Gating in small-animal cardio-thoracic CT.

1
Gating in small-animal cardio-thoracic CT Soenke H. Bartling, Jan Kuntz, Wolfhard Semmler Dr. med. Soenke H. Bartling Dipl. Ing. (FH) Jan Kuntz Prof. Dr. Dr. Wolfhard Semmler Department of Medical Physics in Radiology, German Cancer Research Center (DKFZ) Corresponding author: Dr. Soenke H Bartling Address: Department of Medical Physics in Radiology (E020) German Cancer Research Center (DKFZ) Im Neuenheimer Feld 280 69120 Heidelberg Germany Email: soenkebartling@gmx.de Phone: +49 151 58585585 Fax: +49 6221 42 2613

2
Abstract
Gating is necessary in cardio-thoracic small-animal imaging because of the physiological motions that are present during scanning. In small-animal computed tomography (CT), gating is mainly performed on a projection base because full scans take much longer than the motion cycle. This paper presents and discusses various gating concepts of small-animal CT, and provides examples of concrete implementation. Since a wide variety of small-animal CT scanner systems exist, scanner systems are discussed with respect to the most suitable gating methods. Furthermore, an overview is given of cardio-thoracic imaging and gating applications. The necessary contrast media are discussed as well as gating limitations. Gating in small-animal imaging requires the acquisition of a gating signal during scanning. This can be done extrinsically (additional hardware, e.g. electrocardiogram) or intrinsically from the projection data itself. The gating signal is used retrospectively during CT reconstruction, or prospectively to trigger parts of the scan. Gating can be performed with respect to the phase or the amplitude of the gating signal, providing different advantages and challenges. Gating methods should be optimized with respect to the diagnostic question, scanner system, animal model, type of narcosis and actual setup. The software-based intrinsic gating approaches increasingly employed give the researcher independence from difficult and expensive hardware changes.
Keywords : Small animal, gating, intrinsic, extrinsic, retrospective, prospective, heart, lung, thorax, computed tomography

3
Introduction X-ray computed tomography (CT) is an important technique in cardio-thoracic small-animal imaging, because it provides high-contrast images of structures in the chest such as lung tissue, bones, contrast-enhanced heart chambers and blood vessels [1-7]. The same holds true for most pathologies related to the chest organs. In contrast to most other areas of the body, the physiological movement of organs in the chest is a challenge for small-animal CT imaging [2, 8]. Scan concepts known as cardiac and respiratory gating can overcome this. Gating can improve image quality and be used to acquire four-dimensional (4D) data sets. In this article, gating concepts in small-animal cardiac and respiratory gating are discussed, and various implementations described. Gating-relevant design aspects of small-animal CT scanners are presented. First, we provide a brief overview of applications of small-animal cardio-thoracic imaging with references to further literature. A concrete implementation example is also given.

4
Applications of cardio-thoracic small-animal imaging
Cardio-thoracic CT imaging of small animals is used to assess disease or disease models with respect to the lung, heart [6, 9-12] and vessels [10] within the chest [3, 13, 14]. Lung fibrosis and lung emphysema can be quantified using CT [3, 15]. Since lung metastases and tumors provide a high contrast to lung parenchyma, they can be imaged [16, 17], counted and measured. In addition, high contrast is also seen on CT with other changes to the lung parenchyma such as pneumonia or infiltration for other reasons such as oedema. Effusions within the pleural or pericardial space can be well detected using CT. In most cases it is necessary to use a contrast medium to assess the vessels of the chest. An exception to this is imaging of coronary or aortal calcifications which are better imaged without contrast media.
Contrast media Since most small-animal CT imaging takes much longer than clinical CT imaging, special small-animal CT contrast media must be used. These provide a longer blood-pool time, which means that their clearance from the blood by the kidneys and liver takes much longer than does that of standard iodinated contrast media [18-20]. Some can provide vessel and heart chamber enhancement lasting more than 1 h. However, by injecting standard CT contrast media repetitively and/or continuously, it is still possible to achieve a relatively long and somewhat constant blood-pool enhancement. The exact timing of injections must be figured out individually for every CT scan protocol. Maximum injection volumes certainly represent a limitation for the use of

5
standard iodinated contrast media. Their use is, however, also attractive from a cost perspective, since the blood-pool contrast media currently commercially available are expensive. Certain flat-panel based, small-animal scanner concepts do exist that can acquire a full CT data set in a time that is acceptable for bolus standard CT contrast media [3, 21]. Here, 4D perfusion data could be acquired to provide functional information about organ perfusion [3]. A combination of gating with perfusion imaging has not been described so far in small-animal CT imaging.

6
Gating in cardio-thoracic imaging Gating almost always plays a role when cardio-thoracic imaging is performed because there is movement in almost all chest (and upper-abdomen) organs in freely breathing animals [8]. Image quality can be better with gating than without (Fig. 1) [1, 2, 8, 9, 22-27]. Gating may also make it possible to visualize certain pathologies that would not otherwise be visible. As an example, metastases that are close to the diaphragm can be mixed up with the diaphragm (or rather with the liver under the diaphragm) in non-gated data [2], but they become clearly visible in gated data sets. Additionally, gating reduces the blurring at the edge of pathological structures. Quantitative measurements such as size might, therefore, be more exact with gating than without. Another aspect of gating is the 4D data that can be acquired, which allows assessment of movements (Fig. 2) [24, 27]. In this way, cardiac function data can be acquired such as ejection fraction and heart chamber volumes in diastole and systole. Similarly, in 4D respiration data sets, the maximal difference in lung volumes at different respiratory states results in the respiratory tidal volume. 4D data sets have been used to assess movement of the bronchial tree and the lung in various methods of respiration and ventilation. The use of intubation, ventilation and a ventilation stop during scanning can obviate the need for gating in small-animal cardio-thoracic CT – at least with regard to the effects of respiratory motion [2, 28, 29]. Intubation is, however, complex and increases preparatory work, while at the same time putting the animal at considerable risk. Intubation is, therefore, not considered to be a good alternative in follow-up or high-throughput studies. However, if for some reason the animal must be intubated anyway, stopping respiration during CT might provide good image results without gating. Furthermore, using intubation and stopping respiration allows the lack of

7
reproducibility of organ movement during motion cycles to be overcome and the resolution may be better than in gated scans of freely breathing animals.
Theoretical considerations Gating is necessary in thoracic imaging to prevent a loss of image quality and to acquire 4D data sets. The fundamental reason for this is that the acquisition of CT image data mostly takes much longer than the duration of a respiratory and/or a cardiac phase. Image quality is then restricted by the motion of the heart and lung, just like sports photographs taken with an over-long exposure time. The time required to acquire CT data must be short in comparison to a cardiac or respiratory cycle. Physical and engineering limitations mean that full data acquisition cannot be accelerated to meet the exposure time requirements of cardiac or respiratory movements. This is especially stressed in small-animal CT imaging, because both the heart and respiration rate are higher here than in humans (e.g. the heart rate of a rat can be 400 min-1) [7]. On the other hand, the exposure time of single projections and thus the full scan time is long in small-animal CT. This is because a small X-ray focal spot limits the photon-flux. Therefore, a sufficient number of imaging photons to ensure a good signal-to-noise ratio in high-resolution small-animal CT data sets can only be reached by prolonging the exposure time [30]. For motion-compensating gating several prerequisites must be fulfilled: 1. The organ movement must be cyclic, and deviations of the organ movements above a certain degree can cause motion artefacts – even in gated scans. The organ movements in small animals can be considered cyclic – e.g. the variability of lung position at same phases between several respiratory cycles of mice is 100 µm [8].

8
2. The movement of the organs must be taken into account during scanning - some sort of motion signal must be acquired that relates to the organ movement. 3. The CT data acquisition must be divided into small, partial data acquisition portions that meet the exposure requirements of the cardiac and respiratory motion. These are usually single projections that can be acquired in the order of milliseconds. 4. Multiple projections of the organ showing the same movement situation must be acquired from a sufficient number of angular positions to ensure CT reconstruction. For 4D imaging, this has to be completed for several movement positions of the organ. The acquired motion signal is then combined with the data acquisitions to generate a gated still image and/or 4D data set. In the following, various gating concepts are introduced:
Prospective – retrospective gating concepts
Prospective and retrospective gating varies in terms of how the motion signal influences data acquisition and post-processing of the data. In prospective gating, data acquisition is influenced by the motion-gating signal [9, 22, 23, 25]. An example would be a scanner that waits for a certain heart phase to acquire a projection before it moves to the next angular position. In retrospective gating, data acquisition or scan is performed independently of any motion-gating signal [11, 24, 31, 32]. The motion-gating signal is taken into account during post-processing of the scan data. An example is a scanner that acquires data during continuous rotation. After the scan, the projection data are analyzed in combination with the motion-gating signal(s). Combinations of the prospective and retrospective methods also exist.

9
Amplitude-based gating - phase-based gating
In amplitude-based gating, re-sorting of the projections is modulated with respect to the amplitude of the motion signal. The acquired motion signal should somehow correlate with an aspect of the motion of the organ – this could be, for example, the deflection of the diaphragm in respiratory gating. In phase-based gating, rebinning is done with respect to the time phase between gating signals (Fig. 3) [6, 11, 22, 33]. The time phase between gating events can be defined absolutely (e.g. seconds) or relatively (e.g. percentage between gating signals) (Fig. 3). In amplitude-based gating, the modulation of the scan or the re-sorting is more robust in cases of deflective organ movements [25, 26, 31, 33]. Deflective organ movements occur during pathological breathing cycles. Similarly, cardiac arrhythmias may also cause deflection of the heart movement. Furthermore, amplitude-based gating may be advantageous if the resolution of the scanner system is close to the organ movement reproducibility. In this case, amplitude-based gating might take even slight variations of organ movement into consideration and might improve the image quality compared with phase-based gating. It is not necessary to have any quantitative parameter acquired during a motion cycle in phase-based gating. A binary motion signal is the minimal necessity, making phase-based gating easier to implement. The binary motion signal must occur reliably at a reproducible time point of the motion cycle. Respiratory gating is commonly used with amplitude-based gating, while cardiac gating is more commonly used in a phase-based gating manner. Phase-based gating is usually less disturbed by low- and high-frequency jitter in the motion signal.

10
In both phase-based and amplitude-based gating, the actual phase and amplitude range used for image reconstruction can be determined by trial and error. Additionally observation of the motion signal can be used to find useful time points. Time points for systole, diastole, maximal inspiration and expiration can then be determined by reviewing the 4D data sets after reconstruction. Selecting the gating phases largely influences image quality and time resolution. Wider reconstruction phases usually lead to better image quality in terms of signal-to-noise ratio – however, there are more motion-blurring effects. Good motion phases must be found for each scanner setup, animal model, and diagnostic question.
Intrinsic gating - extrinsic gating
In extrinsic gating, a motion signal is acquired through an external hardware system (Fig. 4) [1, 2, 22-25, 27-29]. For respiration, this can be a pneumatic cushion, a camera, a laser system or some other means of detecting the lung movements. The motion signal can even be a trigger signal delivered by a ventilator in invasively ventilated animals. For cardiac gating, a standard method is deduction of an electrocardiogram (ECG) signal. However, pulse-detection methods that work with an infrared metre, which is attached to the foot or tail of the animal can also be used (Fig. 4). Intrinsic gating methods do not need any additional hardware to derive a gating signal [6, 26, 31, 33, 34]. A gating parameter is derived from the acquired image data itself. A wide variety of post-processing algorithms for intrinsic gating have been described. The obvious advantage of intrinsic methods – the lack of additional hardware – is often offset by the complexity of its implementation: Intrinsic gating methods must be adapted carefully to the characteristics of individual scanner systems. They are

11
usually combined with retrospective gating, because continous motion data are mostly necessary to calculate a gating signal from the projection data.
Different animal narcosis types and their relevance to gating
Narcosis is necessary during animal scanning to prevent the animal from moving. Physiological parameters of the animal are significantly influenced by the type of narcosis. The exact influence depends on the type of narcosis and the actual substances used for it. In small-animal imaging, two types of narcosis are most widely used, namely, gas and injection narcosis. Both have their own advantages – gas narcosis is easier to control, while injection narcosis does not need any additional hardware to be considered in the scanner setup. However, more important is the fact that the actual respiratory motion and frequency as well as heart frequency is significantly influenced by the narcosis and with it the way gating can be implemented. Roughly speaking, gas narcosis (e.g. sevoflurane) reduces the respiratory and cardiac rates, while injection narcosis (e.g. ketamine) usually increases them. Gas narcosis also results in a kind of very regular gasping respiration with long phases of virtually no movement that are interrupted by short phases of deep respiration (Fig. 4). This makes gas narcosis ideally suited for (retrospective, phase-based) gating – at least for the acquisition of still images. On the other hand, injection narcosis leads to fast and irregular respiratory cycles making it well-adapted to amplitude-based gating. If 4D data are to be reconstructed, it must be taken into consideration that long phases of virtually no respiratory motion during the gasping in deep gas narcosis can be very disadvantageous, and the phases with useful inspiratory movements are relatively short. Therefore, a more shallow gas narcosis should be used for 4D respiratory gating.

12
Implementation of gating in small-animal CT imaging In the following, we describe aspects of implementating gating in a small-animal scanner system. The wide variety of small-animal CT scanner systems available and the need to adapt the gating method to the special requirements of each scanner system make a prior discussion of small-animal scanner systems necessary.
Small-animal scanner types In terms of gating, the major difference in small-animal scanner systems is the way in which they acquire projections at different angular positions (Fig. 5). These scanner types are: Type IA) CT scanner systems that acquire projection data in a continuous swipe over a certain angular distance (mostly 180°+∂, where ∂ may be the cone-beam angle) [2, 33]. Type IB) CT scanner systems that provide the option of a continuous scan lasting more than one continuous rotation [6, 11, 24, 27, 31, 34]. Type II) Scanner systems that acquire projections in an angular step-and-shoot technique. Here, several projections can be acquired at any angular position either sequentially or as single projections triggered through a signal. (If only one non-triggered projection is acquired at an angular position, the scanner must be regarded as a “slow”, type I scanner system in terms of gating!). After projection acquisition, the X-ray tube/detector system is then moved to the next angular position [22, 25, 26]. Some scanner setups provide both type I and type II operating modes.

13
Gating concepts in various CT scanner types All the scanner design concepts have their own advantages and disadvantages in terms of small-animal CT gating. Scanners of type I are more suitable for retrospective gating approaches, because it is hardly possible to modulate the scan with regard to organ movement once it has been initiated. Usually several swipes or rotations are necessary to ensure acquisition of a sufficient number of projections at enough organ positions. The projection data from several acquisitions is then re-sorted retrospectively with regard to the gating signal. In theory, prospective gating or scan modulation (such as in clinical scanner systems) can be implemented into these scanner types, but they have not yet been described and seem rather complicated compared with prospective gating approaches in type II scanner systems. In type II scanner systems, both retrospective and prospective gating approaches can be implemented. In the prospective approach, the scanner system waits for a certain movement position of the gated organs and acquires (“triggers”) a projection at an angular position before moving on to the next angular position. In the retrospective approach, a whole set of projections are acquired at an angular position, and the X-ray/detector is then moved to the next angular position. The projections at each angular position are then selected with regard to the motion-gating signal.
Post-processing image data in gating Three-dimensional and four-dimensional reconstruction For gating in type I scanners, it is usually necessary to acquire more data than would actually be necessary for just one motion-gated, still-image, three-dimensional (3D)

14
reconstruction. This is because some projections are acquired in movement phases of the organs that are not useful for reconstruction. This usually means an excessive x-ray dosage compared with prospectively triggered type II scanners. However, this difference can be avoided if 4D data sets are reconstructed: in 4D reconstruction, all the data from type I and type II scanners are used for reconstructing images in several movement phases. The additional availability of functional data would then justify the additional dosage.
Incomplete or excessive data In gating, partially incomplete and/or excessive projection data are sometimes acquired. This occurs because, in retrospective gating setups, the projection and organ movement phases are not taken into account during scanning. “Excessive” projections are those that are acquired at the same or similar angular positions and during the same or similar organ motion phases so that more than one relevant projection may occur in some angular positions. Many excessive projections occur when rotation time and organ motion are more or less concurrent, and relevant projections are not well distributed over the full 180°+∂ reconstruction range. These excessive projection data should not be discarded, but incorporated in the reconstruction by averaging them with other projections at similar positions. In this way, all projection image information (and the dosage used for its acquisition) contributes to the image quality of the reconstructed scan. Projection data are incomplete if at one or more angular position no projection data is acquired at the relevant organ movement phase. These angular positions can be filled in by linearly (with regard to the angular position) interpolated projections from valid neighbouring projections.

15
Artefacts and potential deterioration of image quality might occur. Still by using the proposed post-processing steps optimally, and through careful adaptation to the scanner system and animal model, the gain in image quality outweighs any potential loss of accurate CT values in most cases.