27th CT users group meeting: 16/10/2025

The 27th CT Users Group meeting was held on the 16th of October 2025, at the Newcastle Civic Centre City Council Chamber. The programme is shown below, click the +/- button to see the abtract. Most presentations are availabile to view as a pdf by clicking the link on the presentation name.

Please note: information provided in the slides is not peer-reviewed, is for educational use only and is explicitly not to be used for sales or marketing purposes.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

CT USERS GROUP - ANNUAL CONFERENCE PROGRAMME

Session 1 – CT scanner performance evaluation

09:50 +/- Simplifying CT annual quality assurance tests - Rodney Padgett - Newcastle upon Tyne Hospitals NHS FT Higher physics workloads, higher patient demand on scanner time and several recent accidents with test objects have prompted the team at Newcastle to look at the current CT QA programme and try to reduce the required equipment (particularly the weight of the equipment) and scanner time for taken tests, whilst still delivering a useful annual QA service. Proposed changes, still under investigation, are presented. Reliability of modern scanners means less CTDI measurements are a possibility, and a single image quality (IQ) test is proposed which replaces the current IQ tests and associated suite of test objects.

10:10 +/- Design, Manufacture and Testing of 3D Printed Phantoms for CT Scanner AEC Testing - Nick Keat - Perceptive Discovery Automatic Exposure Controls (AEC) for the automatic selection of exposure parameters have been available on CT scanners for over 20 years. The ability to test, characterise and validate these has been limited by the availability of suitable phantoms for the assessment of the AEC system capabilities. This work describes the production of a phantom suitable for this purpose with an elliptical cross section that increases in diameter along the z-axis. The result is a fillable elliptical cone that approximates a relatively slim adult patient cross section.
The phantom was designed with Autodesk Fusion 360 software (see figure), using parametric dimensions that allow it to be re-sized and re-shaped as required. It was printed with maximum lateral and AP cross section dimensions of 208 to 134 mm (ratio 3:2) in PLA filament on a Bambulabs A1 Mini 3D printer (print volume of 180 x 180 x 180 mm). It has a mount to allow it to be hung off the end of a Phantomlabs Catphan box, or similar, and a prop to allow it to be supported on the patient bed.

Figure: phantom design shown in axial (left) and sagittal (right) directions
The phantom was tested on a Siemens Emotion 16 CT scanner that is part of a Biograph PET-CT scanner. Behaviour of the CARE Dose 4D software was investigated. The limited cross section of this phantom resulted in the Siemens CARE Dose 4D software reducing the tube current to minimum values for most of the phantom when performing body scans. For head examinations, the cross section was closer to the standard size, so the characteristics of the software could be better investigated.
A larger phantom was printed on an Ultimaker S5 printer (build volume of 330 x 240 x 300 mm), which allowed for a phantom cross section of 324 x 216 mm. This was better suited for body protocols, although it is still smaller than the Siemens reference body cross section. This
phantom was used to test the scanner, and the effect of changing various scan parameters and AEC characteristics was investigated.
Overall, 3D printing techniques allowed for the relatively straightforward and flexible production of fillable AEC test phantoms. It also enables the production of phantoms with different cross section characteristics. A 3D printer with a greater build volume would be required to produce a phantom with cross sections comparable to larger patient cross sections.

10:30 +/- Review of expected results for mA variation curves in routine CT AEC testing - Oliver Lawes - University Hospital Southampton NHS FT CT automatic exposure control (AEC) testing is performed using the nested CTDI phantom method [1] at UHS. To confirm AEC reproducibility, the variation of mA along the z-axis is graphed, and deviation from a baseline curve is qualitatively assessed by visual inspection. It was noted that GE scanners produced more variable results than other vendors’ systems, often resulting in a test fail. The aim of this project was to determine the cause of these fails: whether it was down to operator error in the exposure parameter selection, whether the test method was unsuitable for these scanners, or whether the scanner itself is unable to produce reproducible results.
To investigate this, repeatability tests were performed on a GE scanner, and a Siemens scanner which had consistently passed mA variation reproducibility tests. To determine if using an anthropomorphic phantom increased repeatability, a Kyoto Kagaku Lungman [2] phantom was additionally scanned using the same exposure parameters and setup as used for the CTDI phantom.
Quantitative analysis of repeatability results (percentage variation in mA across plateaus and entire curve) showed no significant variation between the GE and Siemens scanners, however visual inspection of the curves showed obvious differences. For Siemens scanners the mA variation curve matched across all repeats, while for GE the turning points matched in amplitude, yet the gradient and amplitude across plateaus showed variation across some phantom sections between repeats. The Lungman phantom did not improve repeatability and was disadvantageous for qualitative visual analysis due to the complex curve shape.
Guidance criteria for visual inspection of the mA curve based on curve characteristics was introduced into routine testing following these results which should reduce the number of false positive results in this test. Retrospective analysis of previous reproducibility tests was also carried out to assess the rate of false positive results in QC reports. This repeatability testing will be performed for Canon scanners in the future to determine similar criteria.
References
[1] Iball, G.R., Moore, A.C. and Crawford, E.J. (2016), A routine quality assurance test for CT automatic exposure control systems. Journal of Applied Clinical Medical Physics, 17: 291-306. https://doi.org/10.1120/jacmp.v17i4.6165
[2] https://www.kyotokagaku.com/en/products_introduction/ph-1/

Session 2 – Radiation dose to patients

11:20 +/- Patient-Specific Monte Carlo Simulation and Effective Dose Calculation for Chest Computed Tomography Using a Hybrid Phantom - Colin Lee - Clatterbridge Cancer Centre NHS FT Effective dose (ED) is a key dosimetric quantity, and an accurate value for it is often sought for optimisation, justification, and research purposes. Numerous methods are available for ED calculation; for application to a specific patient, the accuracy of many of these methods are limited by differences between the modelled phantom and real patient. Patient images generated by computed tomography scans can be used to generate a digital phantom to facilitate personalised Monte Carlo dosimetry. This phantom method inherently comprises the best representation of the patient geometry during the respective scan, however the digital geometry is then restricted to the imaged volume; this presents a challenge with regards to whole body dosimetry and computation of effective dose. This paper presents a phantom method combining real patient and reference phantom data to achieve a patient-representative whole-body model.
A CT x-ray source was constructed for Monte Carlo modelling, as a Phase Space object in GATE v9.0, and was validated against physical measurements throughout a standard CTDI phantom. A chest CT examination was simulated with several digital phantom geometries to ascertain the relative limitations and merits for a range of digital phantoms. An anonymised (standard-sized) patient CT image dataset, from a clinical whole-body nuclear medicine scan, was considered the gold standard geometry due to the guaranteed accuracy with regards to patient representation. The International Commission for Radiological Protection (ICRP) report 110 reference phantom was also used in 3 permutations: 1) its raw form, 2) with a CT couch added, and 3) with arms removed to simulate an arms-up patient position. Finally, a hybrid phantom was generated using real patient data in the chest image volume supplemented with slices from the ICRP reference phantom outside of the image volume.
Simulated doses throughout the CTDI phantom were generally < 10 % different to a physical measurement in the same position. ED for the real patient, ICRP phantoms, and hybrid phantom were 27.1, 24.6-26.6, and 26.0 mSv respectively; the armless ICRP phantom achieved the closest ED to the real patient dataset. Compared with the real patient data, the closest organ mean absorbed dose was achieved by the hybrid phantom for all key organs with the exception of ‘remainder’ and bone.
Simulation using the hybrid phantom successfully achieved the closest organ mean absorbed dose in most cases, and yielded an ED only -4% different to the real patient phantom value. The hybrid phantom method was considered a success in this case, however only one patient and hybrid phantom was assessed in the work presented here. Further work is required to investigate the variability within a population, and particularly with differently sized patients (and matched, clinically scaled reference phantoms). Although the armless ICRP phantom achieved the closest effective dose to the real patient dataset, the relative inaccuracy for each individual organ suggests that this likely would not be reproduced for a range of patient positions and habitus.
Keywords: Computed Tomography, GATE, Phase Space, Monte Carlo, GATE, Hybrid, partial body

11:40 +/- Automated extraction of DICOM heart rate information for coronary CTA audit - J. Dormand, E. McDonagh, E. Shaw, T. Semple, E. Nicholl - The Royal Marsden NHS FT Prospectively gated coronary CT angiography (CCTA) scans can be carried out using a dual source “turbo flash” high pitch spiral or a sequential axial scan technique. Turbo flash scans result in lower patient radiation doses but require low and stable heart rates to ensure diagnostic image quality. Auditing heart rate and heart rate stability of patients who undergo each CCTA scan mode is a useful technique to ensure optimal scan mode choice. We present a method of automated extraction of heart rate information from the DICOM header information of CCTA image data. A script to identify and collect relevant CCTA images from PACS and a script to extract relevant heart rate parameters from the DICOM headers of these
images were developed. CCTA data from 2019 to 2024 audited using this approach was used to analyse heart rate and heart rate variability for each scan mode over time. Analysis shows that percentage use of turbo flash scans is increasing and that average heart rate is decreasing, suggesting heart rate management and scan mode choice at the audited hospital are improving. Limitations of this method and avenues for future work are discussed.

12:00 +/- National CT coronary Angiography (CTCA) dose audit 2025 - Andrew Shah - Barts Health NHS Trust Introduction
CTCA is a front line test for coronary artery disease for which the radiation dose received by patients is dependent on patient habitus, heart rate and heart rate variability. Modern CT scanners are capable of delivering high quality diagnostic scans at a relatively low dose if they are well optimised for a range of patients.
The first UK national survey of CTCA doses was performed in 2014 which informed the current national DRLs. Since 2014, scanner technology has progressed significantly regarding CTCA scan technique which should allow for lower dose scans for a greater proportion of patients. The British Society of Cardiovascular Imaging (BSCI) re-audited CTCA doses in 2025 with an aim of updating the UK NDRLs and providing advice to the imaging community on scanner optimisation.
Methodology
The BSCI national audit was promoted to several professional groups, with encouragement for a collaborative approach to the data collection across radiologists, cardiologists, radiographers and medical physicists. Data was requested for CTCA exams over a consecutive 4 week period from 1st Nov 2024 to 31st Jan 2025 including scanner specification, scan and reconstruction settings, use of heart rate control medication and dose per patient scan.
Summary
Data for 105 CT scanners from NHS Trusts and private Hospitals was received across four CT manufacturers. There is a wide distribution of average doses, with a 12 fold difference in dose from the lowest and highest dose scanner. Scanners routinely achieving < 100 mGycm are all high specification CT systems i.e. wide detector, dual source, fastest rotation time. Higher doses > 300 mGycm were associated with older scanner technology or low specification systems. Third quartile values for different scan techniques will be presented

Session 3 – Clinical practice and optimisation

13:50 +/- Service evaluation of patient positioning in CT scans at Barts Health NHS Trust - B. Thomson, R. Fong, L. Bailey-Pioli, C. Ovenden, V. Mutanga, D. Dihanov, T. Faleye, A. Rose, Q. Munnee, K. Santhirababu, L. Tonks, D. Abbeyquaye, A. Palmer - Barts Health NHS Trust Introduction
Patient misalignment impacts received dose and image quality. The bowtie filter shapes the radiation beam so the thickest region of the patient receives the highest intensity of X-Rays, if a patient is positioned away from isocentre, a higher surface dose is delivered to thinner regions and image noise is increased in thicker regions [1]. If a patient is vertically misaligned to isocentre they will be positively or negatively magnified in the topo- gram, increasing or decreasing the delivered dose due to the Automatic Exposure Control (AEC) [1]. In addition the IR(ME)R (Amendment) Regulations 2024 references the need for particular care to be taken when positioning a patient [2]. For these reasons an evaluation of patient alignment to isocentre is being performed across Barts Health diagnostic radiology departments. The main aims of the project are (i) to establish a process (including the creation of custom tools) for future repetition of this evaluation and (ii) to evaluate alignment accuracy in A&E and Imaging departments.
Method
Approval was sought from Barts Information Governance and the Barts Clinical Effectiveness Unit for the use of patient CT scans. Anonymised adult patient ab- domen and head scans from the period 01/01/2025 - 31/03/2025 were collected from two hospitals (A & B). Data from Hospital B was extended to 30/04/2025 to increase Head sample size. Data for Hospitals A and B has been collected and processed up to this point. Scan analysis was done using custom python code (v3.9.13). Patient images were segmented based on Hounsfield Units (HU) and the centroid for each slice was defined as the mean pixel position of the ROI. Misalignment was defined as the median difference between patient centroid and image centre. Based on radiographers' typical practice, image centre and isocentre are aligned for head and abdomen scans, therefore image centre was assumed to be isocentre.
Results
Head results were excluded as misalignment was found to be minimal. Standard error of the mean is the quoted uncertainty. Sample sizes are 182 for Hospital A Imaging, 229 for Hospital A A&E and 327 for Hospital B Imaging. Figure 1 shows horizontal misalignment is minimal in comparison to vertical misalignment, Hospital B being the largest at -1.13±0.43mm. Vertical misalignment was largest at Hospital A, with A&E and Imaging having means of -17.4±1.1mm and -18.7±1.1mm respectively. Hospital B has a smaller mean misalignment of -12.1±0.8mm.

Figure 1: Violin plots of a) horizontal and b) vertical misalignment at Hospital A A&E and Imaging (orange) and Hospital B (blue). Box and whisker plots are overlaid in black, with mean and median marked.

Figure 2: a) The mean (red) vertical misalignment with slice location for abdomen scans at Hospital B, uncertainty (light blue band) is the standard deviation, overlaid on a projection of all vertical misalignments of scans. Entries were limited to slice locations with >20 measurements. b) Sagittal projection of a Kyoto phantom with centroid (red) and isocentre (blue) marked.
Figure 2a shows the variation in vertical misalignment is significant across the scan. When comparing the shape of the mean misalignment to the phantom centroid in Figure 2b, it can be seen there are significant drops in centroid position due to the shoulders (-100mm) and pelvis (-500mm).
Conclusions
The evaluation process was successful. Patients are posi- tioned systematically below isocentre; this is consistent with literature [3]. Patient anatomy partially contributes to this due to the changes in height at the pelvis and shoulders. Training will be required to improve vertical alignment in CT scans. There is no notable difference in patient positioning between A&E and Imaging depart- ments. This evaluation will be expanded to other hospi- tals in the Trust.
References
[1] Y. Al-Hayek, X. Zheng, C. Hayre and K. Spuur, 2022, DOI: 10.1016/j.jmir 2022.09.027
[2] The Ionising Radiation (Medical Exposure) Regulations 2017 (Latest version), UK Government, 2024. Schedule 3, Table 2.
[3] O. Akin-Akintayo, L. F. Alexander, R. Neill, E. Krupinksi, X. Tang, P. Mittal, W. Small and C. Moreno, 2019, DOI: 10.1067/j.cpradiol.2018.02.007

14:10 +/- Hands up if you need a chest CT! - Gareth Iball, Helen Adamson, Helena Czjetan - University of Bradford Background
It is standard practice for patients undergoing scans of their chest, or chest, abdomen & pelvis to have their arms raised above their head for all of the scan series. When automatic exposure control systems are used, the positioning of the arms above the head should help to keep the dose low. Having arms placed by the patient’s side would increase the dose and the likelihood of experiencing streak artefacts through the abdomen due to the presence of the high attenuation provided by the elbow bones.
The aim of this study was to assess changes in total examination dose, organ dose and image quality for chest, abdomen & pelvis scans undertaken with a variety of arm positions.
Methods
A Kyoto-Kagaku whole body anthropomorphic phantom was scanned on a FujiFilm Scenaria View CT scanner with the phantom’s arms positioned above the head, to mimic the standard scanning technique, and then in a number of alternative positions that could be used in clinical practice.
Scanner recorded dose metrics of CTDIvol and DLP were used to assess the changes in the overall dose for the scans with varying arm positions. An in-house developed python code was used to extract the average mA, calculate the water equivalent diameter and the global noise level for each image slice.
Results
The largest increase in examination dose (56%) was for the scan performed with the arms by the phantom’s sides, whilst keeping one arm up and the other on a pillow across the front of the phantom caused only a 16% dose increase. Scans performed with both arms placed on a pillow yielded a 78% increase in dose to the highly radiosensitive lungs and more than doubled the dose to the liver. The scanner’s noise-based AEC system was able to keep the global noise level for all scans within 5% of that obtained for the standard arm position which demonstrates that it accounted for the additional attenuation caused by the presence of the arms within the scanned field of view.
Conclusion
The results of this study reaffirm that the ideal arm position is above the patient’s head whilst the changes in dose and image quality observed can enable Radiographers to make informed decisions about arm positioning for less-cooperative patients.

14:30 +/- Image quality evaluation of the AiCE deep learning CT reconstruction algorithm - Sarah Fisher^[1,2], Jonathan Cole^[1], Jane Edwards Velez^[1] and Kawal Rhode^[2] - [1] Royal Free London NHS Foundation Trust, [2] King’s College London A new deep learning CT reconstruction algorithm called AiCE (Advanced intelligent Clear-IQ Engine) has been developed by Canon Medical systems[1] and recently introduced into clinical practice at the Royal Free Hospital. AiCE has demonstrated reduced image noise and improved image quality in several manufacturer-lead studies when compared to the most advanced iterative reconstruction methods (for example AIDR). However, there have been concerns raised among radiologists locally that AiCE is also removing or smoothing real structures such as small, low contrast polyps and stones. The purpose of this study was to investigate these concerns further and better understand the performance of AiCE locally.
The effect of AiCE on fine texture image quality was investigated for a phantom study and a sample of clinical images. In the phantom study, CT scans were acquired for several grain samples and a uniform water sample placed inside a CelT phantom. Images were acquired for two different field of view sizes and were reconstructed using both AiCE and AIDR. Power spectra, contrast, homogeneity and texture energy metrics were calculated from the reconstructed images to assess the differences in quantitative image quality between AiCE and AIDR. The clinical image study investigated the image quality of fine texture within five KUB and two CAP CT scans. Each scan had both AiCE and AIDR reconstructions which had been acquired as part of the standard of care. Quantitative image quality assessments of the power spectra, contrast and homogeneity were made for femoral head, lung, liver, kidney and bladder regions of interest and compared for AiCE and AIDR.
This study found high contrast texture was either unaffected or enhanced when using AiCE, and that low contrast texture as well as noise was suppressed by AiCE compared to AIDR. This was the case for both the phantom and clinical images. The phantom study also found that the suppression or enhancement of the texture depended on the field of view size. These results show that for high contrast and uniform texture, AiCE is functioning as expected as it appears to be removing noise from uniform samples while preserving real texture. However, the finding that lower contrast textures also appear smoother is evidence that some fine, low contrast texture may be being misidentified as noise and removed from the image. This may have implications for imaging low contrast lesions within low contrast tissues, for example polyps within the colon.
This study has formed part of a local evaluation of AiCE and has contributed to the understanding of the benefits and limitations of AiCE on the image quality in fine texture regions. Further work is needed to understand the impact of these findings on clinical image quality and assess whether the use of AiCE may be contributing to the image quality issues reported by the radiology team.
References
[1] K. Boedeker, AiCE Deep Learning Reconstruction: Bringing the power of Ultra-High Resolution CT to routine imaging, Technical report, Canon Medical Systems, 2019.

Session 4 – Discussion session on physics support to lung screening services

15:20 Interactive discussion session - Gareth Iball

15:50 Round table items and discussion

     Update on tube current modulation testing results - Laurence King

     Update from UKHSA - Sue Edyvean / Jan Jansen