14th CT users group meeting: 04/10/2012

The 14th meeting of the CT Users Group was held in Edinburgh on 04/10/2012. The programme is shown below with links to pdf version of some of the talks.

Please note: information provided in the slides is not peer-reviewed, is for educational use only and is explicitly not to be used for sales or marketing purposes. Any of the authors can be contacted, via the CTUG if no contact information is provided in the slides, to discuss the contents.

« Previous meeting	–	Meetings Index	–	Next meeting »

Meeting Programme

Automatic Exposure Control

09:30 +/- New designed phantom to test CT ATCM systems - Supawitoo Sookpeng, Colin Martin and David Gentle - Gartnavel Royal Hospital

Since the automatic tube current modulation (ATCM) system is now available on all CT scanners and used in majority of patient, failure to test them is to omit a major component of the image system. At the present time, there is not a standard phantom for the routine quality control of CT scanners ATCM operation. With the ATCM system, the tube current is modulated while scanning according to the patient size, shape and attenuation, although different CT manufacturers work on slightly different basis. The ideal phantom for testing these systems should be capable of evaluating how tube current and image quality as well as dose vary according to changes in patient size and shape. ImPACT developed a phantom for ATCM system assessment, that is a conical in shape, increasing in diameter in the x and y axis with the ratio of 3:2, which approximates to the Abdomen. The phantom has been used by several studies. However, phantoms of this type are expensive to manufacture.
For this study, two new designed phantoms have been developed. The concept of designing is to reflect the ATCM performance in varying dimension along the possible length and shape of the human body. The first phantom comprises of five elliptical sections of different dimensions. The advantage of a phantom of this type is that it is significantly less expensive to manufacture compared with the ImPACT phantom. There is, however, a limitation to the use of the phantom when testing the ATCM systems, as the abrupt change in attenuation provokes a typical ATCM response for some scanners. A second phantom has been developed from the results of the first which provides more effective of the ATCM system and a similar model may be suitable for use in our routine testing.
The study will show assessments of ATCM systems in terms of the dynamic changes in tube current and the image noise using new phantoms as well as evaluations of the new phantoms. The study was performed using three different CT scanner manufacturers; Toshiba, GE and Philips and the results were compared with ones obtained using the ImPACT phantom.

09:55 +/- Optimisation of the Philips CT automatic exposure control system - Tim Wood - Hull & East Yorkshire Hospitals

Introduction: The purpose of this study was to develop a technique for optimising the automatic exposure control (AEC) system on three Philips CT scanners (two Brilliance 16 and one Brilliance 40). This included matching the dose and imaging performance of the DoseRight Automatic Current Selection (ACS) system, and the longitudinal and rotational modulators (Z-DOM and D-DOM).
Method: The Rando anthropomorphic phantom was scanned using appropriate clinical settings to define the ‘user determined image noise reference’; the ACS system uses this to determine the maximum mAs that is required to produce clinically acceptable images of real patients, based on the results of their topogram. This technique replaced the ‘automatic patient size averaging’ system that is the default option on these types of scanner, but was extremely difficult to optimise due to the way in which it is always evolving. The response of the system in terms of radiation dose and image noise was then assessed as a function of ‘patient size’ with a purpose built AEC phantom. The performance of each scanner was then compared to ensure performance was well matched across the clinically relevant range of phantom thickness, and adjustments made to the ACS settings where necessary.
Results: Despite the complexities of the AEC implementation, introduced through the use of a ‘user determined image noise reference’ (as opposed to a single image quality metric, such as standard deviation), it was possible to match the performance of these systems through the use of simple phantoms (Figure 1). The use of the Rando anthropomorphic phantom allows easy adjustments to be made to the reference image for the purposes of optimisation (e.g. reducing mAs), as it does not suffer the significant inter-patient variations associated with the use of real clinical images. Several limitations of the AEC system have also been identified through a detailed analysis of the image noise properties.

Figure 1: Optimisation of the ‘Chest only’ protocol on a 16 slice Philips Brilliance CT scanner. Prior to setting up the ACS system this scanner used a fixed mAs technique (light blue line), but after using the Rando phantom to define the ‘image noise reference’ (dark blue line), performance is now well matched to a 40 slice system setup in the same way (magenta line). (a) A plot of radiation dose (CTDIvol) and (b) image noise (as defined by the standard deviation) as a function of phantom width.

10:20 +/- CT AEC techniques in PET/CT scanning - Gareth Iball, D Tout and H Williams - Leeds Teaching Hospitals NHS Trust and Central Manchester Hospitals NHS Foundation Trust Background: Modern PET/CT scanners include 64 slice CT systems which are equivalent to the diagnostic CT scanners offered by the major CT manufacturers. As such, tube current modulation systems are now used for the vast majority of PET/CT scans that are performed on such systems. The quality of CT images produced by the scanner, and therefore the doses delivered vary according to whether the CT images are used for attenuation correction only, attenuation correction and localisation or for full diagnostic purposes.
In order to compare the tube current modulation techniques of the three major PET/CT manufacturers a collaborative study of patient doses and image quality was undertaken.
Method: Routine patient dose audits for standard sized patients (60-80kg) were undertaken for “eyes-thighs” scans on current PET/CT scanners from three manufacturers. The three scanners involved were: GE Discovery 690, Philips Gemini TF and Siemens Biograph mCT. The patient weight range was subsequently extended so that the full range of patient weights was covered and the patient sample size was thus increased up to 70-100 patients per system
In order to compare the effect of the tube current modulation systems on patient doses, dose information was recorded from scanner workstations, PACS and directly from DICOM headers for 70-100 patients of all sizes. The following information was recorded for each patient: weight, age, gender, average CTDIvol, DLP, average mA (where available) and scan length. In addition to these parameters a measurement of patient size was obtained by measuring the AP and lateral dimensions of the patient on an image at the middle of the liver. In order to compare the effect of the tube current modulation system on image quality the standard deviation of the CT number in the liver was measured on the same image that was used for the patient size measurement. The mA that was used to create this image was also recorded.
An additional, direct, comparison of the AEC systems was made by scanning the same RANDO phantom on all three systems and reading out the mA profile along the patient with the AEC systems activated.
Results: For all three systems patient dose showed a stronger correlation with patient weight than with patient size. The variation in dose with patient weight was very different for the three manufacturers with the GE system showing the largest variation and the Philips system the least - the Siemens system laid between the other two. In terms of image quality the GE system delivered approximately constant image noise for all patient sizes whereas for the Philips and Siemens system the image noise was lowest for small patients and increased with increasing patient size. The combination of these results demonstrates the major differences in the way that the AEC systems function. The mA profiles that were obtained on the RANDO phantom further demonstrated these significant differences.
Conclusions: The CT AEC systems from the three major PET/CT manufacturers yield significantly different tube current modulation patterns and as such deliver different DLPs, organ and effective doses as well as delivering different levels of image quality across the range of patient weights. Users should be aware of how their system works and of steps that could be taken to optimise imaging protocols on all three systems.

10:45 +/- A study of CT dose distributions in an elliptical phantom and the influence of automatic tube current modulation - Colin Martin, Supawitoo Sookpeng, and David Gentle - Gartnavel Royal Hospital

Phantoms used for CT dosimetry take the form of Perspex cylinders of standard dimensions. These provide a useful basis for comparisons of CT output, but such phantoms are not a good representation of the human trunk.
An elliptical Perspex phantom has been constructed with axis dimensions of 330 mm and 220 mm similar to those for the trunk of an average patient. Holes for dosimetry measurements have been included at 10 mm from either end of the major and minor ellipse axes, and at the centre. Measurements have been made with Gafchromic film and ion chambers to compare distributions of dose within the elliptical phantom and a 320 mm diameter cylindrical phantom. Measurements of dose profiles for single rotations have been use to derive results for the CT dose index (CTDI). Results from multiple rotations have been combined to simulate helical scans and derive values for the cumulative doses near the mid-points of the phantoms. Measurements have been made on CT scanners manufactured by Toshiba, GE and Philips. The influence of the automatic tube current modulation (ATCM) systems in the x-y plane of the scan have been studied.
For all scanners, measurements at the lateral peripheries of the elliptical phantom were similar to those in the cylindrical phantom. However, doses at the anterior (top) periphery and the centre of the elliptical phantom were substantially larger than the equivalent in the cylinder, with the peak dose at the centre of the ellipse being about double that at the centre of the cylinder with the Toshiba scanner. The higher dose values occur for two reasons. Firstly the anterior peripheral measurement points in the ellipse lie nearer to the isocentre where the attenuation of the fan beam by the central portion of the bow-tie filter is less. Secondly the attenuation of the thinner elliptical phantom in the anterior posterior direction is lower. The differences between the phantoms are dependent on the CT scanner type with results for the Toshiba Aquilion 64 being the largest. The reason for this appears to be because the central region of the fan beam profile for this scanner is narrower than for the other scanners. When the ATCMs are brought into play, these tend to equalise peripheral doses at the anterior and lateral positions. This has a greater influence on the distributions within the Toshiba scanner than those of the other manufacturers.
Measurements of dose within an elliptical phantom show differences in dose distribution that are not apparent from measurements with a cylindrical phantom. They demonstrate how the shape of the bow tie filter affects dose distributions for different scanners and enable the effect of the ATCM on dose distribution to be assessed.

CT and atomic numbers

11:10 +/- Estimating iodine concentration from CT number enhancement - Rosemary Eaton, A Shah, J Shekhdar - Mount Vernon Hospital

Background: We have undertaken preparatory phantom work for a project that will use dynamic contrast enhanced CT to assess the malignancy or otherwise of solitary pulmonary nodules. Iodine contrast agent will be injected intravenously. The CT number in a region of interest over the lung nodule will be measured pre-contrast and at 1 minute intervals post-contrast for several minutes. The maximum measured CT number will be used as an indicator for malignancy. Previous trials have shown that enhancement of > 15 HU at 120 kV is correlated with malignancy.
This paper will outline our investigations into using a lower kV in order to achieve greater CT number enhancement for the same iodine concentration, and thus better noise statistics.
Method: The correlation between CT number and iodine concentration was measured at 80, 100 and 120 kV. Measurements were made in air and surrounding a water phantom.
Lung nodules were simulated by introducing small volumes of low concentration iodine into an anthropomorphic thorax phantom. CT number enhancement over baseline (water) was measured for three nodule locations (centre of lung, heart border, near rib) at 80, 100 and 120 kV. Measurements were repeated for a ‘large’ phantom, obtained by surrounding the thorax phantom with saline bags.
Results: Greater enhancement in CT number was measured at lower kV. The gradient of CT number against iodine concentration decreased by around 20% when measured next to a water phantom relative to that measured in air.
Anthropomorphic phantom measurements showed that different CT numbers were measured for the same iodine concentration, depending on the location of the region of interest within the thorax. This was particularly acute for the phantom representing a small patient, and at low kV (80 kV).
Conclusions: CT number enhancement over baseline can potentially be used to determine iodine concentration in vivo. However, beam hardening effects mean that calculated concentrations are likely to have poor accuracy for small patients, particularly if low kV is used. The CT number to iodine concentration coefficient must be measured in an appropriate phantom to mitigate these effects as far as possible.

11:35 +/- Materials and methods employed to validate a CT scanner’s approximation of effective atomic number - Robert Loader - Plymouth Hospitals NHS Trust (Derriford Hospital)

Rapid KVp switching Gemstone Spectral Imaging (GSI) dual energy CT, utilising material decomposition and monoenergetic extrapolation, can estimate the effective atomic number of a user defined region of interest (ROI) within a body. Materials and methods are conveyed illustrating techniques to validate the scanner determined effective atomic number for a range of materials utilising a GE HD750 series CT scanner. The basic theory behind material decomposition, the GE definition of effective atomic number, experimental validation techniques and possible clinical uses will be described for this introductory glimpse of quantitative CT.

Iterative reconstruction and dose

13:25 Iterative recon - that means lower doses, right? - Nicola Macdonald

13:50 +/- An investigation into the effectiveness of various advanced CT methods in the reduction of metal artefacts - Matthew Dixon - Plymouth Hospitals NHS Trust (Derriford Hospital)

Metal artefacts are a well known issue in many areas of CT, with scans involving stents, coils and artificial joints representing a significant majority. The theory behind both typical artefacts and advanced CT methods which can potentially reduce them is described and a summary of the results acquired using a specifically designed phantom presented. This will involve the use of a GE HD750 series CT scanner with the use of Model Based Iterative Reconstruction (VEO), Advanced Statistical Iterative Reconstruction (ASIR), rapid kV switching dual energy scanning (Gemstone Spectral Imaging) and high definition scanning.

14:15 +/- EPI-CT: International Epidemiological Paediatric CT Study. Estimates on organ doses and ideas on optimisation in paediatric CT: Work in Norway - TS Istad, EG Friberg, B Toft, Jonathan Turner, A Liland, W Ali, K Kjærheim, HM Olerud - Radiation Protection Authority, Østerås, Norway, Gjøvik University College, Norway, Norwegian University of Life Science, Norway, Centre for Health Informatics, City University London, UK Cancer Registry of Norway, Oslo, and University of Oslo, Norway

Introduction: The “Epidemiological study to quantify risks for paediatric computerized tomography and to optimise doses” (EPI-CT) was set-up by the International Agency for Research on Cancer (IARC) to investigate the relationship between the exposure to ionizing radiation from CT scans in childhood and adolescence and possibly attributable late health effects. We will discuss the work being performed in Norway by the Norwegian Radiation Protection Authority (NRPA) and the Norwegian Cancer Registry (NCR), working in partnership. The NCR has responsibility for the epidemiological methods, establishment of cohort, data collection from RIS, and data analysis and interpretation. The NRPA has responsibility for data collection from PACS, calculation of radiation doses, data analysis and interpretation, optimisation of paediatric CT and dissemination of results.
We will particularly focus on data collection, calculation of radiation doses and optimisation.
Data Collection and Dose Calculation: PACS and RIS systems are well established at all hospitals in Norway and contain information on radiological examinations performed over the last 10 years (PACS) and 20 years or more (RIS). Information stored in these systems allows us to calculate an estimate of the radiation doses delivered for each examination. For this we are using the newly developed PerMos software (Research Centre Henry Tudor, Luxembourg). PerMos performs automatic harvesting of dosimetric data from the DICOM header by querying PACS records, and uses the new NCI-CT software (National Cancer Institute, USA) for calculation of organ doses based on paediatric phantoms of different ages and sizes. For older records that do not exist on PACS, an estimate of radiation dose will be made from RIS records and knowledge about CT-scanners and protocols used in the pre-PACS period. A survey for older scanners and scan protocols is being performed as part of the EPI-CT project.
Epidemiological Study: Radiation doses for patients who were 20 years old or less at the time of their CT examination will be collected. This information will be matched with records from the Norwegian Cancer Registry for statistical analysis. The Norwegian cohort currently consists of more than 30 000 children, and we expect to include more.
Optimisation: As part of the EPI-CT study, we aim to perform optimisation of paediatric CT examinations. For this work, we plan to manufacture and test a paediatric version of the new ICRU dosimetry phantom (John M. Boone, ICRU committee on CT Image Quality and Patient Dosimetry) and scan this with current paediatric CT protocols for the range of current CT scanner models. Software will be developed to automatically evaluate image quality and dose. This part of the work is a collaboration between the partners in Norway and Luxembourg.

Managing CT scanner data

15:10 +/- Transferring protocols from a GE LS16 to a Siemens Definition Flash - Laurence King - Royal Marsden Hospital, London

Aim: To set up protocols on a new Siemens Definition Flash that give comparable doses to well established protocols on the GE LS16.
Method: An initial dose audit was carried out on the LS16 to determine typical CTDIvols for selected protocols. A range of Perspex phantoms were scanned on the LS16 and the CTDIvol was recorded. During commissioning testing of the Definition flash we characterised the available mA modulation curves using the same Perspex phantoms. We then selected the mA modulation settings that gave us the closest behaviour to that on the LS16. Finally we set clinical protocols with the CT Superintendent that gave us the expected CTDIvol for a 75 kg reference patient. Once the scanner went clinical we carried out a dose audit and compared the results to those from the LS16 to confirm that the scanners were matched.
Result: The new protocols were well matched to existing protocols across all patient sizes.

15:35 +/- Quality control in computed tomography by automated monitoring of key performance indicators - Patrik Nowik, R Bujila, H Andersson and C Jonsson - Karolinska University Hospital, Stockholm, Sweden

Background: The purpose of Quality Control (QC) in Computed Tomography (CT) is to verify that a CT system delivers expected image quality with acceptable patient doses. International guidelines and national (Swedish) regulations state that QC should be performed at least once annually on all CT systems. The condition of a CT system during the period between annual QC is unknown by medical physicists in terms of measurable quantities because it is not feasible to perform routine QC on a CT more than once a year using conventional methods.
Purpose: The purpose of this work was to analyze the relationships between QC parameters currently obtainable during annual QC of a CT system and to identify key performance indicators. Key performance indicators pertain to measurable or determinable parameters that are influenced by other parameters, where a stable key performance indicator implies that the underlying parameters are stable as well. Furthermore, to develop a method of performing QC on CT systems by automatically monitoring key performance indicators on a daily basis and a method of troubleshooting a CT system when a key performance indicator deviates, in order to systematically locate the cause of the deviation. The aim of this project is to supplement the currently established QC methodology by making it possible to follow a CT scanners condition between annual QC tests.
Material and Methods: A literature study of QC parameters in CT and the relationship between them, obtainable in a manufacturer’s QC phantom, was made in order to identify key performance indicators. 2 scan protocols were developed, 1 for the monitoring of key performance indicators and 1 for the systematic localization of deviating underlying parameters when a key performance indicator is out of tolerance. The key performance indicators were verified with phantom studies and the results were compared to theory. An application that can automatically analyze phantom images sent to a DICOM receiver and report the results was developed. A pilot study was commenced where a scan of the manufacturer’s QC phantom was made by CT technologists and sent to the developed application on a daily basis. The pilot study remains ongoing.
Results: The key performance indicators include noise, uniformity, CT numbers and positioning. Experimental results show that the key performance indicators compare well with theory and that a deviating underlying parameter can be found systematically. The testing procedure is straightforward and simple and can be made with little effort by CT technologists.
Conclusion: QC of a CT scanner by automatic monitoring of key performance indicators on a daily basis is a powerful tool that can be used to supplement established QC methodologies. Medical physicists, or other concerned parties, are able to obtain an indication of the current and historical (trends) status of a CT system with little effort so that actions can be taken directly to ensure the quality of CT examinations and patient safety.

16:00 Collecting digital dose data. Very nice, but how can I do it? - Ed McDonagh

16:25 Differences in CT numbers under different scan parameters and the impact on routine QC - Patrice Burke

16:35 CT AEC characterisation and optimisation using a noise-power spectra analysis framework - Tim Wood