Research Presentation Session
05:28N. Zalokar, Slovenska Bistrica / SI
Purpose:
Since the breasts are considered one of the most radiosensitive organs in the human body, the aim of this study was to investigate whether the dose to the breasts during head CT is reduced with the use of lead shielding.
Methods and materials:The research was carried out on an anthropomorphic phantom of body and head PBU 60 in two major hospitals (hospital A and B) in Slovenia using axial and helical protocols. The breast implant size of 340 ml was attached to the phantom and served to simulate the breast. The dose to the breasts was measured 20 times during each protocol. Half of the measurements conducted during each protocol were performed with the use of lead shielding of 0.5 mm equivalent lead density. Measurements were carried out on the General Electric Healthcare Revolution EVO and the Toshiba Aquilion 64 slice CT unit; 40 measurements’ data was collected. The absorbed dose was measured with the EDD30 dosimeter, which was positioned to the centre of the breast implant.
Results:A significant decrease of the absorbed dose to the breasts in both hospitals was shown. The use of lead shielding reduced the absorbed dose to the breast by 95% (p<0.001) in hospital A and 86% (p<0.001) in hospital B during the axial head CT examination. A significant dose reduction was also shown during the helical head CT examination; 96% (p<0.001) dose reduction in hospital A and 82% (p<0.001) in hospital B.
Conclusion:The use of lead shielding to the breasts is highly recommended during a head CT examination, regardless of the used protocol, due to the significant dose reduction.
Limitations:The study was only conducted on a phantom.
Ethics committee approvaln/a
Funding:No funding was received for this work.
06:03D. Biddle, Cambridge / UK
Purpose:
A comparison between two trauma CT protocols. Can a multiphase contrast injection, single-pass acquisition be a viable technique in major trauma?
Methods and materials:A retrospective analysis of 100 consecutive major trauma patients referred for CT from January-February 2019 was conducted. 50 patients were imaged with the established two-pass protocol (arterial thoracoabdominal and portal venous abdominopelvic phases) and 50 patients underwent the single-pass, multiphase injection protocol (mixed arterial and venous phase of chest, abdomen, and pelvis). Patients were imaged on a Siemens Definition AS+ 128 slice scanner. Both trauma protocols were bolus-triggered using a region-of-interest (ROI) placed over the descending thoracic aorta. Hounsfield unit measurements were obtained from the main pulmonary artery, thoracic aorta, and portal vein. Two consultant radiologists assessed the overall quality of the studies, rating the studies as diagnostic or non-diagnostic. Splenic enhancement was specifically assessed as a marker of solid organ enhancement. Radiation dose and injury severity score (ISS) data were evaluated for each cohort.
Results:The single-pass cohort had significant dose reduction compared to the two-pass method (DLP 903.84 vs 1856.42 mGycm: p=0.0001). Pulmonary artery enhancement was comparable (p=0.3247), whilst the aortic (p=0.0026) and portal vein (p=0.0118) attenuation was significantly higher in the single-pass cohort. Both cohorts had excellent diagnostic quality: 100%. There was no significance between splenic attenuation (p=0.6845) in both techniques. There was no significance relating to ISS (p=0.1668).
Conclusion:The single-pass, multiphase technique significantly reduces the radiation dose in trauma patients whilst maintaining excellent diagnostic accuracy.
Limitations:The study cohort was limited to 50 patient from each technique. A larger inclusion would provide more diversity of trauma referrals to evaluate the technique
Ethics committee approvaln/a
Funding:No funding was received for this work.
06:01M. Kusk, Esbjerg / DK
Purpose:
Advances have made scanning at kVps other than 120 feasible; optionally with spectral filtering for dose reduction. This has prompted “artificial 120 kVp” kernels scoring to correct Hounsfield values acquired at other kVps, primarily for coronary calcium scoring. Our purpose is to assess the accuracy of this kernel compared to “true” 120 kVp scans over a wider range of kVps and tissue equivalents.
Methods and materials:Images were acquired on Siemens Somatom FORCE scanner. A Gammex-RMI 461 phantom with 5 different density inserts was scanned 5 times in 10 kV steps from 70-150 plus at 100 and 150 kV with a spectral filter. CTDI was kept constant. 3 mm slices were reconstructed at standard and “artificial 120 kVp” kernel. Measurements of Hounsfield value and standard deviation were performed. ANOVA was used to test for differences from 120 kV scans.
Results:For soft-tissue and fat equivalent tissues, "artificial 120 kVp" kernel did not significantly alter Hounsfield values for images acquired without spectral filtration, regardless of kVp. The difference from "true" 120 kVp varied by up to 30%. At 100 kVp with tin filter, Hounsfield values matched within 1%. 150 kVp images with tin filter differed by 35%. Bone/calcium equivalent tissue Hounsfield values matched "true" 120 kVp images to within 4%.
Conclusion:Hounsfield values of soft tissue and adipose tissue are not recovered by the "artificial 120 kVp" kernel and observers should be aware of these limitations when viewing images reconstructed with this kernel. Hounsfield values are accurate for calcifications.
Limitations:As a phantom study, results should be verified on real patients before results can be generalised.
Ethics committee approvaln/a
Funding:No funding was received for this work.
05:44N. Zalokar, Slovenska Bistrica / SI
Purpose:
To establish DRLs based on patients' body mass indexes and compare them with DRLs established based on patients' weight.
Methods and materials:Data for 1,633 patients that were imaged on a single CT unit in a time span of 6 months were collated. Only adult patients were included in the study. Filtration of the data was made based on patients’ weights which was selected according to the ICRP 135 publication; patients with 70±10 kg were included in the study. Another filtration of the data was made in the second part of the study based on the patients’ average BMIs; patients with BMI 27 ± 5 were included in the study. Examinations performed less than 20 times were excluded. The DRLs were established at the 75th percentile for the total DLP.
Results:The data of 621 patients was analysed after the filtration based on patients’ weights. After the filtration based on the average BMI, 1,152 patients remained for the analysis. DRLs were established for 7 examinations (abdomen, abdomen-pelvis, CT angiography, extremity, head, pulmonary angiography, and the thorax). Established DRLs based on patients’ BMIs were increased by 11%, 22%, 10%, 46%, 6%, 22%, and 11% for the abdomen, abdomen-pelvis, CT angiography, extremity, head, pulmonary angiography, and the thorax, respectively, compared to the DRLs based on patients’ weights. It has to be stated that the average patient weight from our collected data was 80.3 kg and the median was 80 kg.
Conclusion:Based on our results and the fact that the population is becoming heavier, we would recommend that DRLs should be established based on patients’ BMIs to get more accurate results.
Limitations:No CTDIvol data.
Ethics committee approvaln/a
Funding:No funding was received for this work.
06:17M. Kusk, Esbjerg / DK
Purpose:
To examine the influence on the sequence of lateral and frontal scan projection radiographs (SPR), with and without patient de-centring, on dose-length-product using CAREDose4D tube current modulation at thoracoabdominal CT.
Methods and materials:Two scanners were examined: A) Definition FLASH and B) Definition AS (Siemens Healthcare AG). A full-body anthropomorphic phantom was marked for placement reproducibility and placed in the standard position. 5 identical thoracoabdominal scans were performed for each of the SPR combinations as follows: A-P, P-A, lateral, A-P plus lateral, lateral plus A-P, PA-plus lateral, and lateral plus P-A. All series were repeated three times: with an iso-centred phantom and 5 cm decentring anterior and posterior, respectively. DLP was noted for each scan. mAs vs Z-position was plotted.
Results:For both scanners, A-P or P-A SPRs alone resulted in higher doses than lateral alone. Scans with P-A or A-P SPR followed by lateral SPR resulted in 30% lower DLP than with A-P or P-A performed first. With decentring, the variation in the dose was less than 5% when lateral SPR performed alone or after A-P or P-A SPR. When P-A or A-P performed alone or after lateral SPR, decentring resulted in differences of up to 50%. Relative variations were largest in denser regions, but lower in the thorax.
Conclusion:Incorrect sequencing of SPRs can seriously alter the patient radiation dose when using CAREDose4D, especially when the patient is not centred. Radiographers must be aware of this when designing protocols or prescribing additional SPRs.
Limitations:As this was only a phantom study and dose variations were not uniform across anatomy, the effective total dose variations are not necessarily linearily correlated with DLP. Other software versions may exhibit other behavior.
Ethics committee approvaln/a
Funding:No funding was received for this work.
06:27L. Kuopusjarvi, Oulu / FI
Purpose:
Abnormal doses in CT examinations have been studied in the past, although often with complex methods. We developed a practical method to detect abnormal radiation doses in CT examinations via Radimetrics.
Methods and materials:4 CT protocols were chosen for evaluation from 1/10/2017-22/08/2019: routine head (n=23985), abdomen in venous phase (n=4445), pulmonary embolism (n=2404), and thorax/abdomen/pelvis suspected malignancy scan (n=3668). Examinations were investigated with the following parameters: CTDIVOL/SSDE vs WED, acquisition count vs scan length, and maximum DLP recorded per protocol. Dose outliers were visually inspected based on the graphs containing these parameters.
Results:Some dose outliers were excluded based on patient-based reasons such as hands inside the scan FOV, wrongly registered protocol name, or unusual imaging instructions given by the radiologist. We observed the following outliers per protocol: routine head (n=5), abdomen in venous phase (n=13), pulmonary embolism (n=26), and thorax/abdomen/pelvis scan (n=8). Out of 52 outliers, 33 included double scanning. Too long scan length or scanning outside of survey radiograph was present in 27 incidents. 8 incidents included contrast media extravasation and in 7 of those, a second scan was required. 17 examinations had to be repeated because of a failed timing of a contrast media bolus. Most commonly, this was because the patient held their hands down. This caused image quality degradations in the monitoring phase of i.v. contrast injection, yielding a scan start delay.
Conclusion:This project led to clarifications in CT-examination instructions, detailed presentations on the causes of dose outliers, and changes in CT imaging protocols. For example, the lateral survey radiographs were optimised in pulmonary embolism to make the pulmonary area more visible.
Limitations:This method is applicable to other CT protocols as well.
Ethics committee approvalThe institutional review board approved the study (204/2019).
Funding:No funding was received for this work.
05:39W. Elshami, Sharjah / AE
Purpose:
Acceptable quality dose (AQD) is a bottom-top optimisation approach based on image quality, radiation dose, and a patient’s weight. The standard dose value is used as an optimisation tool for producing a quality image in the diagnostic examination. The primary purpose of this study is to determine the AQD resulting from various protocols for adult patients undergoing CT examinations in 4 hospitals in the UAE.
Methods and materials:The data used in this study was collected from 4 CT scanners for adult patients undergoing brain, chest, abdomen, and chest-abdomen CT. 320 patients were included in the study. Patient information and exposure parameters were extracted. All images were assessed for image quality according to scoring criteria 1, 2, 3, and 4 corresponding to bad, not acceptable, acceptable, and higher than necessary quality, respectively. Only images with acceptable quality (score of 3) were selected and grouped into 8 weight groups. The median values for CTDIvol and DLP were determined as the AQD.
Results:The results were depicted in table form.
Conclusion:AQD is a bottom-top optimisation approach based on image quality, radiation dose, and a patient’s weight. The integration of image quality scoring with dose might result in the detection of situations with higher-than-necessary image quality. Thus, awareness and orientation of radiologists in image quality are essential to optimise image quality and patient dose.
Limitations:The study focused on the common examinations only, thus, other examinations such as angiography and the spine were not included in the study.
Ethics committee approvalApproval Reference No: MOHP/REC-18/2018.
Funding:No funding was received for this work.
05:57J. Nielsen, Odense / DK
Purpose:
Forensic institutes have widely embraced the use of postmortem CT (PMCT) as a supplement or replacement of autopsy in postmortem examinations. However, imaging of the deceased provides an array of challenges not observed in the living: rigor mortis, damage by fire, severe decomposition, or dismemberment not only complicates the proper identification of anatomy but also affects the positioning and centring of the diseased. Consequently, there is a need for reconstruction methods aimed specifically for postmortem conditions to compensate for the variable states of the bodies. In this study, alternative reconstruction methods were tested.
Methods and materials:4 reconstruction protocols, 3 newly constructed and 1 standard, were compared using a visual grading analysis with 4 observers. The study population was 20 retrospective PMCT scans with different degrees of decomposition. Interobserver agreement was assessed using Fleiss Kappa, with a 95% confidence interval.
Results:The stage of decomposition highly influenced the evaluation from the observer of the usefulness of the reconstruction. Excluding the severely decomposed bodies, 93.4-100% of the 20 PMCTs evaluated were found to be useful for diagnosis using the 4 reconstructions. One newly constructed protocol scored higher in VGAS compared to the standard protocol, with respect to usefulness and beam hardening, indicating a better diagnostic value.
Conclusion:There is a need for improved reconstruction parameters in PMCT, focusing on an increase of details otherwise lost to decay. Moreover, protocols must be targeted to the condition of the body, as highly decomposed, incinerated, and waterlogged bodies all differ structurally from the recently deceased as well as pose-practical problems in relation to positioning.
Limitations:Observers had limited experience in forensic imaging. A limited study population. The question for noise assessment was worded ambiguously.
Ethics committee approvaln/a
Funding:No funding was received for this work.
04:15M. Tsuda, Yonago-Shi Tottori-Ken / JP
Purpose:
To assess the image quality of chest CT using a model-based iterative reconstruction (MBIR) with a lung-image filter.
Methods and materials:The image quality was assessed subjectively and objectively. In the subjective assessment of overall image quality, 25 clinical lung images were assessed by 3 radiologists and scored using a 5-point scale. The images were reconstructed using filtered back projection (FBP), adaptive statistical iterative reconstruction (ASiR-V), and MBIR with a lung-image filter. ASiR-V blending level was 50%. Modulation transfer function (MTF) and noise power spectrum (NPS) were measured in the objective image quality assessment. The MTF measurement phantom has 20, 100, 200, and 500 Hounsfield unit (HU) blocks. The blocks were set into a water phantom and 4 contrast MTFs were obtained using the phantom. NPS measurements were obtained using a water phantom. All measurements in the objective assessment were made using the “CT measure” application (Japanese Society of Technology).
Results:The images obtained by MBIR with lung-image filter received the highest scores in the subjective assessment and showed improved spatial resolution under high-contrast conditions. NPS showed mostly good noise reduction at the MBIR with a lung-image filter.
Conclusion:Using MBIR with a lung-image filter significantly improved the image quality in chest CT.
Limitations:Objective assessments were not made using task-based methods.
Ethics committee approvalThis study has been approved by Certified Review Board, Tottori University Hospital.
Funding:No funding was received for this work.