Responsible AI practice and AI education are central to AI implementation: a rapid review for all medical imaging professionals in Europe.

Artificial intelligence (AI) has transitioned from the lab to the bedside, and it is increasingly being used in healthcare. Radiology and Radiography are on the frontline of AI implementation, because of the use of big data for medical imaging and diagnosis for different patient groups. Safe and effective AI implementation requires that responsible and ethical practices are upheld by all key stakeholders, that there is harmonious collaboration between different professional groups, and customised educational provisions for all involved. This paper outlines key principles of ethical and responsible AI, highlights recent educational initiatives for clinical practitioners and discusses the synergies between all medical imaging professionals as they prepare for the digital future in Europe. Responsible and ethical AI is vital to enhance a culture of safety and trust for healthcare professionals and patients alike. Educational and training provisions for medical imaging professionals on AI is central to the understanding of basic AI principles and applications and there are many offerings currently in Europe. Education can facilitate the transparency of AI tools, but more formalised, university-led training is needed to ensure the academic scrutiny, appropriate pedagogy, multidisciplinarity and customisation to the learners' unique needs are being adhered to. As radiographers and radiologists work together and with other professionals to understand and harness the benefits of AI in medical imaging, it becomes clear that they are faced with the same challenges and that they have the same needs. The digital future belongs to multidisciplinary teams that work seamlessly together, learn together, manage risk collectively and collaborate for the benefit of the patients they serve.

Comparison of image quality assessments between interventional radiographers and interventional radiologists using digital subtraction angiography.

Purpose: The aim of our study was to compare the image quality assessments of vascular anatomy between interventional radiographers and interventional radiologists using digital subtraction angiography (DSA) runs acquired during an interventional radiology procedure. Approach: Visual grading characteristics (VGC) analysis was used to assess image quality by comparing two groups of images, where one group consisted of procedures in which radiation dose was optimized (group A, n = 10 ) and one group where dose optimization was not performed (group B, n = 10 ). The radiation dose parameters were optimized based on theoretical and empirical evidence to achieve radiation dose reductions during uterine artery embolization procedures. The two observer groups comprised of interventional radiologists ( n = 4 ) and interventional radiographers ( n = 4 ). Each observer rated the image quality of 20 DSA runs using a five-point rating scale. Results: The VGC analysis produced an area under the VGC curve ( AUC VGC ) of 0.55 for interventional radiographers ( P = 0.61 ) and AUCVGC of 0.52 for interventional radiologists ( P = 0.83 ). The optimization of radiation dose parameters demonstrated a reduction in kerma-area product by 35% ( P = 0.026 , d = 0.5 ) and reference air kerma (Ka, r ) by 43% ( P = 0.042 , d = 0.5 ) between group A and group B. Conclusions: VGC analysis indicated that the image quality assessments of interventional radiographers were comparable with interventional radiologists, where a reduction in radiation dose revealed no effect on both observer groups regarding their image quality assessment of vascular anatomy.

The impact of body mass index on patient radiation dose in general radiography.

The aim of the present study was to determine the influence of the body mass index (BMI) on the dose area product (DAP) and effective dose (ED) in overweight and obese patients. We also wanted to determine the typical dose values as well as suggest adjustments to clinical practice for overweight and obese patients. In this study we considered 597 patients referred for imaging of the chest in posteroanterior and lateral projection, the lumbar spine in anteroposterior (AP) and lateral projection, the pelvis, the knee in AP and lateral projection, and the shoulder in AP projection. For each examination, the image field size, tube voltage, mAs product, source-to-image receptor distance and values of DAP were collected. Based on their BMI, the patients were divided into three groups (normal weight, overweight and obese). At the end, PCXMC 2.0 software was used to calculate the ED. The study showed a statistically significant DAP and ED increase in overweight and obese patients by 28.9% up to 275.4% in the case of DAP and an increase in ED from 11.0% to 241.9% in all mentioned examinations except knee and shoulder imaging. Typical DAP values ranged from 2.2 to 54.8µGym2for normal-weight patients, from 2.2 to 87.6µGym2for overweight patients, and from 2.2 to 172.5µGym2for obese patients. Spearman's correlation coefficient revealed very weak to very strong correlations when comparing BMI and DAP, as well as when comparing BMI and ED. A strong and very strong correlation was found in the case of examinations of the torso (except for the comparison of BMI and ED in the case of lateral lumbar spine projection).

Predictors of radiation dose for uterine artery embolisation are angiography system-dependent.

This study sought to achieve radiation dose reductions for patients receiving uterine artery embolisation (UAE) by evaluating radiation dose measurements for the preceding generation (Allura) and upgraded (Azurion) angiography system. Previous UAE regression models in the literature could not be applied to this centre's practice due to being based on different angiography systems and radiation dose predictor variables. The aims of this study were to establish whether radiation dose is reduced with the upgraded angiography system and to develop a regression model to determine predictors of radiation dose specific to the upgraded angiography system. A comparison between Group I (Allura,n= 95) and Group II (Azurion,n= 95) demonstrated a significant reduction in kerma-area product (KAP) and Ka, r (reference air kerma) by 63% (143.2 Gy cm2vs 52.9 Gy cm2;P< 0.001,d= 0.8) and 67% (0.6 Gy vs 0.2 Gy;P< 0.001,d= 0.8), respectively. The multivariable linear regression (MLR) model identified the UAE radiation dose predictors for KAP on the upgraded angiography system as total fluoroscopy dose, Ka, r, and total uterus volume. The predictive accuracy of the MLR model was assessed using a Bland-Altman plot. The mean difference was 0.39 Gy cm2and the limits of agreement were +28.49 and -27.71 Gy cm2, and thus illustrated no proportional bias. The resultant MLR model was considered system-dependent and validated the upgraded angiography system and its advance capabilities to significantly reduce radiation dose. Interventional radiologist and interventional radiographer familiarisation of the system's features and the implementation of the newly established MLR model would further facilitate dose optimisation for all centres performing UAE procedures using the upgraded angiography system.

UTERINE ARTERY EMBOLISATION: CONTINUOUS QUALITY IMPROVEMENT REDUCES RADIATION DOSE WHILE MAINTAINING IMAGE QUALITY.

The purpose of this study was to introduce a continuous quality improvement (CQI) program for radiation dose optimisation during uterine artery embolisation (UAE) and assess its impact on dose reduction and image quality. The CQI program investigated the effects of optimising radiation dose parameters on the kerma-area product (KAP) and image quality when comparing a 'CQI intervention' group (n = 50) and 'Control' group (n = 50). Visual grading characteristics (VGC) analysis was used to assess image quality, using the 'Control' group as a reference. A significant reduction in KAP by 17% (P = 0.041, d = 0.2) and reference air kerma (Ka, r) by 20% (P = 0.027, d = 0.2) was shown between the two groups. The VGC analysis resulted in an area under the VGC curve (AUCVGC) of 0.54, indicating no significant difference in image quality between the two groups (P = 0.670). The implementation of the CQI program and optimisation of radiation dose parameters improved the UAE radiation dose practices at our centre. The dose reduction demonstrated no detrimental effects on image quality.

Establishment of typical adult CT dose indicators for PET-CT scans in Slovenia.

The aim was to determine typical values of diagnostic reference level (DRL) quantities for the computed tomography (CT) part of the most common positron emission tomography-computed tomography (PET-CT) procedures in Slovenia. The most common PET-CT procedures were identified, and data collated for 565 patients imaged in all three PET-CT units in Slovenia during a time span of 11 months. As the number of facilities is too low to establish national DRLs, we followed ICRP recommendations and determined typical values of DRL quantities as the median values of the pooled set of data. Mean, median, and standard deviation of CT dose index (CTDIvol) and total dose length product (DLP) for the CT part of the most common PET-CT procedures were determined for pooled data as well as for each PET-CT unit. The data were compared between all three units to identify possible outliers that would likely benefit from optimization. Three most common CT protocols covering approximately 2/3 of all PET-CT imaging performed in Slovenia were considered: from the base of the cranium to the middle of the femur, from the top of the head to the middle of the femur, and for the whole-body PET-CT. The established typical values in terms of total DLP were 295, 359, and 676 mGyccm, respectively; and in terms of CTDIvol3.05, 3.22, and 3.60 mGy, respectively. Comparing the data between all three units showed significantly higher (p< 0.001) patient doses on one unit, indicating a need for optimization. The results present the first-time data on the national typical values of DRL quantities for the CT part of most common PET-CT procedures in Slovenia. While the determined typical values are within the DRL values established in some other countries, significant differences were found between the individual units included in the study.

Does the use of self-compression in mammography affect compression force, breast thickness, and mean glandular dose?

PURPOSE: The purpose of the study was to investigate whether the use of self-compression in craniocaudal (CC) projection has an effect on compression force, breast thickness, radiation dose and image quality compared to the standard mammographic procedure. METHODS: The study was conducted on 200 female patients that were referred for mammographic imaging. Patients were randomly divided into two equal groups. In the first group, self-compression was performed on the right breast and in the second group on the left breast. The data about compression force (N), breast thickness (mm), and mean glandular dose (MGD; mGy) were collected. In addition, the differences in the mentioned variables according to the side of self-compression were compared. All mammographic images were evaluated by two experienced radiologists according to the criteria established by the European Commission. RESULTS: The use of self-compression resulted in a significant increase in compression force by 21.7 % (19.8 N) and a significant reduction of breast thickness by 5% (2.43 mm) and MGD by 6.3 % (0.09 mGy), respectively. There were no statistically significant differences based on the self-compression side, and no differences were observed in image quality assessment. CONCLUSION: This study demonstrates that the imaging protocol in mammography of the CC projection can be adopted by the use of self-compression in order to achieve better results.

Comparison of treatment position with mask immobilization and standard diagnostic setup in intracranial MRI radiotherapy simulation.

PURPOSE: This study aims to compare the quality of images resulting from magnetic resonance imaging of patients who underwent intracranial MRI simulation using two different setups (treatment position with mask immobilization and standard diagnostic setup). Due to a larger number of channels and lack of mask immobilization in the standard diagnostic setup, we would like to evaluate whether this is an appropriate technique for MRI treatment planning. METHODS: In total, 70 patients who underwent MR imaging of the brain at 1.5T were included in the study (48 for 6­channel flex coil, 22 for 24-channel HNU face bill coil). Contrast-enhanced 3D T1w and T2 FLAIR images were acquired. Images were subjectively compared for artifact appearance and general image quality by three radiographers. Objective comparison of contrast rate, contrast-to-noise ratio, and signal-to-noise ratio was also performed. RESULTS: FLAIR and contrast-enhanced 3D T1w images showed various artifacts, such as susceptibility and movement artifacts. There were no statistically significant differences regarding the evaluation of movement artifacts between two coils and two different immobilization methods. There were also no statistically significant differences (p > 0.05) between the 6­channel flex coil and 24-channel HNU face bill coil regarding qualitative general image quality and objective measures. CONCLUSION: There were no statistically significant differences between the occurrence of movement artifacts, overall image quality, and objective image quality in treatment position with mask immobilization and standard diagnostic setup. Based on this result, we can conclude that a standard diagnostic setup is also applicable in intracranial MRI treatment planning with no loss to image quality. Registration of the imaging plans was not performed in this study; therefore, it might still be necessary to perform measurements of tumor delineation matching and geometrical accuracy acceptance in our institution.

Typical air kerma area product values for trauma orthopaedic surgical procedures.

BACKGROUND: The aim of study was to establish the typical radiation quantity values for the most common trauma orthopaedic surgical procedures and to compare them with reference values of equivalent procedures performed in other institutions. In addition, we assess the impact of image intensifier and flat panel detector technology used for fluoroscopically guidance on patient exposure. MATERIALS AND METHODS: Five most frequently performed fluoroscopically guided trauma orthopaedic procedures in University Medical Centre Ljubljana were analysed. Data on 199 cases over a 6 months period from December 2016 to June 2017 were gathered retrospectively. Study covered 40 dynamic hip screw fixations (DHS), 23 proximal femoral nail insertions (PFN), 20 proximal humeral nail insertions (PHN), 77 partial hip endoprosthesis implantations (PEP) and 39 percutaneous posterior spine fixations (PPS). The median and average along with the first and third quartile values of air kerma area product (KAP) for each procedure type were calculated as well as median and average value of fluoroscopy screening time. RESULTS: Typical KAP value for dynamic hip screw fixation was set at 0.52 Gycm2; for proximal femoral nail insertion at 0.53 Gycm2 and for proximal humeral nail insertion at 0.26 Gycm2. For implantation of partial endoprosthesis typical KAP value utilizing flat panel technology was set at 0.08 Gycm2 and at 0.21 Gycm2 when the image intensifier technology was used. Typical KAP value for percutaneous posterior spine fixation was set at 1.26 Gycm2, using flat panel technology and at 3.98 Gycm2 using image intensifier technology. CONCLUSIONS: Established typical KAP levels of surgical orthopaedic procedures in traumatology will serve as a valuable tool for further radiation exposure optimization.

OPTIMAL COLLIMATION SIGNIFICANTLY IMPROVES LUMBAR SPINE RADIOGRAPHY.

PURPOSE: To determine the influence of optimal collimation during lumbar spine radiography on radiation dose and image quality. MATERIAL AND METHODS: 110 lumbar spine patients were split into two groups-the first imaged with standard collimation and the second with optimal collimation. Body mass index, image field size, exposure conditions and dose area product were measured. Effective and absorbed organ doses were calculated. Image quality was assessed. RESULTS: Optimal collimation reduced the primary field by up to 40%. The effective dose was reduced by 48% for the AP projection, while no differences were found for the LAT projection due to incorrect positioning of the central beam with standard collimation. The absorbed dose to selected radiosensitive organs decreased by 41 and 10% in the AP and LAT projections, respectively. Image quality for the LAT projection improved by 24% and maintained for the AP projection. CONCLUSION: Optimal collimation in lumbar spine imaging significantly influences patient exposure to radiation.

The efficiency of lead and non-lead shielding on breast dose in head CT.

The aim of the study was to assess the effect of the shielding material and its thickness on the measured skin dose to the breasts during the CT examination of the head. The helical and axial head CT was performed on an anthropomorphic phantom (PBU 60). Two types of shielding were tested-lead and non-lead (antimony-bismuth) shielding. Measurements with different thicknesses were performed and the shielding efficiency of the materials was compared. Skin dose to the breasts was measured with an educational direct dosimeter (EDD-30). The shielding efficiency during both scanning protocols indicated an increased dose reduction with the thicker equivalent thickness in both shielding materials. Dose reduction was the highest at 0.5 mm equivalent thickness for both materials; lead shielding reduced the dose by 91% and 83%, the antimony-bismuth shielding by 90% and 86%, during the axial and helical head CT protocols, respectively. Statistically significant differences were found between the materials of the same equivalent thickness (0.175, 0.25 and 0.5 mm) during the helical protocol in favor of the antimony-bismuth shielding. During the axial protocol there were no statistically significant differences. Shielding of radiosensitive organs can prevent unnecessary exposure of radiosensitive organs outside the primary beam. Due to the significant decrease in radiation dose to the breasts, and many other positive attributes, use of the antimony-bismuth shielding instead of the lead shielding should be considered, especially during the helical CT scan of the head.

RADIATION DOSE DURING PELVIC RADIOGRAPHY IN RELATION TO BODY MASS INDEX.

The purpose of this research was to investigate the impact of body mass index (BMI) on dose area product (DAP), effective dose (E), dose to the organs and image quality (IQ) on 200 patients referred to pelvic radiography. Patients were classified into three groups according to BMI: normal (<24.99), overweight (25.0-29.99) and obese (>30). The results showed 52% and 135% higher DAP for overweight and obese patients compared to normal-weight patients (p < 0.001). A 46 and 123% rise of E for overweight and obese patients compared to normal-weight patients (p < 0.001) was discovered. Overweight patients received 37% higher dose and obese patients 107% higher dose to the organs compared to normal-weight patients. There were no statistically significant differences between IQ, except between normal weight and overweight patients. A strong correlation (r = 0.733) was found between BMI and DAP and between BMI and E (r = 0.776).

Establishment of national diagnostic reference levels for radiotherapy computed tomography simulation procedures in Slovenia.

PURPOSE: To propose national diagnostic reference levels (DRLs) for radiotherapy (RT) computed tomography (CT) localization purposes, compare both CT units used in the largest RT department in the country and to compare gathered results with other published DRLs in order to discover any need of optimization. METHODS: In total, 1631 patient data (time spend of 4 months) regarding sex, examination type, total dose-length product (DLP) and CTDIvol was collated on two CT units. Those simulation procedures account for more than 80 % of all simulation procedures performed nationwide. Then, total DLP and CTDIvol was calculated and mean, median and 3rd quartile for both units together were presented to determine national DRLs for simulation procedures. The same data was later compared between both units to discover any potential need for optimization. RESULTS: 3rd quartile values of DLP for abdomen, breast, chest, head, head and neck, pelvis and spine were 1116.2, 606.6, 832.4, 1942.4, 969.2, 677.1 and 1042.4 mGy∙cm, respectively. 3rd quartile CTDIvol values for the same sequence of procedures were 18.7, 13.3, 19.2, 76.9, 22.6, 17.9 and 22.2 mGy, respectively. Among the two units, the mentioned dose values were on average significantly higher on one CT unit than on the other unit. CONCLUSIONS: When comparing collected dose values with other studies, RT CT DRLs showed that radiation doses from our institution were similar or even lower. Some variations were found between both CT units in certain protocols, so exposure parameters should be reviewed and optimized.

A PHANTOM STUDY SHOWING THE IMPORTANCE OF BREAST SHIELDING DURING HEAD CT.

This study aimed to investigate the dose to the breasts during head computed tomography (CT) if lead shielding is used. The study was performed in two major hospitals using helical and axial protocols on an anthropomorphic phantom. Measurements were performed with and without the use of a lead shield of 0.5 mm equivalent density. The results showed a significant decrease in dose with the lead shielding in both hospitals. During the helical protocol, the use of shielding significantly reduced the dose by 96% in Hospital A and 82% in Hospital B. The dose reduction during axial protocol was also significant: 95% in Hospital A and 86% in Hospital B with lead shielding. Considering the significant dose reduction of 82% up to 96% during this study, we highly recommend the shielding of breasts regardless of the protocol used during head CT examinations.

Efficacy of breast shielding during head computed tomography examination.

BACKGROUND: Female breasts are exposed to scattered radiation regardless of not being included in the primary field during head CT. This study aimed to investigate whether the use of lead shielding is beneficial in dose reduction to the breasts during head CT. PATIENTS AND METHODS: The study was performed in two different hospitals on two different CT units and included 120 patients. Half of the measurements (n = 60) was conducted without the use of lead shielding and the other half (n = 60) with the use of lead shielding of 0.5 mm equivalent thickness. RESULTS: Significant skin dose reduction to the breasts during head CT in both hospitals with the use of lead shielding was discovered; 81% (338.2 ± 43.7 µGy to 64.3 ± 18.8 µGy) in Hospital A and 74% (from 253.1 ± 35.1 µGy to 65.3 ± 16.9 µGy) in Hospital B. CONCLUSIONS: Considering the assumed carcinogenic effect of low doses of radiation, high frequency of the head CT scans and the significant reduction of radiation doses to the highly radiosensitive breasts, the use of lead shielding is highly recommendable.

LUMBAR SPINE RADIOGRAPHY: LOWER ORGAN DOSE WITH THE USE OF PA PROJECTION.

The purpose of the research was to determine the effect of the posteroanterior (PA) patient position in lumbar spine imaging on effective dose and the absorbed organ dose. The study was performed on 100 patients that were referred to the lumbar spine radiography that were divided into two equal groups of 50. Body Mass Index, Dose Area Product (DAP), exposure index (EXI), tube time-current (mAs), image field size and the source-patient distance were acquired for each patient. The entrance surface dose (ESD), the effective dose and the absorbed organ doses were calculated. There was no statistically significant difference in the BMI and EXI between the AP and PA projection. The results showed a significant reduction of ESD by 33% and the effective dose by 53% when the PA projection was used. Furthermore, there was a 64% average reduction of the absorbed organ doses to the selected organs.

Comparison of anteroposterior and posteroanterior projection in lumbar spine radiography.

Background The aim of the study was to compare patient radiation dose and image quality in planar lumbar spine radiography using the PA and AP projection in a large variety of patients of both sexes and different sizes. Patients and methods In the first phase data of image field size, DAP, effective dose and image quality were gathered for AP and PA projection in lumbar spine imaging of anthropomorphic phantom. In the second phase, data of BMI, image field size, diameter of the patient's abdomen, DAP, effective dose and image quality were gathered for 100 patients of both sexes who were referred to lumbar spine radiography. The patients were divided into two groups of 50 patients, one of which was imaged using the AP projection while the other the PA projection. Results The study on the phantom showed no statistically significant difference in image field size, DAP and image quality. However, the calculated effective dose in the PA projection was 25% lower compared to AP projection (p =0.008). Measurements on the patients showed no statistically significant difference between the BMI and the image field size. In the PA projection, the thickness of abdomen was 10% (p < 10-3) lower, DAP 27% lower (p = 0.009) and the effective dose 53% (p < 10-3) lower than in AP projection. There was no statistically significant difference in image quality between the AP and the PA projection. Conclusions The study results support the use of the PA projection as the preferred method of choice in planar lumbar spine radiography.

The effect of breast shielding during lumbar spine radiography.

BACKGROUND: The aim of the study was to determine the influence of lead shielding on the dose to female breasts in conventional x-ray lumbar spine imaging. The correlation between the body mass index and the dose received by the breast was also investigated. MATERIALS AND METHODS: Breast surface dose was measured by thermoluminescent dosimeters (TLD). In the first phase measurements of breast dose with and without shielding from lumbar spine imaging in two projections were conducted on an anthropomorphic phantom. In the second stage measurements were performed on 100 female patients, randomly divided into two groups of 50, with breast shielding only used in one group. RESULTS: On average, breast exposure dose in lumbar spine imaging in both projections (anteroposterior (AP) and lateral) was found reduced by approximately 80% (p < 0,001) when shielding with 0.5 mm lead equivalent was used (from 0.45±0.25 mGy to 0.09±0.07 mGy on the right and from 0.26±0.14 mGy to 0.06±0.04 mGy on the left breast). No correlation between the body mass index (BMI) and the breast surface radiation dose was observed. CONCLUSIONS: Although during the lumbar spine imaging breasts receive low-dose exposure even when shielding is not used, the dose can be reduced up to 80% by breast shielding with no influence on the image quality.