolic image was subtracted from the systolic image, a differential image (SD) was obtained (Fig. 5-3B). Image areas not having changed their radiation density during the cardiac cycle, should have become extinguished (disregarding statistical variation). Areas with increased isotope density in the systolic image, should be positively manifest in the subtractive image. Areas with decreased isotope density in the systolic image, on the other hand, will not be identified in the SD image, as negative grey values are not permitted.
When the systolic image was subtracted from the diastolic image, another subtraction image (DS) was obtained (Fig. 5-3A). It has positive information in the areas where the SD-image has none, and vice versa.
Both subtraction images demonstrated how the form and thickness of the heart muscle changed during the cardiac cycle.
Data manipulation comprised in each individual case the determination of the maximally contracted image (end-systolic image) and of the maximally dilated image (late-diastolic image).
The R-peak triggering restarts the collection of counts; frames 17 and 18 will perhaps not receive counts in every heart cycle. Therefore those frames were consistently discarded.
This implied a decrease in sampling time for these frames, and thus an artificially low count level (to few counts in these frames).
Step-by-step manipulation was as follows:
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 |
1. [back] A supposed diastolic frame was selected (frame 16), and stored in memory 1 (Fig. 5-2A).
2. [back] A supposed systolic frame was selected (frame 6), and stored in memory 2 (Fig. 5-2B).
3. [back] A subtraction image (SD, systole minus diastole), obtained by subtracting the content of memory 1 from that of memory 2, was stored in memory 3.
4. [back] In the SD-image from memory 3, a Region Of Interest (ROI) was delimited in the area corresponding to the left ventricle, ROI-LV, preferably in the apical region. This ROI was transferred to the "total" image i.e., the image obtained by superposition of the content of all 18 frames (Fig. 5-4). The computer then presented the content of each frame with respect to ROI-LV in curve format (Fig. 5-5, memory 4).
5. [back] Memory 4 content was recalled, and the two frames with the highest number of counts were identified.
Since sampling time for each frame was comparatively long with respect to muscle movement, a new identification was made. These two images, which were alternately subtracted from each other, were recorded. The image that was thereby recognised to have the most concentrated activity distribution within the ventricular area, was chosen as the definitive systolic image. It replaced the supposed systolic image in memory 2.
6. [back] A new operation was aimed at the identification of the frame corresponding to the maximum total volume of the ventricles. Frames 16, 17, 18 and 1 represent the part of the cardiac cycle in which total ventricular volume reaches its maximum.
Frames 17 and 18 had to be discarded, as mentioned. Due to the length of sampling period, Frame 1 may encompass two stages:
In order to decide which of frames 1 and 16 represents the maximum ventricular volume stage, the respective images were compared by alternate subtraction (see above). The image showing maximum distribution of remaining ventricular activity was thereby selected as the definitive diastolic image. It replaced the supposed diastolic image in memory 1.
7. [back] Images in memory 1 and 2 were normalised with respect to an absolute grey scale ( Fig. 5-2A , Fig. 5-2B).
8. [back] Definitive diastolic and systolic images were recalled from memory 1 and 2, respectively. The systolic image was subtracted from the diastolic image, and the DS-image obtained (Fig. 5-3A) was stored in memory 5.
9. [back] Definitive systolic and diastolic images were recalled again. The diastolic image was subtracted from the systolic image, and the SD-image obtained (Fig. 5-3B) replaced the supposed SD-image in memory 3.
10. [back] The differential SD-image was recalled. It displayed changes in the position of parts of the heart muscle between systole and diastole. Parts of the ventricles that gain in muscular mass in systole are prominent. They were delimited manually by means of a joystick both from the background and from each other.
Thus an area A corresponding to a part of the left ventricle, and an area B corresponding to a part of the right ventricle were identified. They were called ROI A and B, and together with the SD-image (Fig. 5-6) stored in memory 6.
11. [back]The SD-image was recalled, and dark areas
(corresponding to lighter toned ones in the DS-image) showing loss of muscular mass in the
systolic phase, were delimited as above.
ROI F was delineated on free hand to show
the displacement zone of the right part of the AV-plane and ROI C of that of the left.
ROI D and E
were delineated on free hand in order to establish zones of outer contour regions A and B.
Furthermore, a typical background area was chosen arbitrarily and designated ROI G.
12. [back] All ROI's were superimposed on the DS-image (Fig. 5-7A) and SD-image (Fig. 5-7B).
13. [back] All frames were analysed with respect for total activity for every ROI. Results were given in diagram form (Fig. 5-8) and were stored in memory 7.
A Change in intensity over the entire cardiac cycle for one specific ROI, implies that the muscular mass pertaining to this ROI must have undergone cyclic changes. Another explanation is that the absorptivity of tissues affecting the radiation recorded could have varied.
It is evident from Fig. 5-8 that ROI A and B increase at systole, and that ROI C and F decrease. ROI D and E are rather constant over the entire cardiac cycle as is background ROI G which, however, has a significantly lower intensity.
The reciprocity of systolic increase of ROI A and B, and of systolic decrease of ROI C and F. That could imply that A and B are approaching the collimator, and that C and F are receding. This possibility is however ruled out by the constancy of ROI D and E. The intensity recorded indicates furthermore the presence of substantial muscular mass.
A better explanation is that the heart is contracting in systole mainly by the AV-plane moving towards the apex. The outer form of the heart is thus essentially unchanged ( Fig. 5-7A , Fig. 5-7B). Tissues adjacent to the AV-plane at its atrial side, together with in-flowing blood, fill up the fictive "free volume" generated behind the moving AV-plane. As these areas are rather low in T1-201 content, they will not show up in the GTS images, and thus do not interfere with the present analysis. ROI D and E represent a part of the heart wall, with muscular mass present both in systole and diastole.
The area between ROI A, B, C and F is a region of very complex anatomy. This is due to the base of the pulmonary artery and the base of the aorta joining the heart at this point. The region is being in rocking motion, coupled to the displacement of the AV-plane.
Furthermore, the volume of the right ventricle is superimposed onto the ventricular septum, which is seen in an almost cross-sectional view; for this reason, the radiation emanating from the septum should be substantially attenuated. If these interference's were not present, one might expect to find a ROI for the septal region similar to ROI C and D due to the cylindrical symmetry of the left ventricular septum.
The shapes of region C and of the zone between ROI A, B, C and F is a result of the specific projection of the heart. The AV-plane is perpendicular with respect to the long heart axis and nearly parallel to the left anterior chest wall. Besides, the camera optical axis can only be made to deviate by approximately 30o from the normal right anterior oblique direction. Therefore the movements will appear in shortened perspective, and the AV-plane movement will become distorted in the DS-image (Fig. 5-9). It is the same as outlined in connection with coronary cineangiography (Chapter 4).
In a position where the recording area of the gamma camera is parallel to the long axis of the heart, there will be a result of a DS-image and a SD-image (as presented in Fig. 5-10).
Sources of error for GTS are related to projection and to the definition of movement. Errors of projection may vary depending on the position of the heart relative to the gamma camera. The errors will have a tendency to yield values that are to low for AV-plane displacement (Fig. 5-9).
The difficulty in defining the positional extremes, in due to the low number of counts per image (cf. Fig. 5-6, authentic image of Patient no. 3).
Within the scope of GTS precision, the outer heart contour was observed to remain motionless in all patients. A further confirmation is provided by the constancy of ROI's D and E ( Fig. 5-7A , Fig. 5-7B).
It was a small number of examined patients and a possible influence of the heart disease. In spite of that, displacement of the AV-plane was calculated as a percentage of the reference distance (major axis) dLV and dRV (Fig. 5-7A , Fig. 5-7B).
The mean values of the AV-plane displacement at RV (22.8 %) and LV (20.1 %) are given in Table XI.
If the major axis length is set to 90 mm (corresponding to a normal heart), AV-plane displacement at LV will be 19 mm and at RV 22 mm. This coincides with the corresponding values 21 mm respectively 25 mm obtained from healthy subjects using echocardiography (Chapter 3), and 16 mm respectively 21 mm in coronary patients using coronary angiography (Chapter 4).
Two frames of the heart, obtained by gated thallium scintigraphy, one in diastole and one in systole, were compared. No outer contour variations of the left and right ventricle were detected.
The only detectable changes in the subtraction pictures was the displacement of the AV-plane, about 19 mm on the left side and 22 mm on the right.
Due to projection errors and the underlying heart diseases of the investigated group, these results are considered to be underestimated.
| Copyright © 1999 Inovacor AB. |
