Cone Beam Computed Tomography in Dentomaxillofacial Imaging
by: Predrag Sukovic
President, Xoran Technologies, Ann Arbor MI, USA
From the Winter 2004 AADMRT Newsletter
Various imaging modalities have been used in the dentomaxillofacial fields over the past few decades, none of them with entirely satisfactory results. This is particularly true for more demanding imaging tasks, such as implant planning, temporomandibular joint imaging, detection of facial fractures, lesions and diseases of soft tissue in the head and neck, and reconstructive facial surgery.
In particular, the use of dental implants is becoming the treatment of choice for the replacement of missing teeth. The successful outcome of a dental implant—the osseointegration of the implant—is heavily dependent on precise presurgical planning. Since the functional load on implants can be high, it is important that the implant be placed in a position where it can contact cortical bone and at an angle where the forces are as perpendicular as possible. Selection of the appropriate size and inclination of the implant in both a bucco-lingual and mesiodistal direction requires precise knowledge of the anatomy of the proposed site, including its dimension in all planes, the presence of knife-edge ridges and undercuts, as well as the location of anatomic structures, such as the nasal fossae, the maxillary sinus, and the mandibular canal. An evaluation of the thickness of the cortical bone and the density of the medullary bone is also critical to the success of the implant.
Commonly used dentomaxillofacial imaging modalities, such as periapical radiography, panoramic radiography, and conventional tomography produce only two dimensional and/or distorted images. As a result, a number of practitioners have resorted to outsourcing CT scans for implant planning and other demanding imaging tasks.
Principles and Brief History of X-ray Computed Tomography
CT scanners consist of an X-ray source and detector mounted on a rotating gantry (Fig. 2). During one rotation of the gantry, the detector detects the flux, I, of X-rays that have passed through the patient. The attenuation of monochromatic X-rays in homogenous objects is governed by:
where I0 is the X-ray intensity without the object, x is the length of the X-ray path through the object, and µ is the linear attenuation coefficient of the material at the X-ray energy employed. For inhomogeneous objects, like the human body, the attenuation of X- rays consequently can be described by:
By taking the logarithm of the flux,
one obtains line integrals of the linear attenuation coefficients. These integrals constitute so-called "raw data" that are then fed into an image reconstruction method that generates cross-sectional images whose pixel values correspond to linear attenuation coefficients. The theoretical background for tomographic image reconstruction was laid out in 1917 when Radon established that a 3D object can be reconstructed from an infinite set of 2D projections obtained at varying angles around the object.
The resulting attenuation coefficients are usually expressed with reference to water, and are given in Hounsfield units (HU):
The first CT scanner was developed in 1967 by Sir. Godfrey N. Hounsfield, an engineer at EMI. Since then, CT technology rapidly underwent four developmental generations. The first generation of CT scanners used a single detector element to capture a beam of X-rays, corresponding to the integral of linear attenuation coefficients along a single line. It then translated horizontally to acquire the next line integral. After acquiring all the line integrals for a given position of the X-ray source, both the detector and source rotated one degree.a design known as the "translate-rotate" or "pencil-beam" scanner. Hounsfield's unit belonged to this generation, as did the first commercial CT scanners introduced in 1972. Interestingly, these first generation CT scanners were designed to scan the head only.
Figure 1: MiniCAT/I-CAT scanner developed by Xoran Technologies, Inc. and Imaging Sciences, Inc. |
A second generation of CT systems was introduced in 1975. These systems, also known as "hybrid" machines, used more than one detector and used small fan-beam, as opposed to pencil-beam, scanning. Like the first generation of CT scanners, these scanners also used a translate-rotate design, and were for the most part head-only scanners. While the first iterations of full body CT scanners also incorporated the translate-rotate design, image quality was poor due to patient motion artifacts caused by the significant amount of time required to take the scan.
A B |
Figure 2: Principles of conventional ("fan beam") CT (A) and cone beam CT (B). |
Third generation CT scanners appeared in 1976 and are the systems most widely used today. These scanners use a large, arc-shaped detector that acquires an entire projection without the need for translation. This rotate-only design, frequently referred to as "fan-beam", utilizes the power of the X-ray tube much more efficiently than the previous generations.
Fourth generation scanners shortly followed third generation scanners, replacing the arc-shaped detector with an entire circle of detectors. In this design the X-ray tube rotates around the patient, while the detector stays stationary. Since these fourth generation scanners tend to be more expensive and suffer from higher levels of Compton scatter artifacts, most of the commercially available CT scanners today are third generation scanners.
After an initial period of rapid development, CT technology quickly became mature, and it was not until the early 1990s that CT research began anew. Recent advances in CT include multirow detectors and spiral scanning. Multirow scanning allows for the acquisition of several cross-sectional slices at the same time, reducing scanning times. Today's state-of-the-art scanners have 64 rows of detectors. Spiral (helical) scanning incorporates a moving table with the rotating X-ray tube, with the net effect that the X-ray tube describes a helical path around the patient.
Conventional CT scanners are large and expensive systems designed for full-body scanning at a high speed to minimize artifacts caused by movement of the heart, lungs, and bowels. They are not well-suited for dedicated dentomaxillofacial imaging, where cost considerations are important, space is often at a premium, and scanning requirements are limited to the head. The advent of cone beam CT (CBCT) technology has paved the way for the development of relatively small and inexpensive CT scanners dedicated for use in dentomaxillofacial imaging.
Principles and Brief History of Cone Beam Computed Tomography
Cone-beam CT scanners utilize a two dimensional detector (Fig. 2B), which allows for a single rotation of the gantry to generate a scan of the entire region of interest, as compared to conventional CT scanners whose multiple "slices". must be stacked to obtain a complete image. In comparison with conventional fan-beam or spiral-scan geometries, cone-beam geometry has higher efficiency in X-ray use, inherent quickness in volumetric data acquisition, and potential for reducing the cost of CT. Conventional fan-beam scans are obtained by illuminating an object with a narrow, fan-shaped, beam of X-rays. The X-ray beam generated by the tube is focused to a fan-shaped beam by rejecting the photons outside the fan, resulting in a highly inefficient use of the X-ray photons. Further, the fan-beam approach requires reconstructing the object slice-by-slice and then stacking the slices to obtain a 3D representation of the object. Each individual slice requires a separate scan and separate 2D reconstruction. The cone beam technique, on the other hand, requires only a single scan to capture the entire object with a cone of X-rays. Thus, the time required to acquire a single cone-beam projection is the same as that required by a single fan-beam projection. But since it takes several fan beam scans to complete the imaging of a single object, the acquisition time for the fan beam tends to be much longer than with the cone beam. Although it may be possible to reduce the acquisition time of the fan beam method by using a higher power X-ray tube, this increases the cost and size of the scanner as well as the electric power consumption, thus making the design unsuitable for a compact scanner.
A number of groups have worked on developing task-specific CBCT scanners over the past two decades. Computed tomography angiography (CTA), in particular, has been an active area of investigation due to its lenient requirements for contrast resolution and strict requirements on spatial resolution – a natural fit for CBCT. The first CBCT scanner ever to be built was built for angiography among other tasks at Mayo in 1982 [1].
Fahrig et al. [2] [3] have developed a CBCT system based on an image intensifier and C-arm for use in angiography. Wiesent [4] have also developed a C-arm plus image intensifier system for interventional angiography. Saint-Felix et al [5] developed a CTA CBCT system based on the gantry of a conventional CT scanner which reconstructs vasculature from a set of digitally subtracted angiography (DSA) images. Ning et al. have developed a CBCT angiography imager based on GE 8800 CT scanner with an image intensifier – CCD chain and later with a flat-panel detector [6] [7] [8]. Schueler et al. have developed a CBCT CTA scanner based on a biplanar C-arm system [9]. Kawata et al. also developed a CBCT CTA system [10].
Jaffray and Siewerdsen have developed a CBCT system for radiotherapy guidance based on an amorphous silicon (a-Si:H) flat-panel detector [11] [12] [13]. Cho et al. have also developed a CBCT system for radiotherapy applications [14]. Efforts are being made towards dedicated CBCT-based imaging systems for mammography [15].
Although CBCT has existed for over two decades, its true potential has not yet been fully tapped. Only recently has it become possible to develop CBCT clinical systems that are both inexpensive and small enough to be used in OR, medical offices, emergency rooms, and intensive care. Four technological and application-specific factors have converged to make this possible. First, compact and high-quality flat-panel detector arrays were developed. Second, the computer power necessary for cone-beam image reconstruction has become widely available and is relatively inexpensive. Third, x-ray tubes necessary for cone-beam scanning are orders-of-magnitude less expensive than those required for conventional CT. Fourth, by focusing on head/neck scanning only, one can eliminate the need for sub-second gantry rotation speeds that are needed for cardiac and thoracic imaging. This significantly reduces the complexity and cost of the gantry.
In short, cone beam CT is ideally suited for high quality and affordable CT scanning of the head and neck in dentomaxillofacial applications.
This value of using cone beam CT for dedicated dentomaxillofacial imaging has been recognized by a number of researchers, and several commercial systems are available commercially from Quantitative Radiology ("NewTom 9000" [3]), Hitachi, J. Morita Co. [4,5], as well as from the collaboration of Xoran Technologies and Imaging sciences International (MiniCATTM / I-CAT (formerly DentoCATTM) [6])
Flat-Panel Detector based on Amorphous Silicon
The quality of the reconstructed CT image depends significantly on the quality of the data that is acquired during the scan. The characteristics of the two-dimensional detector used by a cone beam CT scanner affect therefore directly the quality of the CT scan.
MiniCATTM scanner deploys a flat-panel detector constructed from hydrogenated amorphous silicon (aSi:H). These detectors are essentially self-scanned arrays of n-i-p photodiodes and thin-film transistors switches with a scintillator layer. Essentially the same underlying technology is used to construct flat-panel computer monitors and large-area document imagers. Flat-panel detectors are replacing film and image-intensifiers in conventional radiography and fluoroscopy applications.
The characteristics of aSi:H based flat panel detector arrays make them a much better choice for cone beam imaging than an alternative solution: image intensifier coupled with a CCD. Image intensifiers create geometric distortions that must be addressed when processing the data later in the software, while flat-panel detectors do not suffer from this problem. Also, flat panel detectors afford a greater dynamic range than that offered by the image intensifier + CCD camera approach.
Sample Images
Figures 3-6 illustrate the high spatial resolution that can be obtained with CBCT (MiniCATTM / I-CAT) and compare it to that of conventional CT scanner.
A B
|
Figure 3: Axial, coronal, and sagittal slices of a maxillofacial scan obtained with MiniCATTM / I-CAT. |
A B |
Figure 4: Comparison of a conventional CT scan (A) with a CBCT scan (B) Note the higher resolution along the z-axis for the CBCT scan. |
A B
|
Figure 5: 3D renderings of a MiniCATTM scan of a image-guided sinus surgery phantom. The high isotropic spatial resolution makes it ideal for example for image-guided functional endoscopic sinus surgery. |
Figure 6: 3D renderings of a high-resolution isotropic (voxel size 0.2 mm) scan of a dry mandible. |
Future of Cone Beam CT
Its high spatial resolution, smaller size and lower cost has made CBCT a natural fit dentomaxillofacial imaging. Those same attributes will drive the adoption of CT in other markets as well. Xoran Technologies is successfully marketing the MiniCATTM scanner to otolaryngology offices for in-house imaging of the sinuses and temporal bone. In the future, integration of CBCT with image guided surgery systems for functional endoscopic sinus surgery is anticipated (a scan of an image guided surgery phantom is shown in Figure 5). 3D intraoperative imaging devices based on CBCT are also in development.
At the other end of the spectrum, the "big four" companies that manufacture conventional CT scanners are working on mating cone beam technique with the scanners from their product lines. Once accomplished, this would allow the physicians to "freeze" the motion of the heart for cardiac imaging. The latest generation of conventional CT scanners has 64 slices and can cover 4 cm axially. It is anticipated that the need for sophisticated cardiac imaging tools will drive the introduction of truly cone beam helical scanners within several years.
As CBCT scanning is finding more and more commercial applications in medicine, dentomaxillofacial radiology stands as the privileged field that has driven growth of this exciting technology out of the R&D infancy into the commercial maturity.