Veterinary Computed Tomography

CT Physics and Instrumentation – Mechanical Design

Jimmy Saunders and Stefanie Ohlerth

Basic CT Unit Anatomy

A computed tomography (CT) unit consists of a gantry, a patient table, hardware equipment, an operator console and optionally additional workstations.

The gantry is a doughnut-shaped ring containing the X-ray tube, the detector array and associated equipment. The central hole in the gantry accommodates the patient on a sliding table. The X-ray tube rotates around a slice of patient anatomy. This slice represents the X-Y plane, with the X-axis being horizontal and the Y-axis vertical. The isocenter of the gantry is the central point of this plane. The third dimension is represented by the Z-axis, which is along the orientation of the patient table. The patient bed is a sliding tray on a fixed table with an adjustable height and a defined capacity of forward motion. The operator console is located in another room or behind radioprotective screening, and allows operation of the CT units. Additional workstations can be used to review processed image data, but usually not raw data processing.

X-Ray Tube

Basic Anatomy of the X-Ray Tube

An X-ray tube is a vacuum tube that produces X-rays. It is composed of a cathode (filament) and an anode (target). The cathode cup is negatively charged and incorporates a wound tungsten filament that emits electrons when heated. The anode consists of a disk of tungsten or a tungsten alloy with an annular target, called the focal track, close to the edge. The anode disk is supported on a long stem that is supported by ball bearings within the tube. The anode can be rotated by electromagnetic induction from a series of stator windings outside the evacuated tube.

The X-ray tube is enclosed in a housing unit filled with insulating oil. This oil provides electric shielding from the tube voltage, X-ray protection and transmits heat generated in the housing unit to the unit’s surface. The exterior of the housing unit is cooled with a fan, and insulating oil is cooled by passing it through a heat exchanger.

Low-power applications use stationary anode tubes, while for most mid-range and high-performance applications there is a need to utilize rotating anode tubes.

Basic Physiology of the X-Ray Tube

A current of a few amperes (4–8 A) heats the tungsten filament that releases electrons (thermoionic emission) in the vacuum. A high-voltage power source (‘tube voltage’) ranging from 30 to 150 kilovolts (kV) is connected across cathode and anode to accelerate the electrons producing an electron flow (‘tube current’). These electrons collide with the anode material and about 1% of their kinetic energy is converted into X-rays, usually perpendicular to the path of the electron beam. The remainder of energy is converted into heat, causing the X-ray tube to warm up during operation. The temperature of the focal track can increase quickly to 1000–1500°C. Heat diffuses by conduction throughout the anode body and by thermal radiation (infrared radiation) to the tube housing (80%). Heat is removed from the tube housing by convection to the surrounding atmosphere.

Many X-ray systems, including CT, have built-in safety features that will not allow the equipment to be operated in ‘overheated’ conditions. The temperature cannot be measured directly in the focal track. It has to be evaluated based on indirect values that characterize the ability of the anode to store the heat generated during the X-ray emission, such as the anode heat capacity, the anode dissipation/cooling rate or the tube dissipation.

Specificities for CT

Since the invention of CT, its demands regarding the X-ray source never ceased to increase and are largely superior to those of radiography. These specific requirements can be summarized with higher scan power, shorter rotation times (maximum rotation speed), shorter cool-down times and smaller focal spots without compromise on resolution and image quality. In older CTs, the generator capacities and anode disk’s heat storage capacities were insufficient and long time interruptions were needed.

Scan power: typical values for the maximum power are 20/40–100 kW with the high voltage range ranging from 80 to 150 kV.

Focal spots: X-ray tubes use typical focal spot sizes of 0.5–1.2 mm. Specific innovations for CT are the ‘flying focus’ allowing for control of the focus position on the anode during the scan or the electromagnetic control of the electron beam, which allows switching of the focal spot position both in the fan and in the Z-direction, providing overlapping sampling.

Rotation speed: the traditional glass tube technology is not adequate in terms of required precision and stability to sustain the very high rotation speed, up to 10 500 rotation/min, of high-performance tubes. Despite its higher thermic dissipation and lower cost, glass has been replaced by the metal ceramic technology, which is more precise and better able to sustain the constraints related to the rotation speed.

Cool-down time: different approaches can be used on their own or in combination to shorten the cool-down times and improve the heat storage capacities of CT.

The ‘brute force approach’ was the main way used for three decades. This approach consists of an increase in the thermal capacity of the anode by increasing its diameter and mass. This system has obvious limitations, as it still uses radiative cooling.

The ‘material approach’ is based on a slow evolution in the materials used for the anode.

Direct:

Use of circular grooves in the anode support to increase the contact and improve cooling.

Use of special liquid metal vacuum bearings that allow faster anode rotation.

Focal track made of a mixture of rhenium and tungsten. Rhenium has a higher linear expansibility than tungsten and slows the rate at which anode crazing occurs.

Anode ‘compound’/increased thermal capacities: use of molybdenum or graphite with tungsten in the anode disk. Molybdenum has twice the specific heat capacity and half the density of tungsten. Graphite has an even higher specific heat capacity and a quarter of the density of molybdenum. It increases thermal capacity.

Replacement of the ball bearings by a liquid metal (gallium) that allows the evacuation of heat by conduction.

Indirect:

Multiple detectors allow reduction of the heat produced via reduced scan duration by a factor approximately equivalent to the number of rows. Manufacturers developed systems with up to 1000 rows.

The ‘paradigm shift’ corresponds to innovations in X-ray tubes.

In 2000, Siemens developed the Straton tube, also called the rotating envelope tube, for its high-tech scanner. This tube uses direct convective cooling, exclusively of the anode, with a cooling oil stream at the anode’s back surface. As a result, the cooling rate is vastly increased to 4.8 MHU/min, eliminating the need for large heat storage capacities of the anode disk and reducing waiting times due to anode cooling in the clinical workflow.

This lighter tube also presents a solution for the acceleration/pressure centrifuge high G (above 20 G).

Another innovation on the X-ray tube in relation to CT is the dual-source CT (two tubes, two detector fans); the main advantage of this architecture is its improved temporal resolution. In today’s CT scanners, the gantry rotation time is reduced to about 0.35 s and it is mechanically challenging to reduce that time even further, which justifies the renewed interest in multisource architectures.

Collimators and Filtration

CT systems feature various collimators, filters and shielding designs, which provide filtration of the X-ray spectra, definition of the measured slices, guarding detectors against scattered radiation and general radiation protection. These vary from scan to scan but always offer the same functions.

Collimation

Collimation in CT serves to ensure good image quality and to reduce unnecessary radiation doses for the patient.

Collimators are present between the X-ray source and the patient (tube or pre-patient collimators) and between the patient and the detectors (detector or post-patient collimators).

The tube collimator is used to shape the X-ray fan beam before it penetrates the patient (restrict the X-ray flux applied to a narrow region defines the shape of the X-ray beam). It consists of a set of collimator blades made of highly absorbing materials such as tungsten or molybdenum. The opening of these blades is adjusted according to the selected slice width and the size and position of the focal spot. It defines slice thickness for single-slice CT. Tube collimators define the dose profile according to the required slice thickness. Post-patient collimators improve the slice sensitivity profile by giving a more rectangular shape. Table 1.1 shows the features of CT collimators.

Table 1.1 CT collimator features.

*Can only be implemented in scanners with a rotating detector.

Filtration

The X-ray photons emitted by the X-ray tube exhibit a wide spectrum. The soft, low-energy X-rays, which contribute strongly to the patient dose and scatter radiation but less to the detected signal, should be removed. To achieve this goal, most CT manufacturers use X-ray filters.

The inherent filtration of the X-ray-tube, typically 3 mm aluminium equivalent thickness, is the first filter. In addition, flat or shaped filters can be used. Flat filters, made of copper or aluminium, are placed between the X-ray source and the patient. They modify the X-ray spectrum uniformly across the entire field of view. Because the cross-section of a patient is mostly oval-shaped, some manufacturers use shaped (or bow-tie) filters. These filters have an increased thickness from center to periphery, allowing them to attenuate radiation hardly at all in the center but strongly in the periphery. They are made from a material with a low atomic number and high density, such as Teflon.

In some machines comb-shaped collimators close to the detector array are used to decrease the effective detector element width and thus increase the achievable geometrical resolution.

Detector Systems

The detector is the system for quantitative recoring of the incident ionizing radiation. It acts in two steps.

1. The reception of the incident X-ray photon via X-ray-sensitive detector elements with a specific geometrical configuration.

2. The transformation of the X-ray photon into a corresponding electrical signal, that is then amplified and converted from an analog to a digital form (via analog-to-digital converters). This step is relatively easily specified and submitted to few fluctuations.

There are two detector types.

Ionization chambers, mostly filled with the noble gas xenon under high pressure. Gas detectors have become obsolete due to their limited detection efficiency and the difficulty in manufacturing them for multi-row design.

Scintillation detectors, in the form of crystals such as cesium iodide or cadmium tungstate, and ceramic materials such as gadolinium oxysulfide. These detectors are now predominantly used mainly because of their short decay time, which is an essential factor in subsecond scanning times since. Ultra-fast ceramic (gadolinium oxysulfide based) have superior characteristics in this area, making them the best choice for spatial resolution and image quality.

The alternative detector concept is the flat-panel technology. Potential advantages are the possibility to scan with wider cone angles without the need to develop detectors with 1024 rows or more and the high spatial resolution, particularly in medium to large field of view scans. Flat-panel detectors were developed for digital radiography and their use for CT is currently being explored by manufacturers.

Gantry Anatomy

Third Generation

With third-generation CT, simultaneous rotation of the X-ray tube and detector array became possible (rotate/rotate geometry, rotating gantry). Moreover, the number of detectors and the angle of the fan beam were increased considerably so that the X-ray beam could scan the entire patient. The translational motion of first- and second-generation CT scanners could therefore be eliminated, which reduced the scan times substantially.

In the beginning, third-generation scanners suffered from the problem of ring artifacts. Each detector in third-generation scanners is responsible for the data corresponding to a ring in the image. Detectors close to the center of the detector array are responsible for a ring with a smaller diameter than those detectors towards the periphery of the detector array. Because there is always a certain amount of electronic drift associated with each detector, this causes gain changes between detectors, finally leading to ring artifacts. Today, modern technology has overcome this problem so that third-generation CT scanners are free of ring artifacts.

Multislice CT systems always use a third-generation technology and they provide scan times as short as 0.5 s.

Fourth Generation

Because of the problem of ring artifacts with third-generation scanners, fourth-generation CT scanners were designed. The detectors are placed separately in a stationary 360° ring around the patient and only the X-ray tube rotates (rotate/stationary geometry). Whereas in third-generation scanners, data are acquired by the detector array simultaneously, a single detector collects the data in fourth-generation CT over the period of time that is needed for the X-ray tube to rotate through the arc angle of the fan beam. Each detector also represents its own reference detector. In this way, ring artifacts were avoided in fourth-generation scanners.

This technology requires many detectors because the detector array covers a 360° angle. It is not used nowadays to design multislice CT units because of the high costs for such an immense number of detectors.

Slip-Ring Technology

In third- and fourth-generation scanners, the X-ray tube rotates around the object. This also applies to the detectors in third-generation scanners. In combination, this is also referred to as a ‘rotating gantry’, although not all parts of the gantry rotate. These components require a number of electrical connections for high-voltage power, data transmission and control. In most early CT systems, the connections between the components on the rotating side of the gantry bearing and the power sources, computers, etc., on the stationary side of the bearing were made using cables. They were of finite length and allowed a rotation of perhaps 700°. As a result, these systems had to stop and reverse rotation directions between images.

The alternative to this cable system is the slip-ring technology. It allows the continuous circular rotation of the X-ray tube and other components of a CT system. In a slip ring, electrical brushes allow connections between continuously rotating and stationary components. The slip-ring design made it possible to achieve greater rotational velocities allowing shorter scan times. It finally enabled the design of the modern helical CT scanner.

Multislice CT

Helical CT represents a CT system using slip-ring technology in which continuous X-ray tube rotation is used along with simultaneous and continuous table translation through the gantry. The X-ray tube describes a helical path around the object. The term ‘helical CT’ is equivalent to spiral CT, which is actually an inaccurate term (a spiral decreases in diameter). Helical CT scanners are named single section (single slice, single detector row), dual section (dual slice, dual detector row) or multisection (multislice, multidetector, multi-row) according to the maximum number of slice images generated per gantry rotation.

Helical CT technology makes it possible to image a given volume much more quickly (e.g. 30 s for the entire abdomen). More importantly, though, it allows a volume to be imaged during a more consistent phase of contrast enhancement. It is of significant benefit for CT angiography and multiphase abdominal imaging. The extent of sequential coverage, or the total time of scanning, is generally limited by X-ray tube heating.

The relationship between the incremental table movement and the selected slice width during one rotation of the gantry is described as ‘pitch’. In single-slice CT the pitch describes the ratio of the table movement per 360° gantry rotation to the collimator width (collimator pitch). In multislice CT, the detector pitch describes the ratio of the table movement per 360° gantry rotation to the detector width. The collimator pitch, then, defines the ratio of the detector pitch to the number of detector rows in multislice technology. The pitch influences patient dose, scan time and image quality. Increasing the pitch decreases scan time and reduces motion artifacts. However, the effective section thickness as well as image noise increases, too. For clinical studies, a pitch of 1–1.5 is commonly used.

Since the helical data set does not correspond to sequential plane data, it needs to be reconstructed via interpolation into planar image data sets before the actual CT reconstruction.

With single-slice CT, detectors are rather wide in the Z-axis, e.g. 1 × 20 mm. Almost the entire detector element is actively detecting radiation, and slice thickness is determined by the collimator width. Per default, the collimator width is always smaller than the detector width. Therefore, for single-slice CT, slice thickness can be decreased via smaller collimator width; however, utilization of the X-ray beam is lower, therefore signal-to-noise-ratio decreases as well. This may be partially compensated by increasing the mAs. As an advantage, partial volume averaging decreases and spatial resolution is improved with thinner slice thickness.

With multislice CT, detectors are much smaller (e.g. <1 × 1 mm). The detector size determines the smallest possible slice thickness and the collimators determine the number of detectors used. If only the central two detectors are used, the slice width can be reduced below the detector width. To allow for variation in slice width and to decrease scan time, the signals from multiple rows of detector elements can be combined, so-called binning. Binning can be performed during the requisition or from raw data after scanning.

There are two detector array designs in multislice systems: those with detector elements of equal width (equal-width design) in each detector row and those with detector elements of unequal width in the different detector rows (unequal-width design).

With multislice CT, a much higher anatomic coverage can be achieved with the same pitch and slice thickness than in single-slice CT. For the same anatomic coverage and scan time, with single-slice CT one has to either increase the pitch or the slice thickness. But image quality is then degraded considerably.

There are many advantages with multislice CT. Scanning is faster, providing better temporal and contrast resolution and fewer motion artifacts. Consequently, multiphase studies (e.g. arterial, venous, portal phase studies) became possible. Thinner slice thicknesses are possible, which improves spatial resolution and reduces partial volume averaging. Due to more patient length scanning per rotation, higher X-ray tube current settings may be used, which in turn reduces image noise.

Table 1.2 shows typical performance characteristics for a CT scanner in 2010.

Table 1.2 Performance characteristics for a CT scanner in 2010.

Moving Gantry

Most CT units include a moving table and a fixed gantry housing. However, for certain purposes, moving or sliding gantries have been developed. Instead of the table moving into the gantry, as in conventional CT, in this case the table is fixed and scanning is accomplished by moving the gantry over the patient. In human oncology, for example, it may be an advantage during the course of irradiation if a CT scan can be performed for treatment planning adjustments immediately prior to irradiation. For this, patients are positioned on a common and fixed treatment table, which is integrated in a combined CT and linear accelarator irradiation system. The irradiation system and the CT gantry are positioned on opposite ends of the table so that, by rotating the treatment table, linac radiotherapy or CT scanning can be performed. Moving gantry systems are also designed for usage during surgery or angiography.

Scanning Modes

A routine scan requires a scout radiograph for anatomical orientation and scan region (slice) selection and scan performed in sequential or helical mode.

Scout Radiograph (Survey Radiograph, Localizer Radiograph, Scanogram, Topogram, Scout View)

A survey radiograph, similar to a conventional radiograph, is very useful for selection of single slices or complete scan regions. This radiograph is taken with a low dose and low spatial resolution by transporting the patient slowly through the field of measurement with the X-ray tube in a fixed position with radiation emitted continuously or in pulsed mode. Lateral scanograms are particularly useful to select the gantry tilt according to anatomy.

Sequential Scanning (Axial Scanning, Single-Slice Scanning)

For a long time, CT examinations consisted of scanning single slices sequentially. A single slice is scanned, then the patient is transported for a scan increment, mostly equal to the chosen slice thickness. Then, a second scan is taken and the procedure is repeated. This examination mode is relatively time-consuming and has been largely replaced by the faster helical CT. One fundamental disadvantage is that overlapping images for 3D image reconstruction are generally not available.

Modern scanners offer automated and therefore fast modes for scanning single slices sequentially. Cardiac scanning may be a future indication.

Dynamic Scanning (Serial Scanning)

Dynamic CT is used to record temporal changes in the density characteristics of an object. Typically, dynamic scanning is used to assess contrast medium dynamics. A representative selected slice is scanned repeatedly or multiphase examination of a complete organ is performed before, during and/or after administration of contrast medium. The observed changes may represent physiological processes, such as heart motion or breathing, or pathological processes such as portosystemic shunts. Dual-phase CT angiography is a minimally invasive technique, which provides an excellent 3D representation of portal and hepatic vascular anatomy.

Material-Selective Scanning (Dual-Energy CT)

Dual-energy methods serve to obtain information about the material composition in the tissues examined. To achieve this, a selected slice is scanned with two different spectra, i.e. with different high-voltage values and possibly with different filtration. This can be done in two successive scans or by switching the high voltage rapidly from projection to projection.

Table Design

Many CT tables are made of a carbon fiber material because it will not cause artifacts when scanned. The movement of the table is referred to as incrementation (incrementation indexing). All table designs have weight limits that if exceeded may compromise increment accuracy. The maximum table load on actual CT machines is between 200 kg and 330 kg. Various table attachments and positional aids are available for specific body parts. For large animal CT these are usually custom made (for details, see Chapter 39).

Proprietary CT Terminology (Table 1.3)

Table 1.3 Proprietary CT terminology.

Further Reading

Bushberg JT, Seibert JA, Leidholdt EM and Boone JM (2002) Computed tomography. In: Bushberg JT, Seibert JA, Leidholdt EM and Boone JM (eds) The essential physics of medical imaging (2e), pp 327–72. Philadelphia, PA: Lippincott Williams & Wilkins.

CHAPTER TWO

CT Acquisition Principles

Tobias Schwarz and Robert O’Brien

Image Parameter Selection

Introduction

When running a CT scan the operator of a CT unit is faced with a large number of selectable settings, which can be intimidating. The following guidelines are designed to help in the selection process and can be used to set up individual CT protocols.

Body Part Selection

In all modern CT units, the operator needs to preselect a body part folder, which contains different protocol options. The body part selection includes hardware choices such as selection of specifically shaped bow-tie filters for beam hardening compensation. These are not adapted to veterinary patients and therefore it is worth testing other body parts protocol groups as well. There are often restrictions on protocol uses and topogram selections and it is not always possible to reuse topograms when changing protocol groups.

Tube Rotation Time

The tube rotation time selected should be as short as possible (usually between 0.5 and 1 s) for all body parts with anticipated movement, such as thorax, abdomen and spine, to avoid motion blur. For non-moving body parts in anesthetized patients, such as head, neck and extremities, it can be advantageous to increase the tube rotation time to 2 s to reduce view aliasing artifact (see Chapter 4). However, this might not be necessary in modern CT units, where the software is designed to ensure a sufficient number of views even with shorter rotation times. It is worth running some tests and comparing 1 and 2 s rotation time images for artifact magnitude. When selecting the tube rotation time, the effect on the mAs needs to be considered. In some units, changing the tube rotation time will automatically adapt the mA to result in the same mAs and vice versa, whereas in other units the mAs will change according to the selected product of mA and time.

Current and MAs Product (Figure 2.1)

The current ranges usually from about 50 mA to 400 mA in modern CT units and there are either a large number of selectable settings or it can be selected freely within that range. The deciding factor is the applied current during the entire rotation of the tube, or mAs product. In some units, it is the mAs that is selected, whereas in others the mA is selected.

Figure 2.1 CT images of a canine cervical spine acquired with (A) 100 mAs and (B) 200 mAs, and otherwise identical settings. The image noise is represented by the general graininess of the image, which is reduced in (B).

There is an inverse proportional relationship between mAs and image noise. Thus higher mAs settings will reduce noise. Small slice thickness increases image noise. Therefore if a small slice thickness is selected (2 mm or less) the mAs should be increased to keep the noise at an acceptable level.

The mAs and the kilovoltage both contribute to the heat load of the X-ray tube. Depending on the heat capacity of the tube, the combination of very high mAs and kilovoltage settings is usually limited.

All CT units have special protocols with low mAs setting (usually 80 mAs or lower) for infants to minimize radiation exposure levels. Despite the fact that most small animals usually are close in size to an infant, these low mAs settings do not always generate high-quality images for the head, spine and abdomen in dogs and cats. In small animals, mAs settings between 100 mAs (thorax) and 250 mAs (head and spine) are usually adequate.

All modern CT units have one or two adaptive mAs output options that can modulate the current along the Z plane and/ or the X–Y plane. The shoulder region is a good example where both can be applied. As the patient’s neck travels through the gantry, the increase in body thickness at the shoulder region (Z plane current adaptation) requires an increase in mAs settings to minimize noise. But since the shoulder area is also wider than high, different mAs settings are optimal at different positions of the gantry circle (X–Y plane current adaptation). Adaptive mAs settings are designed to minimize radiation exposure and tube load and therefore do not always improve image quality in veterinary patients. This should be tested before using it.

Kilovoltage

In CT, the kilovoltage settings are usually high and there is a limited number of choices, usually 80, 120 and 140 kV. The 120 kV setting is adequate for almost all small animal patients, and can be applied universally. In large patients, the 140 kV setting would be preferential to ensure adequate penetration; however, the maximum kV settings usually only work in combination with relatively low mAs settings. Maximizing the mAs at the expense of the kilovoltage settings is usually the better option in this case.

Scan Field of View (Figure 2.2)

During a CT image acquisition, the X-ray tube emits radiation that is collimated on all sides to match a fan-shaped array of detectors on the opposite side of the gantry, resulting in a wedge-shaped area of the gantry being exposed to radiation. The X-ray tube and the detectors circle around the gantry center during the exposure, resulting in an area of circularly overlapping wedges of exposed areas within the gantry. The scan field of view (SFOV) is the area of the gantry where there is complete overlap of these wedges. Image reconstruction in a matrix requires all data from all projections. As a result, a CT image can only be selected from within the SFOV. In most modern CT units, the SFOV is no longer selectable, because the maximum SFOV is always preselected. However, in General Electrics scanners, the SFOV is selected by the operator, and it is important to select the SFOV larger than the maximal patient diameter to avoid out-of-field artifacts (see Chapter 4). Keeping the SFOV only as big as necessary improves spatial resolution marginally.

Figure 2.2 Illustration of the gantry with the trace of the rotating X-ray tube (black ring) and wedge-shaped projections obtained along its course (yellow wedges). The SFOV (within the red ring) is the area that is included in all projections. Only for this area is a full set of data acquired and images can only be reconstructed from within the SFOV.

Display Field of View (Figure 2.3)

The display field of view (DFOV, synonym FOV) is the area of the SFOV from which an image is reconstructed. The DFOV cannot exceed the SFOV and should be kept as small as possible. During the reconstruction of a digital image, the matrix size defines the number of pixels used to form the image. For instance, in a 512² matrix, each image will be composed of 262 144 pixels. Displayed over a 40 cm DFOV, the pixel length will be 0.78 mm; displayed over a 15 cm DFOV, the pixel length will be 0.29 mm. Thus sizing of the DFOV has a significant impact on image resolution.

Figure 2.3 Thoracic CT image of a dog obtained with (A) a 42 cm DFOV and (B) a 14 cm DFOV and otherwise identical settings. There is a dramatic increase in image resolution in (B) compared with (A), best visible in the small lung vessels, bronchi and the lung nodule. Coning down the DFOV improves image resolution.

It is often essential to include the whole body cross-sectional area to assess the entire patient, so that coning down too far may not be practical. However, all newer CT scanners allow a targeted retrospective reconstruction from the raw data so that images with a smaller DFOV of specific areas can be obtained after the procedure. Due to the size and geometry of the detectors, decreasing the DFOV below 10 cm usually does not further increase the image resolution.

Gantry Tilt (Figures 2.4 and 2.5)

In most CT units the gantry can be tilted to up to 30° in both directions. The topographic radiographs usually cannot be obtained with tilt, but all CT studies can be done tilted. Some older CT unit such as Elscint scanners even had a table swivel to align in the lateral orientation. Gantry tilt is useful for optimal alignment of the image plane with anatomic planes of interest to achieve optimally aligned sequentially reconstructed images. The most common use of gantry tilt is to align the scanning plane with the lumbar and particular cervical spine. Gantry tilt can be essential to achieve artifact-free images of patients with metallic foreign material or to avoid inclusion of artifact-inducing patient structures such as the head in elbow CT.

Figure 2.4 The creative use of gantry tilt can facilitate diagnostic images in challenging patients, such as in this dog with a paraspinal bullet seen on the topogram in (A). The transverse scan plane (red line) results in a non-diagnostic transverse CT image (B) for the area of interest due to the strong metallic artifact. The tilted (blue line in A) CT image (C) eliminates the artifact. These images can be retrospectively reconstructed into a transverse aligned CT image (D). A combination of the sagittal reconstructions from both tilted acquisitions (E) displays the diagnostic information from almost all areas obtained with this strategy.

Figure 2.5 There are many ways to CT scan a canine elbow joint. (A) To avoid streak artifacts on elbow CT images emanating from the head and neck, the head is fixed in a laterally flexed position and the forelimbs are slightly elevated. (B) Lateral topogram illustrating the slightly tilted scan plane (colored lines) to avoid the presence of the head and neck in the scan plane without extreme flexion of the neck. (C) Resulting artifact-free elbow CT image.

There are several caveats when using gantry tilt. It is preferable to adjust tilt at the gantry itself for patient safety reasons. Since the gantry tilts only in one plane, it is essential to position the patient in the appropriate recumbency. For instance, aligning with intervertebral disk spaces via gantry tilt is not possible in lateral recumbency. When the gantry is tilted, the scan direction remains along the Z plane, not the tilted plane. This requires for instance for longer spine sequences to leave the DFOV sufficiently large to accommodate the anatomic area of interest within the DFOV over its entire length. Some reconstruction softwares are either unable to reconstruct orthogonal planes from tilted acquisitions or introduce image distortion (e.g. in Osirix 3.7). This can be remedied with additional software (macros).

The image reconstruction from helical scans with gantry tilt always requires a longer reconstruction time because the necessary mathematical interpolation is very complicated: the X-Y plane of the tilted gantry needs to be normalized to the X-Y plane of the moving patient table. In some CT scanners multiplanar reconstructions or helical acquisitions are not even possible with gantry tilt. The latest generations of CT scanners with 80 or more detector rows often do not have the costly mechanical gantry tilt. An isotropic image resolution can be achieved with any orthogonal image reconstruction so there is less of a need for this feature. For veterinary use with non-standardized patients and problems, gantry tilt remains a useful tool.

Slice Thickness

General Considerations (Figures 2.6, 2.9)

Slice thickness (synonym slice width) is arguably the single most important setting to select for a CT scan. Slice thickness is directly proportional to the magnitude of volume averaging (synonym partial volume artifact) and inversely proportional to the magnitude of image noise. Thus thick-slice CT images are blurry but contain little noise, whereas thin-slice images are sharp but noisy. There are several strategies to deal with this dilemma.

Thin-slice images should be obtained with higher mAs settings to keep the noise at an acceptable level.

Structures with a wide inherent object contrast, such as bone, nasal turbinates and lungs, should be viewed with wide window settings. Wide window settings suppress the visibility of noise. Therefore a higher noise level is tolerable as long as these image series are only reviewed with these window settings.

Structures with a narrow inherent object contrast, such as the brain, spinal cord, liver and other soft tissue organs, require viewing with a narrow window setting, which enhances noise. Therefore a thicker slice width (3 to 5 mm) should be selected in most circumstances for these structures.

Most CT users have a probably subconscious preference for sharp, thin-sliced images with wide window settings, because these are aesthetically more pleasing than blurry thick-slice images windowed narrowly to the tolerance level for image noise. It is important to select slice width and window width rationally and not intuitively.

Figure 2.6 Canine elbow CT images acquired with (A) 2 mm and (B) 1 mm slice thickness and otherwise identical settings. The increased slice thickness results in more blur but less image noise. For orthopedic settings, the thin-slice image is preferable.

Another limitation of thin-slice imaging is the limited coverage of the patient or the prolonged time to scan a defined area. The tube heat capacity is also limited, so it is not always possible to scan a large area with very thin slices. Helical and, in particular, helical multi-slice CT have vastly improved our ability to scan a large patient volume in a short time span with thinner slices. Selection of the appropriate slice interval in sequential CT and pitch in helical CT helps to find the optimal compromise in slice thickness selection.

Another artifact that can be minimized with thin-slice imaging is non-linear partial volume (see Chapter 4), which contributes to the streaks in the caudal fossa of the calvarium. By obtaining images with a thin slice width originally and then fusing the raw data into bigger sections, non-linear partial volume artifact is minimized and noise is reduced. Because of the binning of thin detectors, all thick-slice images in multislice CT are obtained as a fusion from smaller slices, and non-linear partial volume is reduced automatically.

Thick-slice images can be appropriate for specific purposes and the partial volume artifact can work favorably to achieve this. It is, for instance, not necessary to select the smallest slice width to detect pulmonary nodules. Due to the partial volume artifact and the large density differences in the lungs, a pulmonary nodule that is smaller than the slice width will still create a visible, albeit blurry, opacity. Therefore a 3 to 5 mm slice width is acceptable for lung metastasis screening in dogs and cats. A short thin-slice series can then be obtained as a follow-up.

Slice Thickness Selection and Detector Design

Single-Slice CT (Figure 2.7)

In a single-slice CT unit, submillimeter-thin and usually 10 mm long detector cells are aligned in a row in the X-Y plane of the gantry. The position of a mechanical collimator, made of two lead jaws, determines the exposed width of the detectors and thereby the slice thickness. Slice thickness cannot be modified retrospectively from raw data after the image acquisition in a single-slice CT unit.

Figure 2.7 Illustration of a single-slice CT detector array. Two lead jaws restrict the X-ray beam exposure of the detector elements and thereby determine the slice thickness.

Multislice CT (Figure 2.8)

In a multislice CT unit, several rows of detectors are lined up along the Z-axis of the gantry. The slice thickness is determined by the combination of detectors from which data are pooled together, a process called binning. It is possible to reconstruct CT images from the same acquisition both as thin- and thick-slice images. The ability to change the slice thickness retrospectively from the raw data is one of the major advantages of multislice CT. Dual reconstruction should be part of the standard protocol for almost all body parts with multislice CT.

Figure 2.8 Multi-slice CT detector row design and collimation options. (A) Equal-width detector-row design with eight rows of 1.25 mm wide detectors. (B) Unequal-width detector-row design with rows of two each 1 mm, 1.5 mm, 2.5 mm and 5 mm wide detectors. This is the typical design of a ‘four-slice’ CT unit as only a maximum of four consecutive images can be reconstructed per tube rotation from these eight rows of detectors. (C) The collimator is used to determine the number of detector rows to be exposed in multislice CT. (D) In addition, the outermost pair of detectors can be partially collimated, resulting in more possible slice thickness options. In this case, four images of 1 mm slice thickness are generated. (E) The collimator can also be used to generate ultra-thin slice images if only the central detector pair is used. This will increase scan time as it does not use the full array of detectors.

A basic understanding of the detector alignment helps in selecting the appropriate slice thickness. Two design types exist. In an unequal-detector-width design only the central detectors are very thin and the detector width enlarges symmetrically peripherally. This design is cheaper and prevalent in older units and scanners with fewer detector rows. In an equal-detector-width design all detectors have a very small width (0.5–1 mm). This design is more expensive and prevalent in the newest generation of scanners with many detector rows. The mechanical collimator is used to determine the number of detector rows to be exposed.

A CT image sequence can only be reconstructed with slices of equal thickness. Therefore in an unequal-detector-width-design CT scanner, only certain binning combinations are numerically possible. The number in the multislice CT unit name (e.g. ‘four-slice CT’) usually corresponds to the maximum number of detector row combinations that can be binned together, not the number of detector rows, which is usually greater. For instance, a so-called ‘four-slice scanner’ usually has eight detector rows with two 1 mm, two 1.5 mm, two 2.5 mm and two 5 mm detectors. Possible binning combinations are then 4 × 2.5 mm, 4 × 5 mm, 2 × 10 mm and 1 × 20 mm. The outermost detector row can be partially collimated by the collimator, adding further binned slice thickness options of 2 × 0.5 mm, 4 × 1 mm, 4 × 8 mm.

Slice Thickness and Slice Sensitivity Profile (Figure 2.9)

Due to the fact that the focal spot of a CT unit is much smaller than the selected slice thickness, the X-ray beam in CT is not truly a slice but a wedge and is thus thinner at the isocenter of the gantry than at the detector level. The slice sensitivity profile expresses this relationship mathematically and states that the contrast resolution degrades towards the slice periphery. It is possible to reconstruct image data only from the center of the slice (half-width-height-maximum) and some manufacturers (e.g. Siemens) offer two different reconstruction options for each selected slice thickness: the thin-slice option reconstructs in half-width-height-maximum mode, which increases image resolution; and the thick-slice option reconstructs the whole slice width data, allowing the use of higher pitches in helical CT.

Figure 2.9 (A) The 5 mm slice thickness image of the nose reveals ambiguous features of the nasal turbinates due to partial volume related image blur. (B) The 1 mm thick image supports the diagnosis of bacterial rhinitis based on the presence of normal turbinate and secretion. Leaving a gap between thin slices is a better option than obtaining continuous thick slice images of mediocre quality. (C) Frontally reconstructed view of a canine elbow specimen obtained with sequential scanning and a 1 mm slice width and interval. (D) CT arthrography imaged sequentially with a 1 mm slice width and a 0.5 mm interval shows a smoother outline of the subchondral bone due to a reduction in stair step artifact

(images courtesy of Andrew Gendler).

Slice Interval (Figure 2.9)

The slice interval (synonym slice increment) is the interval at which CT images are acquired in sequential mode CT. In helical mode, data are obtained continuously, hence slice interval is not applicable here. The slice interval is selected in relation to slice width. As a default most units will preselect a slice interval equal to the slice width, ensuring a continuous image acquisition.

Continuous scanning is essential if the anatomy changes rapidly along the Z plane, which is the case for most body parts. However, some body parts, such as the nose, do not rapidly change their anatomic features along the Z plane. In older CT units, where tube heating is an issue it might then be better to obtain 1 mm thick images for maximal anatomic detail of the nose, but leave a 2 mm slice interval to not overload the tube. If continuous scanning is essential, such as for a metastasis check of the lungs, then helical mode should be considered.

For some body parts another compromise that can be made between limited tube heat capacity and optimal image detail is to obtain several small series of continuous slices and leave gaps between them. For instance for spinal CT the intervertebral disk spaces can be scanned continuously and gaps can be left over the vertebral bodies. Leaving a slice gap will always limit the ability to perform orthogonal reconstructions.

Overlapping scanning is rarely performed because of tube heat issues, but this might be useful if maximum image detail is required for a small area. Orthogonal reconstructions in particular will improve with overlapping images.

Sequential Versus Helical Mode

In sequential CT (synonym axial) mode, a complete set of data is acquired from all angles of the tube position, resulting in optimal image resolution. In helical CT, the data represent a helix of data, and the incomplete data set has to be completed by mathematical guesses or interpolations. Therefore helical CT images have a lower image resolution than sequentially scanned CT images. In general, if maximum image detail is required, a sequential CT should be performed. However, there are other factors that influence image quality. Body parts with anticipated motion are preferentially scanned helically in any scanner, as the likelihood of occurrence of motion is lower during the shorter helical acquisition time. Also, the ability to perform interleaved image reconstructions helically and the fact that with a multislice CT there is data redundancy of each anatomic area can make helical mode equal to or better than sequential scanning for overall image quality. In CT units with more than 16 detector rows, there is little more benefit of performing a sequential CT. In units with lower numbers of detectors, it is often still preferential to obtain head and spine images in sequential mode, but this needs to be tested.

Helical Pitch

Introduction (Figure 2.10)

In helical CT, the patient moves through the gantry while parts of the gantry rotate around it. This results in a helical (synonym spiral) set of continuous data that represents the entire volume of the scanned area. The relationship between the table increment during one full gantry rotation and the slice thickness is expressed as the pitch, a unitless number. Pitch is directly proportional to image blur, therefore a highly pitched CT scan results in a very blurry image. In single-slice CT, the collimation and slice thickness are synonymous. In multislice CT, however, this is not the case, and pitch can be defined as the relation of table increment to the collimated width of all included detectors combined (collimator pitch, similar to single-slice CT) or in relation to the slice thickness of each detector (or binned combinations thereof, detector pitch) (Boxes 2.1 and 2.2).

Figure 2.10 Illustration of a helical CT image reconstruction. Data are acquired as a helical set of volumetric data. The width of the helical band represents the slice thickness. The CT image (red box) is reconstructed in the transverse plane from the helical data set (black c02uf001 ), by extrapolating from and weighting of existing data at similar locations along the helix (green c02uf002 and blue c02uf003 ). The higher the pitch, the more stretched the helix, the more imprecise are these calculations, resulting in image blur.

Box 2.1

Box 2.2 Examples

Single-slice CT:

2.5 mm slice width and 2.5 mm table increment = collimator pitch of 1

2.5 mm slice width and 5 mm table increment = collimator pitch of 2

Four-slice CT:

2.5 mm detector width and 7.5 mm table increment

= detector pitch of 3

= collimator pitch of 0.75

2.5 mm detector width and 10 mm table increment

= detector pitch of 4

= collimator pitch of 1

The stretch of the helix in Figure 2.10 illustrates the magnitude of the pitch and its relationship to slice thickness. The width of the helical band represents the selected slice width or collimation width. A pitch of zero would result in a complete ring of data, a sequential CT scan. With a pitch of 1, the helix is stretched to a degree that after one rotation the table will have moved exactly by one slice or collimation width. The sequential plane CT image is then reconstructed with mathematical interpolations from available data obtained at other positions of the helix. The higher the pitch, the more stretched is the helical data set and the more incorrect are these mathematical guesses, resulting in a blurred image.

Pitch Selection in Single-Slice CT (Figure 2.11)

In single-slice CT units the optimal compromise between the coverage of the object and the image detail is a pitch of 1.4. If more image detail is required, a pitch of 1 should be selected. Pitches of less than 0.75 result in a diminishing gain of image quality. If maximal image quality is required, then a sequential CT scan should be performed. A pitch of 2 is the maximum that is advisable for a single-slice CT unit; beyond this the image blur becomes too dominant.

Figure 2.11 Examples of different pitch selections (otherwise identical settings) and their effect on image quality of sagittally reconstructed cervical spinal CT images acquired with a single-slice CT unit. (A) Sequential CT scan (pitch = 0). (B) Helical series with a pitch of 1 introduces slight image blur, reducing the visibility of the intervertebral