PREFACE

The following definitions apply for the purposes of the present publication:

CT number

Number used to represent the mean x-ray attenuation associated with each elemental area of the CT image. Numbers are normally expressed in terms of Hounsfield units (HU). Measured values of attenuation are transformed into CT numbers using the Hounsfield scale:

where μ is the effective linear attenuation coefficient of the measured material relative to water for the utilised x-ray beam. The CT number scale is defined so that water has a value of 0 HU and air a value of −1000 HU.

(1) Modern CT scanners employ multiple rows of detector arrays allowing rapid scanning and wider scan coverage.

(2) All new CT systems have multiple detectors with a single or dual x-ray source, and a number of new dose reduction tools have become available commercially.

Back to Article Outline

1.1. Background 

(3) Computed Tomography (CT) technology and its clinical applications have shown enormous resilience against alternative diagnostic methods and, at the moment, the use of CT continues to rise. Current technology provides high power x-ray tubes, enormous computing power, multichannel detectors to give sub-millimetre slices with wider scan coverage and faster rotation times to complete one rotation in one-third of a second. CT now has an important role in dynamic applications such as cardiology and three-dimensional imaging of vascular and musculoskeletal anatomy.

(4) A number of terms are used for this technology, namely multi-detector-row computed tomography (MDRCT), multi-detector CT (MDCT), multi-detector-array spiral CT, multi-channel CT, and multi-slice CT (MSCT). The number of simultaneous but independent measurements along the patient long axis is often referred to as the number of ‘slices’, and this value is commonly used to represent the technical capabilities of a system (e.g., 64-slice MDCT). In this report, the Commission has chosen to use the terminology MDCT when referring to the technology generically, and 64-MDCT when referring to a specific technical implementation of MDCT.

(5) In 2000, ICRP published ‘Managing Patient Dose in Computed Tomography’ (ICRP, 2000a). At that time there was an urgent need to focus the attention of radiologists, physicians, medical physicists, and other personnel involved in CT on the relatively higher doses to organs in individual patients, increasing frequency of CT examinations, changes in clinical applications, and the increasing contribution of CT to the collective dose. Further, the technology in use dominantly utilised a single row of detectors (SDCT), permitting scanning of only a single slice at a time in either a discrete (sequential acquisition) or continuous fashion (spiral acquisition). Multiple detector rows along the z axis (longitudinal axis of the patient, i.e., head to toe) permit simultaneous scanning of more than one slice. MDCT was in its infancy at the time of the 2000 report (ICRP, 2000a) and thus there was only brief mention in the report of its impact on radiation dose. The data and experience were insufficient to make any evaluation. In the following years there was a phenomenal increase in use of MDCT, and technology has been advancing very rapidly to move from 4 slices to 8, 16, 32, 40, and 64 slices. Furthermore, dual-source CT has been recently made available and a 256-slice MDCT system will soon be available. The improved speed of MDCT scanning has also meant new applications (cardiac CT, whole body scanning) as well as improved patient throughput and workflow. In the past two decades, the use of CT scanning has increased by more than 800% globally (Frush, 2003). In the United States, over the period of 1991 to 2002, a 10–20% growth per year in CT procedures has been documented (Fox, 2003). Also in the United States during this period, CT scanning for vascular indications has shown a 235% growth, followed by a 145% growth in cardiac applications. An increase has also been demonstrated in abdominal (25%), pelvic (27%), thoracic (26%), and head and neck (7%) applications (Fox, 2003). With 64-slice MDCT, a further substantial increase is expected in cardiac applications. A 10% annual growth in the global CT market was reported in the year 2002 and this trend seems set to continue.

Back to Article Outline

1.2. Introduction to MDCT Technology 

(6) MDCT systems are CT scanners having more than one detector array, or row, in the arc around the patient. In CT, the detector array consists of 600 to 900 detector elements and corresponds to one transaxial plane. In SDCT, there is one such row of detectors. The ‘multi-detector-row’ nature of MDCT scanners refers to the use of multiple detector arrays (rows) in the longitudinal direction (that is, along the length of the patient lying on the patient table). MDCT scanners in most cases utilise CT geometry in which the arc of detectors and the x-ray tube rotate together. All MDCT scanners use a slip-ring gantry, allowing spiral acquisition at rotation speeds as fast as 0.33 second for a full rotation of 360 degrees of the x-ray tube around the patient. A scanner with two rows of detectors (Elscint CT Twin) had already been available since 1992 when several manufacturers introduced MDCT scanners with four detector rows in 1998. The primary advantage of these scanners is the ability to scan more than one slice simultaneously and hence use the radiation delivered from the x-ray tube more efficiently (Fig. 1.1). The time required to scan a given volume could thus be reduced considerably. The number of slices, or data channels, acquired per axial rotation continues to increase, with 64-detector systems now being widely available (Flohr et al., 2005a, Flohr et al., 2005b). It is likely, in the coming years, that even larger arrays of detectors having longitudinal coverage per rotation > 4 cm will be commercially available. Preliminary results from a 256-detector scanner (12.8 cm longitudinal coverage at the centre of rotation) have already been published (Mori et al., 2004). Further, an MDCT system with two x-ray sources is now commercially available (Flohr et al., 2006), signalling continued evolution of CT technology and applications.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1.1. 

  Schematics of detector elements in the arc around the patient, and detector rows perpendicular to the z axis. All CT scanners have many detector elements in the arc around the patient. However, only MDCT systems have more than one row perpendicular to the z axis.

(7) MDCT scanners can also be used to cover a specific anatomic volume with thinner slices. This improves the spatial resolution in the longitudinal direction considerably without the drawback of extended scan times. Improved resolution in the longitudinal direction is of great value in multiplanar reformatting (perpendicular or oblique to the transaxial plane) and in three-dimensional (3D) representations. Spiral scanning additionally allows overlapping data sets – which improve multiplanar reconstruction and 3D image quality – to be reconstructed without additional radiation dose to the patient.

Back to Article Outline

1.3. Differences between SDCT and MDCT 

(8) One essential difference between SDCT and MDCT is how the thickness represented by an image, or slice, is determined. For SDCT, slice thickness is determined by a combination of pre-patient and post-patient collimation. Therefore, the dimension of the detector array along the longitudinal axis can extend beyond the anticipated width of the x-ray beam or image slice (Fig. 1.1), i.e., the detector width is greater than the beam width. For MDCT, the converse is true and the x-ray beam width must be large enough to allow irradiation of all ‘active’ detector rows, i.e., all those being used for a particular scan acquisition; slice thickness is instead determined by the widths of the individual active detector rows used for image reconstruction.

(9) In Fig. 1.1 the single-detector row CT (SDCT) system on the left has one detector row, which is perpendicular to the longitudinal axis (indicated by z). Within this one row are many detector elements in the arc around the patient. On the right the MDCT system has 16 detector rows, which are all perpendicular to the z axis. For SDCT, the width of the detector (relative to the centre of the gantry) is 20 mm, although the pre-patient collimated beam is generally no more than 10 mm. Thus, in SDCT, the detector is wider than the x-ray beam, ensuring that the entire primary x-ray beam is detected. The multiple-detector-row CT (MDCT) system on the right has 16 detector elements each 1.25 mm along the longitudinal axis for each of the approximately 900 positions around the patient. The width of the detector is also 20 mm at the isocentre. Depending on the scanner model and collimation used, this may result in some portion of the primary radiation beam not being detected. This overbeaming phenomenon is described in Section 3.2.1.

(10) The four data channels in 4-MDCT allow the acquisition of four simultaneous slices, of 1.25, 2.5, 3.75 or 5 mm width. Wider image thicknesses (2.5 mm, 5 mm, 10 mm) are generated by electronically combining the signal from more than one of these rows. Therefore the image thickness used for the purposes of image review often differs from the slice thickness used for data acquisition. In this document, the term ‘slice thickness’ always refers to that used for data acquisition (slice collimation).

(11) Owing to the narrow width of the detector rows and the use of third-generation geometry, gas ionisation detectors are no longer used for MDCT scanners. In order to generate an image of a 1-mm slice of anatomy, detector rows of not much more than 1 mm in width must be used (detector dimensions are normalised relative to their coverage at the centre of the CT gantry).

(12) Another design for an MDCT detector array is illustrated in Fig. 1.2. When small slices are desired, only the central portion of the array is used. It is therefore not necessary to have narrow rows in the outer portions of the array. The wider detectors at the periphery allow simultaneous acquisition of four slices each of 5 mm thickness. This design is somewhat less expensive and geometrically more efficient.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1.2. 

  Diagram of the detector geometry used in a MDCT from two major manufacturers. The detector array is 20 mm wide along the longitudinal (z) axis and uses eight detector rows of varying widths to allow simultaneous scanning of 4 slices up to 5 mm thick.

(13) Currently, MDCT systems are capable of acquiring up to 64 slices simultaneously along the z direction (Fig. 1.3). Three of the four manufacturers use 64 rows of either 0.625 mm or 0.5 mm detectors. The fourth manufacturer uses 32 rows of 0.6 mm detectors and oscillates the focal spot to acquire 64 overlapping slices (Flohr et al., 2005b). This results in the reduction of spiral artefacts and improved spatial resolution along the longitudinal axis (Flohr et al., 2005b).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1.3. 

  Diagram of the detector geometries used in 64-MDCT from four major manufacturers. The Siemens 64-MDCT uses 32 sub-mm detectors and a moving focal spot to achieve 64 overlapping slice measurements.

(14) For sequential or axial data acquisitions (e.g., the table is stationary during the rotation of the x-ray tube around the patient), each channel collects sufficient data to create one ‘slice’ or image, so as many as 64 independent images along the z axis could theoretically be reconstructed. For narrow slice widths, geometrical ‘cone-beam’ considerations may limit the number of allowed images per rotation to less than 64. For example, one manufacturer’s 16-MDCT scanner allows only 12 data channels to be used in sequential scanning because of cone beam considerations (Flohr et al., 2005a, Flohr et al., 2005b).

(15) The primary attribute of MDCT systems is not the number of physical detector rows, but the number of slices that are acquired simultaneously. The speed needed to cover a given volume is improved by a factor equivalent to the number of slices included in the scan simultaneously. The reason why the number of simultaneous slices was initially limited to four was the amount of data to be acquired and transferred simultaneously. At that time, engineering and cost considerations limited the systems to four simultaneous data collection systems. Additionally, cone beam artefacts were not severe in 4-MDCT systems, but as the number of simultaneous slices increased, these artefacts become more problematic using conventional fan-beam reconstruction methods. Once 3-D cone-beam reconstruction algorithms (or advanced fan-beam algorithms with cone-beam corrections), and the increased computational power needed for these algorithms, became available, 8- and 16-MDCT scanners were introduced.

(16) The advent of spiral CT initially introduced an additional acquisition parameter into the CT vocabulary, pitch. Pitch is the ratio of the distance of table travel per x-ray tube rotation to the x-ray beam width. With MDCT, a significant amount of confusion was introduced regarding the definition of pitch, as some manufacturers used an altered definition of pitch that related the distance of table travel per x-ray tube rotation to the width of an individual data channel, resulting in calculated pitch values of 3 or 6. The International Electrotechnical Commission CT Safety Standard re-established the original definition of pitch (distance of table travel normalised to the total beam width) as the only acceptable definition of pitch (IEC, 2002, McCollough and Zink, 1999). This definition of pitch conveys the degree of overlap of the radiation beam: a pitch of 1 indicates contiguous radiation beams, a pitch less than 1 indicates overlap of the radiation beams, and a pitch greater than 1 indicates gaps between the radiation beams.

(17) Two manufacturers report the tube current–exposure time product (milliampere second – mAs) as the average mAs per unit length along the longitudinal axis, called either effective mAs or mAs/slice, and calculated as actual mAs/pitch. This distinction between mAs and mAs per unit length is important because, as the pitch is increased, scanner software may automatically increase the mA such that the image noise (and patient dose) remains constant with increasing pitch values (Flohr et al., 2003a, Flohr et al., 2003b, Mahesh et al., 2001). When the effective mAs or mAs/slice is displayed, the user may be unaware that the actual mA changes when the pitch value is changed. On other MDCT systems, the mA value is automatically adjusted to the value that will keep image noise constant as pitch or slice width is changed, and the selection box is turned orange to alert the user to the change in the prescribed mA value.

Back to Article Outline

1.4. Upcoming developments 

(18) Recently, cone-beam computed tomography (CBCT) has been introduced that employs a large-area detector with approximately 1000 or more detector ‘rows’ (compared to 16–64 rows in current clinical MDCT systems). CBCT with one or two flat-panel detectors is a developing imaging modality. There is no true cone beam commercial system available yet. The available reports from experimental systems indicate its usefulness in intraoperative imaging, interventional radiology, bone and lung imaging, mammography and radiation therapy (Siewerdsen et al., 2005, Daly et al., 2006, Ross et al., 2006, Guerrero et al., 2006, Glick et al., 2007). The advantages include decreased scan time, wide z-axis coverage and higher near-isotropic spatial resolution. Current weaknesses of the experimental platform arise from lack of integration with an in-room navigation system, difficulties with scatter rejection, the presence of artefacts, the use of fixed collimation at the output of the x-ray source, and inaccuracy of the dosimetry method using a standard 100-mm-long CT chamber (Ross et al., 2006, Siewerdsen et al., 2005).

(19) Volumetric imaging of the whole breast has been demonstrated by a number of investigators using a flat panel detector for CBCT breast imaging (Glick et al., 2007, Kwan et al., 2007, Shaw et al., 2005). The framework for determining the optimal design and combination of parameters was the patient dose criteria, i.e., mean glandular dose (MGD) constrained to that given for a typical two-view conventional mammography study (Glick et al., 2007). The application of CBCT in dental and maxillofacial imaging is also increasing (Guerrero et al., 2006, Sukovic, 2003). Some studies promoting dose reduction have been published (Tsiklakis et al., 2005, Ludlow et al., 2003).

Back to Article Outline

1.5. What is the motivation for this report? 

(20) With ICRP Publication 87 (ICRP, 2000a), an editorial in the British Medical Journal (Rehani and Berry, 2000), and the February 2001 issue of AJR (AJR, 2001), considerable attention was focused on the topic of dose management in CT. Two papers addressed the lack of appropriate parameter selection in paediatric CT examinations (Paterson et al., 2001, Donnelly et al., 2001). Further, Brenner et al. (2001) reported on the potential risk of cancer induction from the use of CT in the paediatric population. These publications note that the use of CT has significantly increased in children (for clinically valid reasons), but they warned that this increased usage carries with it a potential for excessive exposure to radiation and an increased risk of cancer in the paediatric population. In the editorial by Lee F. Rogers in the same issue of AJR (Rogers, 2001), he stated: ‘sorry to say, but kids get overlooked’. These reports aroused media attention and the clinical and radiological protection communities recognised that radiation doses in CT should be more carefully scrutinised. The number of publications on radiation exposure in CT, and management thereof, has been steadily increasing. Manufacturers now put more emphasis on radiation exposure reduction and improved image optimisation in addition to scan time reduction. In recent years, improved management of radiation dose in CT has been high on the agenda for all CT manufacturers.

(21) In 2005, the Commission realised that essentially all new CT systems are MDCT, and that a number of new dose reduction tools have become available commercially. Thus, to address these new tools, the continued growth in CT applications, and the consequent growth in the contribution of CT to the collective dose from medical applications, it was decided to update ICRP Publication 87 (ICRP, 2000a). In addition to reviewing these technology changes in CT, a number of issues will be addressed, such as:

(22) As in its previous report (ICRP, 2000a), the primary audience for this document is imaging professionals: radiologists, cardiologists, operators, medical physicists and researchers involved in patient dose management. However, this document provides reference material that may be useful for referring physicians, physicians who may own a CT scanner, regulators and national authorities, manufacturers and hospital administrators.