CT scanning prototypes were developed in the late 1960s and this post describes the stages of technological development for CT scanners (Generations 1-5). After demonstrating the differences between these generations of scanner we describe how the 3rd generation became the dominant architecture and discuss the common projection geometries of parallel beam, fan beam and cone beam.
This post is designed for those that are new to CT or want a refresher on the high-level scan configurations used in different generations of CT scanners. As in our other posts we will embed the youTube videos but if you are interested in skipping directly to the videos that is also an option. If you would rather read about it than watch the videos scroll on by the videos.
The differences in CT generations can also be clearly highlighted in table form for a quick means to understand the geometric differences.
|Generation||Year||Why Developed||Anatomy||Source-Detector Movement||Time to acquire 1 image||Why it died?|
|1 st Gen||1971||To show CT works||Head Only||Translate-Rotate||~5 min||Slow|
|2nd Gen||1974||Image Faster||Head Only||Translate-Rotate||20sec-2min||Slow|
|3rd Gen||1975||Image Faster||All Anatomy||Rotate-Rotate||1 sec||This Geometry won.|
|4th Gen||1976||Make images without rings||All Anatomy||Rotate-Stationary||1 sec||Expensive, not good for scatter.|
|5th Gen||1980s||Fast Cardiac CT||Cardiac Only||Stationary-Stationary||50 ms||Cardiac specific, low x-ray flux.|
For a historical perspective of the scientific developments in Radiology please see Brief History of Radiology. In this post we will give an overview of the different generations in the development of CT from 1st to 5th generation, along with a review of basic scan modes to perform standard CT scans on modern CT.
This is the first time that CT was performed and is the basis for the Nobel Prize in Medicine.
The very first CT scans were performed on a first generation geometry on a CT benchtop. In the benchtop systems patients were not imaged but rather an object to be imaged is placed on a stage that can rotate (i.e. like a slow and well calibrated record player).
Then in order to image patients (rather than biological samples) a rotating gantry is needed so that the patient can lie on an imaging table and the x-ray source and detector will rotate around the patient.
Both of the benchtop and first generation rotating CT systems share a common configuration that is referred to as a translate/rotate acquisition.
In first generation CT scanners, there was one X-ray source and one X-ray detector. So, in order to acquire an axial image of the patient, one ray would go through body of patient and be measured using a single detector. The x-ray source and the detector moved together to collect the data. In order to reconstruction one slice the x-ray source would have to translate many times for each view. Then the source and tube were rotated with respect to the patient (or another object being imaged).
This process was very time-consuming and scanners were slow. Such scanners with only one source and one detector proved that there was tremendous value in CT but they could be very slow taking ½ hr for an average acquisition of several slices.
Second generation CT was a refinement on first generation CT but still using the same general concepts. The translate and rotate acquisition was still used but while 1st generation CT had only one x-ray source and detector, in 2nd generation CT there was a small fan beam appeared that enabled more coverage than just one detector (5-53 detectors at a time).
Second generation CT was significantly faster taking an average exam from a significant fraction of an hour to the order of minutes. An average scan during on this system was ~1.5 minutes. Each slice went from taking 5 minutes on 1st generation to as low as 20 seconds on 2nd generation.
Second generation CT is still a translate rotate acquisition but was significantly faster than 1st generation CT. Multiple versions of second generation CT were built where more detectors used at once lead to more speedup.
Then a significant improvement was again made going from 2nd generation CT to 3rd generation CT where the translation of source within each view was eliminated by having a fan-beam shaped x-ray beam acquiring all the data (for a slice) within each view.
This acquisition mode can be termed rotate-rotate as both the x-ray source and x-ray detector are rotating together. Using a rigid ring the x-ray tube and detector can be mounted such that they rotate around the patient.
In 3rd generation CT a scan takes a few seconds on average compared 2nd generation CT where it was on the order of minutes.
3rd generation CT survived as the base of modern CT for several reasons but mainly the speed of acquisition and mechanical simplicity.
So, the modern CT is typically an X-ray source and X-ray detectors mounted in an arc facing the source. In the sections below we will describe the standard data acquisition methods on these systems.
The modern CT systems may even include new configurations such as 2 tubes and 2 detectors mounted on the same gantry.
Other state-of-the-art systems now include x-ray detectors large enough to cover an entire organ (e.g. the brain or the heart) in a single rotation of the system.
While some have made different classification systems, we believe that all these systems are based on 3rd generation CT where the x-ray source and detector are rigidly mounted on the gantry across from one another. We don’t introduce new generation terminology for 2T2D systems or whole organ coverage systems.
The 4th generation CT geometry is considerably different from 3rd generation geometry in that the x-ray detectors surround the entire circle (much like a P.E.T. detector).
The x-ray source rotates in 4th generation CT and the detector is stationary, so this generation we term rotate-stationary.
The scans times were similar on 3rd generation CT and 4th generation CT and 3rd generation CT required many more detector elements (~ three times as many) to cover the full ring around the patient.
The final generation of CT which is truly a different acquisition method is that of 5th generation CT. In all of the other methods above there is significant mechanical motion of the parts on the gantry.
In 5th generation CT both the x-ray source material and the detector are stationary. In this sense this is a stationary-stationary design.
The x-tube in this design is a scanning x-ray tube, where the electrons are steered magnetically (like in old TVs) rather than physically moving the x-ray tube. This method allows for very fast acquisitions and is ideal for cardiac scanning (with a temporal resolution of a given slice as low as 17ms).
The niche of 5th generation CT was dedicated cardiac scanning. However, these scanners did not have full volumetric coverage and the flux that could be delivered was more limited.
In the end, the third generation CT ended up winning out compared with the relatively niche design of 5th generation CT.
Especially when we went wider coverages because you can get higher power on the third generation CT and the gantry rotation time just keeps getting faster and faster.
With the difference in temporal resolution shrinking between what the electron beam CT could provide and what a good third generation CT could provide the 3rd generation geometries have become the heart of modern CT scanners.
Parallel, Fan beam, Cone beam geometries
After going through the different generations of CT scanners we also wanted to focus on a few terms related to ways that x-rays cover the body during scanning. These go from parallel-beam, to fan-beam to cone-beam geometry in terms of speed of acquisition (i.e. the time required to get all the data needed to make an image).
Parallel beams of x-rays were used in the first generation of CT. The x-ray source was collimated and turned into single ray, a.k.a. a pencil beam. In this case the source and detector had to move to get another single ray.
In first generation CT we need to do all that just in order to get just one view of data, then the whole gantry rotates and to get the second view of the data we have to repeat again and again until we go at least half way around the patient.
After all of that in parallel beam CT we still have acquired only one slice of the patient. This then needs to be repeated for each slice after translating the table. That is why parallel beam CT was replaced with fan-beam CT.
Fan-Beam CT was introduced in 3rd generation CT. The reason it is called Fan-Beam, is because the beam coming out of the x-ray tube makes the shape of a fan.
Instead of translating you just turn on the x-rays and then the whole slice in this direction and here is covered in x-rays so we can get one rotation relatively quickly and don’t have to wait for those translations anymore.
Each rotation in fan-beam CT provides just one slice. Therefore, many rotations are needed to cover the full anatomy. This is where cone-beam CT comes in for even more acceleration in the speed of acquisition.
Then what came after fan-beam CT is called Cone-Beam CT and the geometry is the same as the third generation CT (i.e. an x-ray source and x-ray detector mounted straight across one another). However, in this case the detector has many more rows so the shape of the x-ray beam coming out looks more like a cone than a fan.
Each rotation is still fast (just like in fan-beam CT) but fewer rotations are needed in order to scan the entire anatomy of interest. As you can see in the figure more of the volume is covered at one time and thus we are need fewer rotations in cone-beam CT.
On a modern (state-of-the-art) CT scanner one can scan the whole heart or the whole head in just one single rotation (more discussion on this below).
Now Cone-Beam CT is the basis of modern CT scanners and it is, also called multi-detector CT (MDCT), as there was a gradual progression of rows of CT detectors on systems from 1,2,4,16, 64, until today where state-of-the art systems can have 256 or more rows of CT detectors. That is a huge leap in coverage from the early days of 3rd generation CT.
In this summary table we highlight the differences from parallel-beam to fan-beam to cone-beam. With a sum-what arbitrary distinction drawn between fan-beam and cone-beam when the number of detector rows is larger than 16. There were more major differences required in the CT reconstruction when the number of rows became greater than 16, so that is where we will draw the distinction between fan-beam and cone-beam systems.
|X-Ray Tube /|
Detector within each View
|X-Ray Tube and Detector from view to view||Num of detector rows|
|Fan Beam||—-||Rotate||1-16 detector rows|
|Cone Beam||—-||Rotate||More than 16 detector rows|
We also note that sometimes in the literature when people use the phrase cone-beam CT they are referring to systems which use a flat-panel detectors (that were originally designed for radiography and fluoroscopy procedures) in order to perform the CT. These systems were always cone-beam in nature as they evolved from interventional radiography systems rather than from early 3rd generation CT scanners. In future posts we will discuss these other scanners that also have very large cone-angles as cone-beam CT has been applied to a number of disciplines including: interventional radiography, cardiography, radiation therapy guidance and dental imaging.
At a high-level we went from first-generation CT being Parallel-beam CT to then Fan-Beam CT much faster to acquire a single slice. Now the state-of-the art is Cone-Beam (multi detector) CT (MDCT) where the a whole organ can be scanned in a fraction of a second.