This section is created by Rahul Kumar, MD; Patna, Bihar, India. CEO at G S Neuroscience Clinic and Research Centre Private Limited
What is Catheter Cone Beam CT Angiography (CBCTA) ?
This term is used colloquially in the Angio suite but let us take a step back and try to understand the origin of this term. For this, we must first understand the working principle of a conventional CT scanner and then try to apply the same knowledge to the Flat Panel Detector (FPD), and in the process try to understand the differences.
Conventional CT Scan Physics and technique
Depending on the generation of CT scanner being used, the physics and the hardware are configured differently. In the oldest clinically available models, the X-Ray source used to be a point source and it used to shoot a single beam through the object that used to be picked up by a single small detector, and then the whole construct used to move around the object and keep on repeating the process (Fig 1)
(Fig 1 – Conventional 1st generation CT Scanner – the X-Ray source is seen on the left and the detector is seen on the right of the panel. The source emits a single beam that is captured by the detector, then the process continues to acquire parallel slices. Once the entire object has been X-rayed, then the entire construct moves by an angle of the arc and the whole process is repeated till the entire circumference is acquired.)
As it will be very obvious from the images shown here in Fig 1 (a poor attempt to demonstrate something quite so complex !!), the entire workflow of the first-generation scanner is very slow and cumbersome. This was the only thing available in the late 70’s and early 80’s.
Fast forward a couple of decades and now we are in the era of the 4th generation scanners that are widely used for clinical purposes. (Fifth generation photon counting, electron beam scanners are also available but that remains another discussion). The main differences that are seen in the fourth generation scanners is the presence of a circumferential detector around the patient in the gantry ( which remains stationary ) and a point X-Ray source that spins around the patient in a continuous fashion acquiring the images as it rotates (Fig 2).
(Fig 2 – 4th generation CT Scanner – the X-Ray source is seen on the left and the detector is seen circumferentially around the object being scanned. The source emits a single beam that is captured by the circumferential detector and the source keeps on moving continuously around the object being scanned )
In essence, despite the major differences in the construct and workflow of different scanners, one principle remains fairly constant in conventional CT scanners. Since a point source is used as the X-Ray generator and a single detector moving in an arc(1st gen) or a point source as the X-Ray source with a circumferential detector (4th gen), the distance travelled by the x-ray beam from source to destination, through the object, remains constant. Due to this, the shape of the beam as it transcribes its orbit is almost like a fan in shape, and therefore the conventional CT scan is also called as the “Fan Beam CT” (Fig 3)
(Fig 3 – as the image is generated in a conventional scanner, each of the X-Ray beams labelled a, b and c emanate from the generator, pass through the object and reach the curved detector. Since the origin is a point and the destination is curved, the length travelled by all the three representative beams remains the same. Therefore, the sector being scanned has a fanlike shape, with the narrowest point at the origin of the beam and the terminal part remaining curved )
So, What is Cone Beam CTA ?
Now that we have understood the generation of conventional spiral CT, it is the right time to apply these principles in the angio suite. As we are all aware, neither is the detector in the angio suite a doughnut shaped circumferential detector, not is the detector curved like we see in conventional CT. The X-Ray generator remains almost identical, just miniaturized in size, to oversimplify things. The main differentiating point is in the Flat Panel Detector. As the name suggests, it is flat instead of being curved and this makes things different, and interesting.
The basic principles remain as in spiral CT. There is a source, there is an object and there is a detector. Also this is where the similarities end. The imaging principles have been summarized in the following image (Fig 4)
The sector that is transcribed by the scanner in a flat panel detector, unlike in a curved detector, takes the shape of a cone instead of a fan, and therefore any CT acquisition using a flat panel detector, is essentially a Cone Beam CT Exploration.
(Fig 4 – as the image is generated in a flat panel scanner, each of the X-Ray beams labelled a, b and c emanate from the generator, pass through the object and reach the flat panel detector. Since the origin is a point and the destination is flat, the length travelled the two peripheral beams a and c is considerably longer than the beam in the center of the field, b. Therefore, in this flat panel situation, the sector being scanned has a conical shape, with the narrowest point at the origin of the beam and the terminal part of the scanned sector remaining flat, like the base of a cone)
Conversion of X-Rays into a 2D and subsequently in a 3D image set – how does it happen ?
As we have discussed in the previous sections, in essence, all that a CT Scanner does is shoot an X-Ray beam through an object, and that beam gets to travel through the object and finally get picked up by a detector. On its way to the detector, part of the beam gets reflected, part gets absorbed/modulated in terms of its frequency and part gets to go through and through. It is this modulation that is interpreted by the detector as varying densities (Fig 5).
(Fig 5 – Three representative X-Ray beams named a, b and c are shown here. Beam a does not penetrate the object at all and gets reflected completely back and does not reach the detector. This is the reason for the so called “scatter radiation” in the angio suite. Beam b passes through in its entirety and reaches the detector almost unchanged. Finally, beam c, upon entering the object, gets partially absorbed by the object and a portion of it gets to reach the detector.)
Once this concept is clear, we can begin to build on this and understand how this gets converted into a grayscale representation at the end of the pipeline. The key point here is that at the end of the acquisition, there will be a set of numbers generated by the detector all around the object, and these numbers will essentially be the intensity and frequency of the x-rays that have made it to the detector (Fig 6)
(Fig 6 – The left-hand panel shows the multiple hypothetical locations from where the images have been acquired. The panel on the right-hand side shows the arbitrary numbers picked up by the detectors (D) all around the object of interest.)
Once these numbers, that actually represent the X-Ray attenuation values circumferentially around the object of interest have been acquired, it becomes a matter of simple mathematics to reverse-calculate the values of the boxes that lie within this area.
The more the number of directions at which the data has been acquired, the more is the number of hypothetical squares that can be generated within the area of interest; the more this number, the better is the image. Once the data has been calculated for each square within the area of interest, it is given a value on a grayscale gradient, and the computer converts this into an image of multiple small squares, called pixels, of varying densities. The composite from this is the final CT scan image that we know of and use daily in our clinical work (Fig 7). Each pixel, depending on the sophistication of hardware can be of varying sizes, and this ability to image very small pixels improves with improvements in hardware and software developments. Each pixel represents an area in the anatomy that is being studied, and this part of the body that is represented by a single pixel is called a voxel.
(Fig 7 – The panel on the left hand side shows the reverse calculation of the attenuation values of the individual pixels and their conversion into a grayscale format, and the panel on the right hand side shows the conceptual application of this principle in the final generation of a collection of pixels of varying shades from white to black, finally giving rise to a cross sectional conglomerate of pixels representing an anatomical structure)
Post processing – the concept of “binning”
Now that we have understood how images are generated, and how voxels are converted into pixels, we also need to understand that the images that are presented to us for clinical use are not the same ones that have been acquired using all the principles that we elucidated earlier, in the sense that a lot of post processing goes in between the raw dataset and the finished image on the console. One of the techniques that is used almost universally is “binning”. Let us understand what this is.
The raw dataset out of the scanner at the point of acquisition, especially in the current generation of machines contains too much data for a smooth image. The initial pictures are typically very noisy as the pixel sizes are very small. To circumvent this noise in the final image, developers devised a technique by which the adjacent pixels are merged and the densities are averaged, giving origin to larger pixels and more uniform densities, thereby reducing the noise in the image. While in most cases this does not impact the clinical interpretation and diagnostic use of the data, in situations where we need very high spatial resolution and noise is not much of a concern, unbinned data sets are preferred over binned data sets. The following illustration summarizes the concept of binning and its application to imaging protocols (Fig 8).
(Fig 8 – raw data acquisition at 12×12 pixels, an arbitrary figure produces a raw matrix of 144 pixels. If this is post processed at 2X, you get a 6×6 pixel format with a matrix of 36 pixels. If the same is post processed at 4X, you get a matrix of 9 pixels).
Flat Panel CBCTA exploration – specifics – the concept of “Isocenter”
The isocenter of a biplane catheterization lab is the theoretical point in space where the central X-ray beams of both the frontal (antero-posterior) and lateral imaging systems intersect. This is marked as a small red dot on both the frontal and lateral planes. Since this is a point in space where the maximum number of beams intersect during rotation of the C-arms, it automatically implies that the maximum amount of data about a given voxel will be generated when this voxel is positioned at the isocenter, both in the anteroposterior and the lateral projections. The following sets of diagrams show the concept of the isocenter as well as the correct way to position the object/area of interest in the correct positions in both AP and lateral projections (Figures 9, 10 and 11). All these concepts will be illustrated in the subsequent sections with real world examples.
(Fig 9 – the relative positions of the isocenters in the AP and lateral planes, always marked by a red dot on the gantry)
(Fig 10 – Imaging of a small structure like the spinal cord vasculature. The left hand panel showing the oD isocenter location of the area of interest, and the panel on the right hand side showing the correct positioning in the isocenter, thus optimizing the imaging of the spinal vasculature)
(Fig 11 – Imaging of a small structure like the spinal cord vasculature. The left-hand panel showing the oD isocenter location of the area of interest, in the lateral plane. The patients back is not positioned in a way that keeps the spinal canal in the isocenter. the panel on the right-hand side showing the correct positioning in the isocenter, thus optimizing the imaging of the spinal vasculature).
Clinical example illustrating the concept of isocenter
An elderly male patient presented with progressive paraparesis and double incontinence of subacute onset. MRI of the Dorso-lumbar spine showed the typical features of a dural fistula. A spinal DSA was performed which showed the Adamkiewicz originating from the Left T11 and a radiculo-meningeal branch going to the fistula from the same level. No other pedicles were found to be involved. Patient has severe reservations about surgery, and therefore endovascular approach was considered apter a due explanation of the pros and cons. In view of the complex anatomy and the Adamkiewicz from the same pedicle, a CBCT exploration was done to examining the anatomy further for embolization.
The study was performed under general anesthesia and apnea. Despite these fantastic conditions, my fellow who was performing the study was unable to get satisfactory images as seen on the right-hand side panel. The ASA was seen as was the dural branch but there was no clarity. On closed examination of the source images as shown on the left-hand side panel, it was obvious where the problem was. As it can be seen with the imaginary isocenter lines superimposed, the area of interest was mush much below the point of intersection and thus the image quality suffered.
The study was repeated with the attention being paid to these issues and the images obtained were radically different, with much better delineation of the anatomy, despite using the same physical conditions, image is shown below.
In this exploration, it can be seen that the spinal canal is well within the cross-hairs and the resultant image on the right hand side panel is far more informative that the previous example. The following images will compare them side by side and then the differences will be much more appreciable.
In the images on the right-hand side, with the so called “On” isocenter exploration, it can be very well appreciated that the anterior radiculomedullary branch ascends anteriorly and superiorly to reach the ASA axis. At the same time the radiculomeningeal branch hooks over the ligamentum denticulatum, and travels posteriorly, in a completely different plane to reach the dura at the Dorso-lateral aspect and supply the fistula. With this clarity, the subsequent embolization becomes very safe and effective.
