COMPUTER ASSISTED PULSED DIGITAL ANGIOGRAPHY
The present invention relates to an improved technique of X-ray angiography.
The radiological assessment of patients with peripheral vascular disease is currently performed using a technique called angiography. This technique requires an injection of a large quantity, typically 50-150 ml, of radio-opaque dye into an artery via a cannula, using a motorized syringe. The injection takes several seconds and its objective is to fill the artery and its branches with dye so that there is sufficient contrast between the artery and surrounding tissues for it to be visible on an X-ray. After the injection, the radio-opaque dye which acts as a contrast agent is carried with the blood into the smaller branches of the arterial tree and, ultimately, into the tissues and veins. A limited number, usually between one and ten, of single frame X-ray pictures are taken. The patient is usually moved between frames so that each X-ray picture is of a different region but, in each case, the blood vessels are well filled with the contrast agent.
The X-ray images so generated display the anatomy of the arterial tree from which an experienced clinician is able to derive a limited, subjective assessment of the functional impairment that occlusive arterial disease may be causing. Arterial disease is usually irregularly distributed and functionally significant lesions are easily missed, especially if they overlie other X-ray dense structures such as bones. Even with X-rays taken in two planes at right angles to one another, the functional severity of the arterial disease is not measurable objectively.
Digital X-ray imaging has been used to improve the quality of angiograms in terms of the anatomical information that they contain. The technique of digital subtraction angiography (DSA) uses the same dye injection method as is outlined above but records two images - one before the injection and one after. Converting the
- 1
images to a digital form and subtracting the first from the second leaves a representation of just the image of the dye bolus. The appreciation of arterial anatomy is easier with such an image as the effect of overlying tissues is removed.
There are also in existence digital vascular imaging (DVI) techniques that use a sequence of many images, generated from an X-ray image intensifier, the images are either recorded or subject to immediate image processing.
The advantage of these techniques is a reduction in the X-ray dose and the ability to subject the images to manipulation and therefore make use of some of the dynamic information contained in the sequence of images. Such techniques have been claimed to be able to give a representation of blood flow but are, in fact, unable to make objective measurements of flow.
One system which utilises digital subtraction techniques is described in United Kingdom patent application 2020945A. This patent application describes a system that collects a series of images before the dye injection and which integrates these images to form a low noise mask image. This is then subtracted from subsequent shorter sequences of images, in real-time, to provide a dynamic digital subtraction angiogram. The dye injection is timed to be visible, between the mask image and subsequent images.
The major disadvantage of images generated by an image intensifier is the high noise content of the individual images. Previously described systems integrate a number of such images to reduce the noise but in so doing lose much of the dynamic information. This integration step is therefore only confined to the first or 'mask' image and not to the subsequent images.
The important parameter that governs the severity of arterial disease is not the configuration of the individual lesions seen on an angiogram but the blood flow pattern through the arterial network. In arterial disease, the overall volume of blood flow does not diminish at rest but the distribution of the blood flow via the
possible parallel pathways does change and is a useful indicator of the functional severity of the arterial disease. Measurements of relative volume blood flow in different arteries is the most useful information for planning an appropriate surgical procedure.
Several systems have been described which compare the image in a number of successive frames of a sequence of X-ray images. In general, these techniques measure the velocity of the leading edge of a bolus of dye in the artery. Unfortunately, the velocity (cm/sec) of the blood travelling along an artery is not equivalent to the volume flow of blood. In order to measure the volume flow using the linear velocity the area of the artery must also be known. X-ray images have a relatively low spatial resolution and inherent perspective distortions and are not suitable for the necessary accurate measurements needed in order to calculate the volume flow. In addition, the linear velocity can only be measured accurately if the angle at which the artery lies relative to the X-ray beam is known. Thus, all the systems that use bolus edge motion techniques can only give a limited assessment of blood flow and cannot give an accurate objective measurement of volume blood flow.
The flow of blood in a major artery such as the aorta is not steady but shows a marked forward flow phase in the first half of the cardiac cycle followed by a period of almost zero or even slight reverse flow in the second half. During the first part of the cycle the blood is accelerating and this has the effect of stabilising the flow (laminar flow) so that injection of contrast agent during this period results in excessive dilution of the agent and inadequate mixing of the agent with the blood: both of which reduce the quality of the final angiogram image. Conventional angiography attempts to get over this problem by injecting a large volume of dye (50 to 150ml) at a rapid rate (10-15ml/sec) over several cardiac cycles (5-10 seconds). Previously described real-time digital angiogram systems use the same dye injection method as conventional angiography and differ only in the manipulation technique used on the image sequence.
In accordance with the invention, there is provided a method
and apparatus for monitoring blood flow through a blood vessel by angiography, characterised in that a bolus of radio-opaque dye is injected into a blood vessel over a period less than one cardiac cycle, and a sequence of X-ray images are formed at predetermined intervals to provide an indication of movement of the bolus through the vessel and its branches. Prefereably, the cardiac cycle is monitored and injection of the bolus is effected at a predetermined point in said cycle.
The sequence of X-ray images may be converted into digital format, the set of digital values representing each of the X-ray images being subtracted from those representing a successive X-ray image in the sequence and any negative digital values being converted to digital zeros. The density of radio-opaque dye within a predetermined notional area at least a part of which overlies the area occupied by the blood vessel in the X-ray images may then be calculated for each set of digital values generated by the said subtraction, the calculated densities providing an indication of volume flow through the blood vessel.
The proposed technique takes digital subtraction angiography further than existing techniques.
Blood flow in an artery is biphasic with an initial forward flow phase and a shorter reverse flow phase. Instead of a continuous injection of dye, a short pulse of dye is introduced at the time when the blood flow is almost zero. The dye forms a localised bolus in the artery which is carried through the arterial tree on the next pulse. Instead of two images, a continuous sequence of images is recorded, one every l/50th second, direct from the X-ray image intensifier as the dye bolus passes. These images can be digitised and stored directly in computer memory, or recorded on videotape and digitised at a later date. Each image is taken in turn and subtracted from the following one and the negative values in this image ignored. The result is an image that shows only the forward movement of the dye bolus in the given time interval. As the time interval is so short the effect of a slight movement of the patient is effectively eliminated. The set of subtraction images
can then be treated in two ways:
a) the images can be added together producing an image that shows the entire course of the dye bolus over the cycle. This is equivalent to a conventional digital subtraction angiogram.
b) the velocity of the dye bolus in each time interval is found by finding the centre of gravity of the bolus in each frame and calculating how far it has moved.
The advantages of this technique are that it uses even less dye than a conventional angiogram, provides both anatomical information and blood flow velocity over the pulse cycle. For any given point in the cardiac cycle, the change in velocity from one point in an artery to another is related to their relative areas, i.e. the degree of occlusion of the artery by arterial disease. Thus both the anatomical distribution of the disease and its functional effect can be measured. The latter is the most important factor in deciding what form of surgical treatment is required.
The computer assisted dynamic digital angiography technique described can be implemented using existing equipment. The ECG monitor, X-ray Image intensifier and motorised syringe are standard pieces of equipment in an angiography suite. A computer is fitted with a frame grabber interface board which allows it to capture images directly from the image intensifier or from the videotape, and an analogue/digital and digital/analogue interface which allows it to read the ECG signal and control the video recorder and motorised syringe driver.
A system in accordance with the invention will now be described in detail, by way of example, with reference to the drawings, in which:
Figure 1 is a schematic diagram of apparatus in accordance with the invention;
Figure 2 shows diagra matically injection of a contrast agent during a single cardiac cycle;
Figure 3 shows X-ray image density plots for two unsubtracted frames;
Figure 4 shows the effect of subtracting the X-ray image density plots of Figure 3;
Figure 5 shows schematically a dye bolus passing a pre-defined region of interest ('ROI'); and
Figure 6 shows a plot of ROI density against time (frame number) .
The apparatus used in this angiography technique is illustrated schematically in Figure 1 and includes a conventional diagnostic X-ray system (10) with an X-ray image intensifier and television camera for converting an X-ray image into a video signal comprising a series of television fields; a computer (12) having an analogue or digital memory capable of storing a plurality of successive fields and a digital processor that can manipulate any of fields in any order; and ECG monitor (14) having an analogue/digital interface to the digital processor; a motorised contrast medium injector (16) and interface to the digital processor; and an output which can be displayed on a television monitor after the image processing has been completed.
The method of injecting the radio opaque dye which acts as the contrast agent is the key to the technique. Instead of a large bolus of dye being injected over several cycles, a small bolus of contrast agent is injected during the latter part of a single cardiac cycle. As shown in Figure 2, the blood in the artery decelerates rapidly during the latter half of the cardiac cycle and this has the effect of destabilising the fluid flow and creating fluid turbulence so that injection of dye at this point will result in efficient mixing of the dye with the blood. In addition, because the dye is only injected during one cycle at a point when the blood
flow is at its least, minimum dispersion of the dye in the vessel •takes place. The result is a very localised bolus (e.g. 1-2 cm long) of high dye concentration. The synchronisation of the dye bolus injection with the cardiac cycle is achieved by using the ECG output as input to the digital processor, and an output from the digital processor to trigger the motorised dye injection syringe. The detailed timing of the signals depends on the site of the injection and the characteristics of the syringe driver.
On the next heartbeat following the dye pulse injection the blood and dye are accelerated rapidly, and as the flow is stabilised by acceleration, the bolus retains its compact configuration as it is carried into the arterial tree. As the dye is well mixed with the blood the bolus is distributed along branches of the arterial tree according to their relative volume blood flows. A sequence of X-ray images is recorded in real-time during this process. The compact nature of the bolus and the rapidity with which the blood moves makes such images difficult to interpret by an unaided observer in real-time so digital image processing is required to extract the desired anatomical and flow information.
The sequence of X-ray images must be combined and manipulated in order to give a useful representation of the overall anatomy of the arterial tree, as each image only has dye filling a small section of the arterial tree. Images from two different positions in the sequence (as shown in Figure 3) are subtracted from each other. As the bolus is in a different position in the two images the resulting image consists of a region of positive density and a region of negative density corresponding to the forward movement of the head and tail of the bolus (Fig 4). All structures which are present in the same position in both images are absent from the subtracted image. The negative density regions are set to zero to leave just a positive image of the movement of the dye. This process is repeated for different pairs of images in the sequence; the result is a set of positive contrast images that include the whole of the arterial tree. Finally these images are integrated to build a complete map of the arterial tree and to reduce the background noise in the final image.
This composite anatomical image is equivalent to the image provided in a conventional digital subtraction angiogram (DSA).
Several techniques can be used to measure flow: the most limited are those that detect the edges of the dye bolus and measure the distance it moves over a number of frames. This information is neither accurate due to image distortion and the unknown angle between vessel and X-ray beam, nor sufficient to calculate the desired volume flow.
The technique we have adopted is to measure the movement of the dye bolus past an arbitrary point in a vessel: the method calculates the centre of gravity of the dye bolus by using densitometric measurement over a small, static region of interest ('ROI') in a contiguous sequence of subtracted frames (see Figure 5). The ROI is a small notional rectangle that spans the vessel in question: the orientation of the rectangle to the vessel is not critical, nor is it necessary that the rectangle should exactly match the diameter of the vessel. Over the sequence of images, the entire dye bolus passes through the ROI, and by integrating the X-ray density over the ROI for each image, and summing the value for each image in the sequence a plot of the ROI intensity against time can be generated (Figure 6). The area under this graph is a measure of the volume of dye (and hence blood) that has passed through that part of the artery in a defined time. Such measurements are repeated for different vessls of interest. The ratio of the resulting areas for each vessel is in proportion to the ratio of the time averaged volume flow in the vessels. Thus if the process is applied to a major vessel and its branches, the relative distribution of flow between each of the branches can be calculated.
As this technique relies on X-ray density measurements and not on distance measurements it is not affected by image distortion and/or the angle of the artery vessel to the X-ray beam it is therefore far more accurate than velocity measurements. The deliberate use of a well localised, high concentration contrast bolus enhances this method of relative volume flow measurement.
The combination of this relative mean volume flow measurement technique with instantaneous bolus edge velocity estimates at various points along an artery provides a measure of the relative area of one part on an artery to another. This method of calculation is more accurate than densitometric measurements made directly from the subtracted images, especially for small arteries.
The combined use of synchronised single cycle pulsed dye injection and digital image processing of a sequence of real-time digitised X-ray images makes it possible to extract both anatomical and haemodynamic information from otherwise conventional angiography. The use of a digital processing system to co-ordinate the image generation and processing allows this information to be extracted from the image sequence and the results displayed in a suitable composite format during the radiological procedure.