26th CT users group meeting: 18/10/2024

The 26th CTUG meeting was held on the 18th of October 2024 in Exeter. The meeting programme is below, with abstracts for the talks (click the +/- button to expand abstracts). pdfs of some of the presentations can be opened by clicking the links.

The meeting was preceded on Thursday 17th October by the second CTUG training course, at the same venue, following a similar programme to the last training meeting in 2022.

Please note: information provided in the slides is not peer-reviewed, is for educational use only and is explicitly not to be used for sales or marketing purposes.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.

MEETING PROGRAMME

Session 1 – Anthropomorphic Phantoms and New Technologies

09:50 +/- Experience of the Kyoto anthropomorphic phantoms in CT - Ruby Fong - Barts Health NHS Trust Barts Health NHS Trust carried out a series of CT scanning in May 2024 using Kyoto anthropomorphic phantoms, PBU-60 (CT Whole Body Adult), PBU-70 (5y paediatric whole body) and PBU-80 (Newborn Whole Body Phantom). Physicists and radiographers worked together and the phantoms were scanned using clinical protocols for a selection of common CT examinations on a variety of scanners at different hospital sites within the Trust. Eight scanners of 5 different models at three hospital sites were involved. The aim was (i) to establish how harmonised the scan protocols and doses were, (ii) to compare the resultant dose indicators with diagnostic reference levels, (iii) to explore the effects of different scanning techniques and (iv) to evaluate the effects on image quality and dose by changing scanning parameters. The outcome of these endeavours feeds into our ongoing optimisation programme.
Results revealed some interesting discoveries. For some examinations, CTDIvol and DLP were broadly similar while for some other examinations, significant differences showed up between scanners within the same site. Work has begun to address the differences found, working with radiographers and radiologists.
The presentation will recount our experience with using the Kyoto phantoms, present some scan results and explore further work.
The anthropomorphic phantoms have allowed the same “patient” to be scanned by the same operator at each hospital site using clinical protocols. While the phantoms are not cheap, they can be valuable tools in aiding Physics and Radiology to determine the current situation and to collaborate in harmonising and optimising CT scan protocols.

10:10 +/- Optimising CT head protocols with the Kyoto PBU-60 phantom - Adedotun Adedokun - Imperial College Healthcare NHS Trust CT dose audits were completed on three GE scanners across two sites, including two GE Revolution EVOs and a GE Discovery 750HD using Bayer Radimetrics and DoseTrack dose management systems. The median DLP from CT head examinations were found to vary between the two GE Revolution EVOs and feedback from a neurology consultant also indicated there were concerns with image quality on the GE EVO with the lower median DLP. A review was therefore conducted to identify the differences in radiation dose across the three scanners and to identify opportunities to improve image quality.
An investigation into the discrepancy between the scanners was first conducted by examining the acquisition and reconstruction parameters employed across the three scanners. A Kyoto PBU-60 head phantom was then used to acquire images with varying acquisition and reconstruction parameters for neurology consultants working across the three scanners to review and score. A scoring matrix for consultants to use was created after reviewing published guidelines and feedback from a neurology consultant.
This work presents the initial results of the optimisation project and our future plans to achieve optimisation across the three scanners.

10:30 +/- Comparison of Image Quality Between Photon Counting and Energy Integrating CT for Prostate Artery Embolisation Planning with Detectability Index (d’) - Angus Fraser - Oxford University Hospitals Foundation Trust Photon-counting CT is a rare example of a true technological leap, offering significant improvements in both image quality and dose efficiency. On top of performance gains, its spectroscopic capabilities make existing state-of-the-art techniques look primitive in comparison, offering true spectral separation and does not rely on assumptions about two distinct, overlapping spectra in dual energy CT.
Given the novelty of the technology, its potential is far from being realised, both in terms of new applications, and performance in existing ones. For example, virtual monoenergetic image (VMI) reconstructions are becoming increasingly prevalent, and their utility may even lead to them becoming the default method of reconstruction in the future. Given the intrinsic spectral sensitivity of photon-counting CT, it excels in VMI, and would likely play a key role in its continued proliferation. It is therefore necessary for physicists to consider whether standard practice for image quality assessment is adequate for this new paradigm.
Another consideration is for applications where CT was used sparingly, or not at all, due to inadequate dose or image quality performance. There is a clear pattern in the literature showing that many of such applications are continually being shown to be viable with photon-counting CT. One such example that we are working on is prostate artery embolisation (PAE) planning. Its use provides a number of potential benefits for the patient, but not without complicating factors which make it difficult [1].
PAE is a minimally invasive treatment for benign prostatic hyperplasia (BPH) with reported benefits of lower risk of complications and sequelae than surgical resection [2]. PAE is a technically challenging procedure that demands a highly skilled operator with precise knowledge of the patient’s prostatic vasculature [3]. Currently, the procedure is planned through contrast enhanced CT angiography. Visualising the arteries of interest is particularly challenging for CT systems due to their small size, their varied origin, and typical comorbidities of patients with BPH [4]. Recently, patients have been directed to a novel photon-counting CT scanner for PAE planning as its technical specifications indicate that it is particularly well suited for this task.
This project will involve the assessment of image quality for already acquired PAE planning CT images on a qualitative and quantitative basis. The chosen metric for quantitative assessment of image quality is detectability index (d’) for visualisation of small blood vessels in and around the pelvis, and normalised for contrast concentration in the aorta. Noise Power Spectrum (NPS) and Task Transfer Function (TTF) will be measured using the iodine insert of a Mercury 4.0 phantom. Calculation of d’ will be carried out using David Platten’s detectability index plugin for ImageJ. For the identification of anastomoses to the prostatic artery, PAE planning CT has been shown to have excellent specificity of 94.2%, but relatively poor sensitivity of 59% [1]. In principle, this should mean that d’ for small blood vessels in and around the prostate should serve as an excellent, clinically relevant, driver for optimising image quality in PAE planning CT.
References
[1] D. Maclean et al., “Planning Prostate Artery Embolisation: Is it Essential to Perform a Pre-procedural CTA?,” CardioVascular and Interventional Radiology, vol. 41, no. 4, pp. 628-632, 2018/04/01 2018, https://doi.org/10.1007/s00270-017-1842-7.
[2] “Prostate artery embolisation for lower urinary tract symptoms caused by benign prostatic hyperplasia.” NICE.
[3] B. Moulin, M. Di Primio, O. Vignaux, J. L. Sarrazin, G. Angelopoulos, and A. Hakime, “Prostate Artery Embolization: Challenges, Tips, Tricks, and Perspectives,” Journal of Personalised Medicine, vol. 13, no. 1, p. 87, 2023.
[4] A. Kobe, G. Puippe, E. Klotz, H. Alkadhi, and T. Pfammatter, “Computed Tomography for 4-Dimensional Angiography and Perfusion Imaging of the Prostate for Embolization Planning of Benign Prostatic Hyperplasia,” Investigative Radiology, vol. 54, no. 10, pp. 661-668, 2019, https://doi.org/10.1097/rli.0000000000000582.

Session 2 – Clinical Optimisation and Image Quality

11:20 +/- Evaluation and optimisation of low dose neck CT for SPECT - Louise Giansante, Jan Taprogge, Abigail Glover, Iain Murray, Elly Castellano - Royal Marsden NHS FT Context
Routine dose surveys for various procedures are conducted in our department approximately every three years to review and update Local Diagnostic Reference Levels (LDRLs) as needed.
During a recent dose survey focused low-dose CT procedures for SPECT, we identified that representative doses for low-dose neck CT scans were high, exceeding the National Diagnostic Reference Levels (NDRLs) by 30% for CTDIvol and 50% for DLP.
Method
A meeting was organised with Nuclear Medicine physicists and a Nuclear Medicine consultant to discuss the dose levels and potential issues. It was determined that there was no justifiable reason for the representative doses to exceed NDRLs, and a multidisciplinary optimisation task was put in place.
After confirming that the correct protocol had been used and that dose indicators could be optimised for the required clinical objective, we undertook the following steps:
· Analysed patient dose reports and images to assess tube current modulation across the scanned body area;
· Conducted phantom work using both cylindrical and anthropomorphic phantoms to evaluate the impact on CTDI vol when LAT and AP topograms were used;
· Performed further tests with anthropomorphic phantoms, considering and changing various parameters such as organ characteristic, kV, topogram directions, pitch, rotation time, and quality reference mAs to assess the impact on dose indicators.
The optimised protocol was then applied to all workflows involving low-dose neck CT for SPECT procedures.
CTDI vol and DLP values were then periodically monitored (every three months) to ensure dose levels were acceptable. Image quality was also monitored by Nuclear Medicine consultants.
Results and Conclusions
One of the issues identified in practice was that these neck CT scans often extended beyond the neck to include areas down to the diaphragm, and occasionally the pelvis. Additionally, these scans were not solely used for localisation and attenuation correction; radiologists still required diagnostic image quality for the lung portion of the scan.
After about one year, we successfully reduced the representative dose indicators to align with the published NDRLs, achieving the initial project goal. Image quality was still deemed acceptable and adequate for the required clinical task.
References:
Iball, G.R. et al. A national survey of computed tomography doses in hybrid PET-CT and SPECTCT examinations in the UK. Nuclear Medicine Communications. 2017;38(6):459 to 470

11:40 +/- Radiation dose and image quality audit and optimisation in CTCA - Andrew Shah - East and North Hertfordshire NHS Trust Computed Tomography Coronary Angiography (CTCA) is a front line test for coronary artery disease. Its clinical use is expected to grow following changes to national guidance in 2016 recommending its use for detection of coronary artery disease on patients with stable chest pain. The first national audit of CTCA practice in the UK showed significant variation in average radiation dose between approved scanners for use in CTCA. This study explores optimisation of two scanners capable of single heartbeat CTCA acquisition with the goal of reducing exposure to ionising radiation whilst maintaining diagnostic image quality. Optimisation of parameters was via multi-disciplinary consensus and multiple protocol changes were made with small changes to dose at each iteration. Detailed audit of radiation dose was undertaken following each protocol change with statistical comparison between audits. The GE Revolution and Canon Aquilion One Genesis wide detector systems were included in the audit
Image quality was assessed by two reporting radiologists and cardiologists for each scanner. Images from forty patients on each scanner were retrospectively evaluated and scored for image noise and motion artefact. Twenty patients were scanned on the initial protocol, and twenty on the final protocol on each scanner as part of standard of care.
Radiation dose was significantly reduced whilst maintaining diagnostic image quality for all patients. Optimisation was implemented via cautious changes to CTCA scan technique following several multi-disciplinary reviews of radiation dose and scan parameters and comparison to the literature and manufacturer recommendations.
Information on the next national CTCA dose audit will also be shared.

12:10 +/- Using iDose to optimise image quality in radiotherapy planning CT scans - Adam Frankland - The Royal Wolverhampton NHS Trust Radiotherapy patients will undergo a planning CT scan to identify tumour position and organs at risk. The planning CT scan is also used to obtain electron density information for dose calculation in the treatment planning system.
iDose is an iterative reconstruction algorithm on Philips CT scanners for removing image noise. It ranges from level 1 to level 7, where level 1 removes a small amount of noise and level 7 removes the most noise but can cause images to look “plasticky”. This research assessed if changing the iDose level can allow for a reduction in patient dose while retaining current CT image quality.
Head and neck, thorax and pelvis CT protocols were investigated. Image quality was assessed by scanning a Catphan phantom and measuring contrast-to-noise ratios and noise power spectrums for the CT number linearity and image uniformity modules, respectively. The clinical protocols were adjusted to reduce the tube current and increase the iDose level.
Standard treatment plans were created using anthropomorphic phantom images to compare point dose and dose-volume histogram information. Anonymised patient plans were recalculated with different iDose level images. Patient images were analysed by MVision AI-based contouring software to evaluate how iDose level impacts autocontoured structures.
Mean pixel values for Catphan inserts were consistent within ± 1 HU for all levels of iDose. When going from iDose level 3 to level 4, it is possible to lower the tube current by 10% with no statistically significant difference in contrast-to noise (p<0.05). By increasing from level 3 to level 5, the tube current can be reduced by 30% without any significant difference in image quality. For all tested phantom and patient plans, all dose variations because of changing iDose level and tube current were well below 1%. No adjustments to the CT number-electron density calibration in the planning system were required. Slight deviations in autocontour volumes were observed when adjusting iDose reconstruction level, but this was deemed not large enough to significantly change planned dose distributions.
The outlined optimisation methodology can be used regardless of vendor or reconstruction algorithm. Propose new CT protocols using iDose level 4 and a lower tube current to reduce patient dose by 10% with non-inferior image quality compared to current clinical protocols. This research has been used to implement a new CT protocol for HDR gynae brachytherapy. Further studies required in visually grading images with clinicians and investigating 4D CT protocols

Session 3 – Metrics of Patient Dose and Image Quality

14:00 +/- Optimisation of abdomino-pelvic CT protocols: presentation of a recently published comprehensive and up-to-date inventory of pertinent metrics for the perusal of the clinical medical physicist - Eric Pace - University of Malta Objective
Patient-specific clinical protocol optimisation requires consideration of body region (in this case abdomino-pelvis) and a judicious selection of metrics to represent body habitus (BH), image quality (IQ), and risk/dose (RD). This work set out to make available to the clinical medical physicist a comprehensive and up-to-date inventory of such metrics based on an extensive review of the relevant literature.
Method
A Pubmed search was carried out with scope delimited to the imaging of adults and the period 2010–2024. The search words used were: ‘comput* tomography’, ‘CT’, ‘abdom*’, ‘dose’, ‘risk’, ‘SSDE’, ‘image quality’, ‘water equivalent diameter’, ‘size’, ‘body composition’, ‘habit*’, ‘BMI’, ‘obes*’, ‘overweight’. Inclusion criteria for identified metrics were: for BH, should be easily measurable either directly on patient or pre-scan radiograph or reconstructed image, and being a potential predictor of IQ or RD; for IQ, the metrics should be measurable on the final reconstructed image and in the spatial domain (since this reflects what is available to the radiologist and radiographer); for RD, metrics should be calculable solely from the DICOM header and reconstructed image data. For both IQ and RD, the metrics should at least in principle be automatable.
Results
An inventory of metrics from the three categories was established. 11 BH metrics were identified and organised as being either regional or global: patient weight, waist circumference, waist to hip circumference ratio, sagittal and lateral diameters, patient cross sectional area, area of circumscribing ellipse, effective diameter, ellipticity ratio, water equivalent diameter, body mass index; Nine IQ metrics were identified and categorised as measurements of noise, contrast or sharpness: noise standard deviation, Tian and Samei noise, global noise level, local task-based autocovariance, contrast to noise ratio, local task-based line profile, margin sharpness, image blur metric, and structure sharpness index. Six RD metrics were identified: dose length product, size specific dose estimate, individual organ dose, effective dose, risk index, relative effective dose.
Conclusion
An inventory of metrics from the three categories relevant to patient-specific clinical protocol optimisation is presented. It would seem the current general consensus at present favours the use of water equivalent diameter for BH, global noise level for generic IQ and size specific dose estimate for RD. Nonetheless, the purpose of the inventory is intended to present other alternatives to the clinical medical physicist for possibly improved patient-specific protocol optimisation. Such optimisation should ideally be targeted at addressing the image quality criteria for specific clinical tasks provided by either radiology colleagues or internationally used guidance documents.

14:20 +/- Comparison of methodologies for calculating CT effective doses for unintended and accidental diagnostic procedures - Tolulope Falola - Somerset NHS Foundation Trust According to the UK Health Security Agency, about 16% of the average UK radiation dose is acquired from medical radiation with CT being the largest contributor to this figure in diagnostic imaging.
An informal census in the UK shows that different Trusts use different methodologies to calculate the estimated effective doses of accidental and unintended CT exposures. This project compares the estimated effective doses using three methodologies.
· ImPACT calculator.
· ‘Updated estimates of typical effective doses for common CT examinations in the UK following the 2011 national review’ by Shrimpton et al.
· How Radiology Works calculator.
It highlights the variation in the methods and the effect of these variations on the categorising of these incidents per the guidance published by CQC under the Significant Accidental and Unintended Exposure (SAUE) codes.
This project also highlights the discrepancies in the individual methodologies used. For example, the various versions/modifications of the ImPACT calculator and the different tables used to select the conversion factor for the effective dose when using the paper published by Shrimpton et al. titled ‘Updated estimates of typical effective doses for common CT examinations in the UK following the 2011 national review’.

14:40 +/- Use of novel image quality metric the Global Noise Index to develop image quality reference levels – should we be performing image quality audits? - Katherine Baker - NHS Lothian Motive
Current image optimisation work is often guided by results from patient dose audits and clinical user requests. As a result, scanners which may benefit from optimisation may not be highlighted from dose audits alone [1].
The primary aim of the project is to implement a previously developed technique for obtaining measurements of an innovative image quality metric, the Global Noise Index (GNI) [2], to phantom and clinical images. The secondary aim of the project is to evaluate the contribution that measurements of GNI obtained using the software might make to current CT optimisation work, and whether it can be a useful tool to develop image quality audit methodology in tandem with patient dose audits.
Method
This project makes use of in-house written software to determine GNI for each CT scan. The GNI software was validated using synthetic and phantom images and the utility and limitations of the image quality metric explored.
Anthropomorphic phantom images and then clinical images were analysed using the GNI software from scanners which were already the subject of optimisation work. Values of GNI were then obtained for scanners and protocols for comparison, as parallel metrics to audit alongside dose audit measurements [3].
Discussion
The GNI proves a simple to obtain image quality metric using the automated software [4]. It gives an objective measure of image quality and can be used to compare scanners and protocols, directing optimisation work.
We hope to develop this work further to implement GNI as part of an image quality audit programme.
[1] Alsaihati, N.; Ria, F.; Solomon, J.; Ding, A.; Frush, D.; Samei, E. Making CT Dose Monitoring Meaningful: Augmenting Dose with Imaging Quality. Tomography 2023, 9, 798-809. https://doi.org/10.3390/tomography9020065
[2] Christianson O, Winslow J, Frush DP, Samei E. Automated Technique to Measure Noise in Clinical CT Examinations. AJR Am J Roentgenol. 2015 Jul;205(1):W93-9. https://doi.org/10.2214/AJR.14.13613 PMID: 26102424.
[3] Ria F, Davis JT, Solomon JB, Wilson JM, Smith TB, Frush DP, Samei E. Expanding the Concept of Diagnostic Reference Levels to Noise and Dose Reference Levels in CT. AJR Am J Roentgenol. 2019 Oct;213(4):889-894. https://doi.org/10.2214/AJR.18.21030 Epub 2019 Jun 10. PMID: 31180737.
[4] Solomon JB, Li X, Samei E. Relating noise to image quality indicators in CT examinations with tube current modulation. AJR Am J Roentgenol. 2013 Mar;200(3):592-600. https://doi.org/10.2214/AJR.12.8580. PMID: 23436849

Session 4 – QA Techniques - Dual Energy and Tube Current Modulation

15:30 +/- Construction of a phantom for dual energy CT quality assurance tests - Anne Hill - University Hospitals Bristol and Weston NHS FT Dual energy CT (DECT) uses two energy levels (typically 80 and 140kVp) to acquire two separate image datasets with differing attenuation profiles. Material properties (such as effective atomic number and density) can be derived from the two datasets. A variety of applications are available. For example, it is possible to analyse material composition, quantify the amount of contrast agent such as iodine in organs and lesions, generate material density iodine maps and create virtual mono-energetic images.
A GE Apex CT scanner was installed and commissioned in North Bristol NHS Trust in June 2024. This is the first scanner in the Bristol Trusts with dual energy capability.
Several dual energy quality assurance tests were devised with reference to relevant literature and a recent CT Users Group presentation on dual energy CT testing.
Some of the tests used existing test objects: CTDI Perspex, Catphan and GE standard water phantom. However, it was decided that a dedicated DECT phantom would be required for a fuller characterisation of the dual energy capability of the scanner.
After reviewing the market, it was decided to construct a DECT phantom in-house with the aim of using it to assess material quantification accuracy for iodine contrast agent, as well as CT number accuracy for images reconstructed across the range of available mono-energetic X-ray energies.
The cylindrical phantom (approximately 10cm diameter) housed five inserts, all with a diameter of 3cm and each containing a different concentration of iodine in water. Iodine concentrations were selected, after a literature review, on their applicability to current clinical applications of DECT. They ranged from 0.5 to 15 mgI/mL. The inserts were evenly spaced within a deionised water background.
The iodine inserts were filled with different combinations of left over iodine contrast agent (Omnipaque, concentration: 300 mgI/mL) and deionised water. Errors in the resulting nominal iodine concentrations were estimated to range from 1.2 % for the 15 mgI/mL insert to 1.8% for the 0.5 mgI/mL insert.
The measurements made at commissioning using this phantom indicated that it was a useful and cost effective additional tool for assessing the dual energy capability of a DECT scanner. Results demonstrated a linear relationship between measured and nominal iodine concentrations. Measured values were similar to nominal ones and within published tolerances. For each iodine concentration, CT numbers plotted against kV showed a similar trend to National Institute of Standards and Technology (NIST)-derived nominal values.
This presentation covers methods of phantom construction including analysis of the errors in iodine concentrations, how it was used in practice and a discussion of the results acquired at commissioning.

15:50 +/- Developing a Phantom for the Evaluation of Iodine Quantification on Philips Spectral CT Scanners - Alexandra Palmer - Barts Health NHS Trust Objective
This study aims to develop a phantom to evaluate SDCT iodine quantification that can be used for developing a routine quality control procedure to assess the accuracy of iodine quantification and the limits of detectability.
Materials and Method
An acrylic phantom was developed to assess the Philips SDCT ability to accurately measure and detect iodine concentrations. The phantom was filled with Omnipaque iodine solution mixed with distilled water. The lower concentration limits were defined between 0 – 1 mg/ml. Higher limits of 2, 5, 10, and 15 mg/ml solutions were also measured. The developed phantom was also compared to a solid water phantom using a typical CCTA cardiac protocol to assess how the choice of phantom effects iodine density measurements. The solid water phantom also investigated the variation of phantom size and insert diameter. An abdomen protocol was then used to assess the variation in kV, choice of iodine map, and consistency through the phantom on iodine quantification.
Results
The results of the preliminary study showed a linear relationship between the measured iodine densities compared to the nominal iodine densities, with an R2 > 0.99, and with the measured iodine density compared to conventional Hounsfield units. There was also no statistical significance (p>0.05) between the iodine measurements for 100 kV, 120 kV, and 140 kV with the same delivered dose.
From the phantom comparison study, the solid water phantom size affected the LOB with a result of 0.77 mg/ml for the small phantom and 1.44 mg/ml for the large phantom. The LOD was also greater for the larger solid water phantom and increased with decreasing insert diameter. The project phantom showed a greater percentage error in the iodine measurements compared to the solid water phantom, but despite this had a lower LOD (0.23 mg/ml) and LOB (0.28 mg/ml) than the solid water phantom.
Conclusion
This project has demonstrated a cheap method for developing a SDCT phantom that can be used to assess quantitative features and had a low LOD compared to a solid water phantom. The project phantom demonstrated that the accuracy of iodine density measurements were impervious to the variation in tube potential, but did affect the LOD and LOB, which was lowest at a 140 kV. In the phantom comparison study, Grueter’s phantom was able to assess the accuracy of iodine and found that was more accurate for larger diameter inserts and in smaller phantoms. The size of the phantom, even with the dose adapted to phantom size, affected the detectability and accuracy of iodine measurements

16:10 +/- Experiences of CT tube current modulation testing - Laurence King - Royal United Hospitals Bath NHS Foundation Trust A local protocol for routine testing of CT tube current modulation (TCM) performance has been developed and put into use at the Royal United Hospitals Bath NHS Foundation Trust. In this presentation, the development of this protocol will be described including phantom selection and use of automated image quality analysis routines. The local protocol will be compared to protocols described in literature including Iball et al, (2016) and report of AAPM task group 233 (2019). Example results of TCM performance from three CT vendors will be presented and the merits of each referenced protocols will be discussed. The results of tests carried out on a novel hybrid SPECT-CT system will be presented that demonstrates the importance of characterising CT TCM performance during commissioning of a new CT system.