S 8 - My scientific paper in the field of dose optimisation, abdominal imaging, Doppler ultrasound and musculoskeletal imaging

Purpose:

To evaluate a tube voltage-tailored contrast media (CM) application protocol utilising software support for patient-specific injection during coronary CT angiography (CCTA).

Methods and materials:

In this HIPAA-compliant single-centre study, 120 patients referred to CCTA were prospectively assigned to a tube voltage-tailored CM injection protocol. Automated tube voltage selection (ATVS), which automatically adjusts kV to the patient's body habitus, ranged from 70-130kV by applying 10kV intervals. The iodine delivery rate (IDR) was adapted to the tube voltage level using dedicated software (Bayer/Medrad, Indianola, PA). The administrated CM volume ranged from 33mL at 70kV (IDR, 0.8gI/s) to 65mL at 130kV (IDR, 1.9gI/s). Attenuation measurements were performed in the aorta and coronary arteries to calculate quantitative signal-to-noise (SNR) and contrast-to-noise (CNR) ratios. Five-point scales were used to evaluate the overall image quality. Radiation metrics were assessed (Radimetrics™, Bayer) and compared between the protocols.

Results:

The mean age of the study patients was 62.5±11.9 years. Image quality was rated as diagnostic in all patients. Contrast enhancement peaked at 7 kV (p<0.001), while SNR and CNR parameters showed no significant differences between the tube voltages (p=0.085). Additionally, no significant differences were found for subjective image quality parameters between the different protocols (p=0.139). The lowest radiation dose values were observed in the group of patients assigned to the 70kV protocol with a median effective radiation dose of 1.1mSv (p<0.001).

Conclusion:

The proposed tube voltage-tailored injection protocol allows for an individualised scanning of patients undergoing CCTA with significant reductions of CM and radiation dose while simultaneously maintaining a high diagnostic image quality.

Limitations:

The diagnostic accuracy of different tube voltage-tailored CM protocols was not investigated. The vendor-specific parameters for CM administration.

Ethics committee approval

IRB-approved, all subjects provided informed consent.

Funding:

No funding was received for this work.

Purpose:

To evaluate both retrospective and prospective radiation exposure data in a cohort of patients undergoing IC examinations performed by different operators to analyse factors impacting radiation dose with a special focus on operator dependency and to seek operator perceptions on radiation protection training.

Methods and materials:

Retrospective and prospective data collation and analysis of radiation parameters from 179 DCAs and 43 PCIs performed in a single centre (Sept 2018-June 2019) by 3 interventional cardiologists within a Siemens Artis imaging suite were conducted. The radiation parameters available for all patients were dose area product (DAP), total fluoroscopy time (FT), the number of cine runs, and the cumulative air kerma. Data collated for retrospective examinations included D.O.B, gender, previous cardiac history, operator, procedure type, access route, contrast agent, and contrast dose. In addition, patient BMI and the total examination duration and complexity was recorded for prospective data. Dose reference levels were calculated at the 3rd quartile. Operators compared against each other and with national/international standards. Cardiologists with consistently low DRLs were followed up with a semistructured interview.

Results:

DRLs were determined by gender, patient BMI, previous cardiac history, type of IV approach, contrast dose, and the operator, and were within Irish and European DRLs. Significant interoperator variability (p <0.001) was noted. The cardiologist acknowledges his trainer and mentor who influenced him a lot with his technique, decision making, efficiency, and low radiation levels.

Conclusion:

Radiation safety in interventional cardiology procedures is operator dependent. Further research is warranted in the standardisation of operator training with evolving technologies.

Limitations:

The retrospective data collection, discontinuous sampling, procedure-specific rather than patient-specific statistical analysis, and the non-availability of occupational dosimetry.

Ethics committee approval

Ethics approval from UCD School of Medicine and Departmental with permission from Bon Secours Hospital, Dublin.

Funding:

No funding was received for this work.

Purpose:

To investigate scattered radiation in the mobile chest at different distances and heights around the patient’s bed with varying kV values.

Methods and materials:

The study utilised a MobileArt evolution high power type 32kW (manufactured 2010) mobile x-ray unit. Measurements were taken with a phantom laying supine and half-sitting on a bucky table for an AP chest x-ray. The mAs value was 1.2mAs, while the tube potential varied from 95kV-125kV. Scattered radiation was measured both from the phantoms side and caudally with RaySafe i2 dosimeters (868.3MHz). The distances were measured from the centre of the radiation field at 0.5 m, 1 m, 1.5 m, and 2 m. At each distance, doses were measured at 70 cm (bed level), 100 cm, 130 cm, and 160 cm from the floor.

Results:

The amount of scattered radiation increased when kV was raised in both patient positions. With the patient supine with 95kV, the dose rate was 3,222mSv/h, while at 125kV, it was 5,154mSv/h. Scattered radiation decreased significantly when the distance increased. For example, with 125kV and measuring at 160 cm, with 0.5 m distance, the dose rate was 6,244mSv/h, but at 2 meters only it was 0,944mSv/h. The highest dose rates for both supine (5,154mSv/h) and half-sitting (6,244mSv/h) positions were measured at 0.5 m and 160 cm from the floor.

Conclusion:

As expected, the higher the tube potential, the more and further radiation scatters in both positions. The highest dose rates were measured at 160 cm from the floor (radiographers neck-head level) and the lowest doses were at the patient’s bed level. Noting the direction of scattering radiation is important to protect the surrounding patients and personnel.

Limitations:

n/a

Ethics committee approval

No ethical approvement needed.

Funding:

No funding was received for this work.

Purpose:

The manual segmentation of CT images for volumetric analysis of the liver can be time-consuming. Alternatively, partially, or fully computer-aided methods can reduce labour and processing time.

Methods and materials:

62 CT scans from patients examined for diffuse or focal liver diseases were
retrospectively collected. The portal venous phase scans were reconstructed with a slice thickness of 5 mm and without an interslice gap. The area of the whole liver was manually segmented as the ground truth with 3D Slicer software. A grow-from-seeds method was used for semi-automatic three-dimensional segmentation of the livers. Either every third or pre-selected five axial slices were used in the grow-cut process, followed by an automatic hole filling, extrusion removing, and surface smoothing of the predicted volume. We developed in-house a volumetric, fully convolutional neural network (V-net) for automated liver segmentation. The network was trained on 51 scans after image preprocessing and data augmentation. The remaining 11 scans were used in the test set and prediction accuracy was estimated after 10-fold cross-validation.

Results:

The predicted volume of 62 livers in semi-automatic segmentation where the area of the whole liver was manually delineated on every third slice showed a Dice coefficient of (DSC)=mean±SD=0.96±0.007, Hausdorff average distance of (Hd)=0.95±0.22 mm, and a Jaccard coefficient of (Jc)=0.93±0.012. The accuracy of the semi-automatic segmentation using only five manually contoured slices was also excellent: DSC=0.94±0.013, Hd=1.83±0.67 mm, and
Jc=0.88±0.023. Meanwhile, the V-net based segmentation has achieved DSC=0.87±0.024, Hd=4.25±0.76 mm, and Jc=0.77±0.043.

Conclusion:

We conclude that a V-net model can be used for liver segmentation with similar accuracy to manual and semi-automatic methods, thus, expediting the clinical workflow.

Limitations:

The size of the training and validation cohorts should be increased before a clinical application can be introduced.

Ethics committee approval

Approval obtained from the research ethics committee of our institution.

Funding:

No funding was received for this work.

Purpose:

Superb microvascular imaging (SMI) is an innovative Doppler technique for imaging minute vessels with slow velocity. The primary aim of this study was to elucidate the application of SMI in comparison with conventional colour Doppler (CD) and power Doppler (PD) in detecting neovascularity in the supraspinatus tendon. The secondary aim was to investigate the association between neovascularity and shoulder pain and functional disability.

Methods and materials:

The supraspinatus tendon of consecutive patients presented with clinical suspicion of rotator cuff problems were evaluated with grey-scale (tendon enlargement, homogeneity, and hypoechogenicity), CD, PD, and SMI. Two radiologists independently graded the images. The relationship between neovascularity (on CD, PD, and SMI) and pain/disability as determined by a visual analog scale (VAS) and the Oxford shoulder score (OSS) was established.

Results:

59 patients (mean age 53, 39 females) were recruited. 34 (34/59: 58%) patients demonstrated no flow in all Doppler techniques while in 22 (22/59: 37%) patients, vascularity was detected in SMI but not with CD or PD (p<0.01). SMI positively correlated with hypoechogenicity on grey-scale (p<0.01), VAS (p<0.001), and OSS scores (p<0.001), while CDU and PDU did not show a significant relationship. The Kappa-coefficients of grey-scale, CD, PD, and SMI between the two reviewers were 0.87, 0.79, 0.74, and 0.85, respectively.

Conclusion:

Neovascularity on SMI is associated with a hypoechogenic alteration of the supraspinatus tendon and is more related to shoulder pain and functional deficit compared to CD and PD. SMI is a feasible and reliable clinical tool. The supplementation of SMI to conventional ultrasound may increase the performance for diagnosing painful rotator cuff tendinopathy.

Limitations:

A single-centre study with a small sample size.

Ethics committee approval

Approved by the Singapore Institute of Technology (SIT) Institutional Review Board.

Funding:

No funding was received for this work.

Purpose:

To predict the clinical outcome in elderly lung transplantation (LuTx) recipients after transplantation by semi-automated quantification of body composition in preoperative chest CT scans.

Methods and materials:

This retrospective study included 114 patients ≥60 years who received LuTx at the Medical University of Vienna from December 1998-December 2018 and who underwent a chest CT within one year before transplantation. The mediastinal fat areas at the level of the carina and the dorsal muscle group (DMG) areas at the level of the 12th thoracic vertebral body were calculated semi-automatically using OsiriX (Pixmeo, Switzerland). The adjusted and normalised data was correlated with clinical parameters.

Results:

The mean age at transplantation was 63±2.7 years. Fibrosis (40.4%) and COPD (50.9%) were the most common diagnoses. We identified three groups of patients according to body composition (“low risk”: high muscle mass and low mediastinal fat (n=8), “high risk”: low muscle mass and high mediastinal fat (n=12), and “intermediate”: any other combination (n=94)). The “high risk” group patients compared to the “low risk” group patients had a significantly increased risk for wound infections (p=0.001), delirium (p=0.042), and tracheostomy (p=0.032). The time of ventilation (p=0.022; 480 vs 43 hours), the stay in the ICU (p=0.001; 38 vs 5 days), and the stay in hospital (p=0.001; 66 vs 19 days) were significantly longer in the “high risk” group compared to the “low risk” group.

Conclusion:

The body composition in pre-transplantation chest CT scans of LuTx candidates is an objective and easily applicable tool for the identification of patients with a poor outcome after LuTx. In addition, it accentuates the pivotal importance of physical activity and rehabilitation in LuTx.

Limitations:

The small sample size and single centre.

Ethics committee approval

Ethics committee approval obtained.

Funding:

No funding was received for this work.

Purpose:

To investigate the feasibility of a protocol for coronary artery calcium scoring (CACS) at 100kV with tin filtration (Sn100kV) to provide accurate Agatston scores and to asses its potential for radiation dose reduction using a software-based correction algorithm and a kV-independent kernel compared to the standard 120kV acquisition.

Methods and materials:

We analysed image data of 114 patients who underwent a clinically-indicated CACS acquisition using the standard 120kV protocol and an additional Sn100kV CACS research scan. Datasets of the Sn100kV scans were reconstructed using a kV-independent kernel, which produced images with Hounsfield units equivalent to 120kV for bone and calcium. This enabled Agatston scoring without changing the original weighting threshold of 130HU, regardless of the original tube voltage chosen for image acquisition. The Agatston scores and radiation dose values were calculated and compared between the two protocols.

Results:

Median Agatston scores derived from the standard 120kV and the Sn100kV protocol with the kV-independent kernel were 24.7 (IQR:0-171.1) and 21.4 (IQR:0-173.8), respectively (p=0.18). We found an excellent correlation for Agatston scores derived from the two different protocols with a Pearson's correlation of r=0.99. The dose-length-product was 11.4±4mGycm using the Sn100kV and 50.4±24.9mGycm using the standard 120kV protocol (p<0.01), resulting in significantly lowering the effective radiation dose by 77% (0.16±0.06mSv vs 0.7±0.35mSv, P<0.01) for scanning at Sn100kV. Additionally, 99% of patients were classified into the same risk category using the Sn100kV protocol.

Conclusion:

CACS at Sn100kV using the kV-independent kernel is feasible and shows a high correlation compared to standard 120kV scanning. Furthermore, the radiation dose was significantly reduced using the low-kV protocol.

Limitations:

A single-centre study with images read by only one radiologist.

Ethics committee approval

IRB-approved, informed consent waived.

Funding:

No funding was received for this work.