scholarly journals Performance of different myocardial tissue tracking algorithms and acquisition-based strain imaging to characterise myocardial pathology

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
S.J Backhaus ◽  
G Metschies ◽  
V Zieschang ◽  
J Erley ◽  
S.M Zamani ◽  
...  

Abstract Background Myocardial deformation imaging is superior in risk-stratification compared to volumetric approaches. Myocardial Feature-Tracking (FT) allows easy post-processing of routinely acquired cine images. Since there is no clear recommendation regarding FT post-processing we sought to compare different FT-strains with reference standard techniques including tagging and strain encoded (SENC) magnetic resonance imaging. Methods CMR-FT software from 4 different vendors (TomTec, Medis, Circle, Neosoft), CMR tagging (Segment) and fastSENC (MyoStrain) were used to determine left ventricular (LV) global longitudinal and circumferential strains (GLS and GCS) in 12 healthy volunteers and 12 heart failure patients. Variability and agreements were assessed using intraclass correlation coefficients, coefficients of variation and Bland Altman plots. Results Compared to tagging, FT-based strain was software independently significantly higher except for GCS using Medis (p=0.178). Compared to fSENC, mean-differences of GLS were smaller within a range of ±1.5%. For GCS this only applied to CVI and Medis (<1.5%) but not TomTec (>7%) or Neosoft (>4%). Absolute agreements comparing FT to tagging were best for CVI (GLS ICC0.70) and Medis (GCS ICC0.85). Compared to fSENC agreement of GLS was generally excellent (ICC>0.77), but only CVI and Medis revealed excellent agreement for GCS (ICC0.88 and 0.85). Consistency and correlation of GLS were software independently high compared with tagging and fSENC (ICC>0.86, r>0.76) while being lower for GCS (ICC>0.68, r>0.72). Conclusion Although agreement differs between deformation assessment approaches, consistency and correlation are high irrespective of the method chosen, thus indicating reliable strain assessment. Further standardisation and introduction of uniform references is warranted for clinical routine implementation. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): DZHK - German Centre for Cardiovascular Research

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
I Del Canto Serrano ◽  
P Tejero ◽  
M.P Lopez-Lereu ◽  
J.V Monmeneu ◽  
V Bodi ◽  
...  

Abstract Background Quantification of regional myocardial function allows risk stratification in heart disease. CMR tagging (TAG) enables the evaluation of segmental cardiac deformation, but it has not reached clinical routine due to the long acquisition and post-processing times. Conversely, CMR feature-tracking (FT) is a post-processing method based on standard cine-MR imaging. Purpose To compare myocardial strain and torsion obtained with CMR-TAG and CMR-FT in healthy volunteers and myocardial infarction (MI). Methods 42 subjects (18 healthy; 24 MI) underwent CMR (1.5T, cine/TAG sequences). Global and segmental (16-segment) circumferential strain (CS), and torsion were measured using FT (CVI42, Canada) and tagging (InTag, France). Inter-method agreement was assessed using 2-way-mixed intraclass correlation coefficient (ICC). Results The agreement for segmental and global CS measurements was good to excellent in both groups (Table). Torsion angle showed excellent (0.763) and good (0.697) agreement for healthy and MI. Conclusion CMR-FT strain and torsion measurements showed high agreement with CMR-tagging. Thus, CMR-FT provides a potential clinical alternative in the assessment of regional ventricular function. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): Carlos III Health Institute, Spanish Ministry of Economy and Competiveness; Agencia Valenciana de la Innovaciόn, Generalitat Valenciana


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
F Sedaghat-Hamedani ◽  
J Trebing ◽  
A Kindermann ◽  
E Kayvanpour ◽  
K Tan ◽  
...  

Abstract Introduction Cardiomyopathies (CMPs) are leading causes of heart failure (HF) and sudden cardiac death (SCD). Comparative data of the multiple cardiomyopathy forms are largely missing. The TranslatiOnal Registry for CardiomyopatHies (TORCH) is the largest prospective multicentre CMP registry world-wide. Enrolled patients are comprehensively phenotyped by clinical examinations, state-of-the-art imaging, and molecular investigations. In this study, we present the baseline and 1-year follow-up data. Methods TORCH is a national, prospective, multicentre registry within the German Centre for Cardiovascular Research (DZHK) and includes 2300 patients with non-ischemic (primary and secondary) CMP from 20 centres. The minimum follow up was one year. The DZHK-wide harmonization of datasets and SOPs ensure a high level of data quality and comparability across different CMP forms. Results Dilated cardiomyopathy (DCM) has the highest prevalence with 64% of all enrolled patients, followed by hypertrophic cardiomyopathy (HCM) with 16%. At baseline, patients with arrhythmogenic right ventricular cardiomyopathy (ARVC) were treated more often with ICD implantation and showed high rates of adequate ICD therapies (65.8%, p<0.05 and 47.8%, p<0.05, respectively). The prevalence of stroke or transient ischemic attack (TIA) was in multivariate analysis significantly higher (p<0.05) in left ventricular non-compaction cardiomyopathy (LVNC, 14.9%), while atrial fibrillation was lower than in other cardiomyopathy forms. Patients with amyloidosis had the worst outcome (HR: 6; 95% CI: 2.5–14.5, P<0.05) with annual mortality of >15% and 12% receiving heart transplantation. In DCM, reverse remodelling with improvement of functional parameters and biomarkers was more often observed in idiopathic and inflammatory cases compared to familial ones. HCM patients had the most favourable outcome. Conclusion and outlook TORCH is the largest prospective study focusing on CMPs. We provided for the first time prospectively the clinical data of patients with diverse cardiomyopathies with outcome. Furthermore, comparing the different CMP forms on the clinical and molecular level will be an important step to enable translational research projects. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): German Centre for Cardiovascular Research (DZHK)


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
L.K Lewis ◽  
S.D Raudsepp ◽  
T.G Yandle ◽  
C.J Pemberton ◽  
R.N Doughty ◽  
...  

Abstract Background Heart failure (HF) is a leading cause of morbidity and mortality worldwide. Measurement of BNP and NTproBNP are used in HF for diagnosis and prognosis but levels of these peptides are inappropriately low in obesity, a condition which is associated with increased HF. Cleavage of proBNP to produce BNP and NT-proBNP requires proBNP to be unglycosylated at threonine 71 (T71). Gycosylation at T71 is affected by obesity, resulting in lower plasma NT-proBNP concentrations in patients with higher BMI. However the relationships between BMI, proBNP glycosylation and BNP (particularly the bioactive cardio-protective peptide BNP1-32) have not previously been described. Methods Validated in-house assays for BNP, BNP1-32, proBNP, proBNP unglycosylated at T71 (NG-T71) and the commercial Roche assay for NT-proBNP were applied to plasma samples obtained from patients with HF (n=321, PEOPLE study: Prospective Evaluation of Outcome in Patients with Left Ventricular Ejection Fraction). Results Median (IQR) concentrations of BNP, BNP1-32, proBNP, NG-T71 and NTproBNP were 10.7 (5–21), 5 (2–9), 27.8 (9–62), 6.2 (3–22) and 217 (104–425) pmol/L respectively. BMI was inversely related to NG-T71, NT-proBNP, BNP and BNP1-32 (r=−0.19, −0.40, −0.36 and −0.34 respectively, all p<0.01) but not proBNP (r=0.11, ns). ProBNP levels in patients with BMI above or below 30 kg/m2 were similar (29.8 (11.2–56.6) and 22.5 (3.9–65) pmol/L, p=0.51), whereas NG-T71, NT-proBNP, BNP and BNP1-32 levels were increased (p<0.001) in patients with BMI <30 (11.6 (3–25.6), 263 (153–486), 13.8 (6.5–25.5) and 6.3 (2.8–10.4)) compared to BMI >30 (3 (1–16), 127 (63–274), 7.8 (3–14) and 3.6 (1.1–7) respectively. The BMI >30 group had increased ProBNP:NT-proBNP, ProBNP:BNP and ProBNP:BNP1-32 ratios (all p<0.001) and proBNP:NG-T71 (p=0.037), whereas ratios of NG-T71 to BNP, BNP1-32 or NT-proBNP were not related to BMI. Patients with diabetes (n=90) also had lower BNP, BNP1-32 (both p<0.01), NG-T71 and NT-proBNP concentrations (both p<0.05), but not proBNP (p=0.46), and a trend towards a higher proBNP:BNP1-32 ratio (p=0.06). Discussion and conclusion The negative association between BMI and plasma NT-proBNP and BNP is not well understood. We recently reported that obese patients with HF have reduced circulating levels of proBNP unglycosylated at T71. In this expanded sample we show that whilst proBNP remains unaffected by BMI, both immunoreactive BNP and more specifically bioactive BNP1–32 levels, and NT-proBNP, are decreased with obesity in conjunction with increased glycosylation at T71. Increased glycosylation at proBNP-T71 reduces the amount of proBNP cleaved to form NT-proBNP and BNP resulting in decreased production and lowered circulating concentrations of these clinically used marker peptides. Our results provide a robust mechanism to explain the reduction in NT-proBNP and BNP levels observed in obese patients and confirm this is associated with reduced bioactive BNP1–32. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): National Heart Foundation of New Zealand, nHealth Research Council of New Zealand


2021 ◽  
Vol 22 (Supplement_1) ◽  
Author(s):  
SJ Backhaus ◽  
G Metschies ◽  
V Zieschang ◽  
J Erley ◽  
SM Zamani ◽  
...  

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): German Centre for Cardiovascular Research Purpose Myocardial Feature-Tracking (FT) deformation imaging is superior for risk-stratification compared to volumetric approaches. Since there is no clear recommendation regarding FT post-processing, we compared different FT-strain analyses with reference standard techniques, including tagging and strain encoded (SENC) magnetic resonance imaging. Methods FT software from 4 different vendors (TomTec/Medis/Circle(CVI)/Neosoft), tagging (Segment), and fastSENC (MyoStrain) were used to determine left ventricular global circumferential and longitudinal strains (GCS/GLS) in 12 healthy volunteers and 12 heart failure patients. Variability and agreements were assessed using intraclass correlation coefficients for absolute agreement (ICCa) and consistency (ICCc) as well as pearson correlation coefficients. Results For FT-GCS, consistency was excellent comparing different FT-vendors (ICCc = 0.84-0.97, r = 0.86-0.95) and compared to fSENC (ICCc = 0.78-0.89, r = 0.73-0.81). FT-GCS consistency was excellent compared to tagging (ICCc = 0.79-0.85, r = 0.74-0.77) except for TomTec (ICCc = 0.68, r = 0.72). Absolute FT-GCS agreements between FT-vendors were highest for CVI and Medis (ICCa = 0.96) and lowest for TomTec and Neosoft (ICCa = 0.32). Similarly, absolute FT-GCS agreements were excellent for CVI and Medis compared to both tagging and fSENC (ICCa = 0.84-0.88), good to excellent for Neosoft (ICCa = 0.77 and 0.64) and lowest for TomTec (ICCa = 0.41 and 0.47). For FT-GLS, consistency was excellent (ICCc≥0.86, r≥0.76). Absolute agreements between FT-vendors were excellent (ICCa = 0.91-0.93) or good to excellent for TomTec (ICCa = 0.69-0.85). Absolute agreements (ICCa) were good (CVI 0.70, Medis 0.60) and fair (TomTec 0.41, Neosoft 0.59) compared to tagging but excellent compared to fSENC (ICCa = 0.77-0.90). Conclusion Although absolute agreements differ depending on deformation assessment approaches, consistency and correlation are consistently high irrespective of the method chosen, thus indicating reliable strain assessment. Further standardisation and introduction of uniform references is warranted for routine clinical implementation.


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
V Parisi ◽  
S Cabaro ◽  
V D'Esposito ◽  
L Petraglia ◽  
M Conte ◽  
...  

Abstract Background The role of epicardial adipose tissue (EAT) in myocardial diseases is well established, and several evidence suggest that EAT may negatively affect left ventricular (LV) remodelling through an imbalanced production and secretion of pro and anti-inflammatory cytokines. Of these, the IL-13 it is known to play a positive activity on cardiac remodelling. Nowadays, the crosstalk between EAT and the myocardium is still poorly understood and the effects of myocardial ischemia on morphological and functional properties of EAT are almost unknown. Purpose In the present study we explored whether an increase of EAT thickness after STEMI might be associated with unfavourable LV remodelling at 3 months (T1). We also evaluated the relationship between changes (Δ) of EAT thickness and systemic levels of Interleukin (IL)-13 which is known to play a favourable activity on LV remodelling after STEMI. Methods We enrolled 66 patients with first STEMI, undergoing primary percutaneous angioplasty. At baseline and at 3 months we performed a complete echocardiogram, including EAT maximal thickness assessment, and determined circulating levels of IL-13. Results At 3 months after STEMI, the population was stratified into two groups according to different EAT remodelling after cardiac event: Group 1, patients with an increased EAT thickness (Δ EAT>1; 30 patients) and Group 2, patients with unchanged or decreased EAT thickness (Δ EAT<1). The two groups did not differ for age, gender and atherosclerotic risk factors. Group 1 had a worse LV remodelling at 3 months with higher LV diastolic and systolic volumes, lower LV ejection fraction (p=0.003; p=0.013; p=0.013 respectively) and worse diastolic function (E/e'; p=0.011). Of interest, EAT thickness increase was paralleled by circulating IL-13 reduction (p=0.022). Conclusion Myocardial injury can result in EAT increase which is associated to worse LV remodelling probably through the loss of the protective role of IL-13. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): This research has been funded by the University of Naples “Federico II” and “Compagnia di San Paolo e l'Istituto Banco di Napoli” within the competitive grant STAR 2018; Valentina Parisi is the principal investigator


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
A Riva ◽  
A Camporeale ◽  
F Sturla ◽  
S Pica ◽  
L Tondi ◽  
...  

Abstract Background Ischemic cardiomyopathy (ICM) is often associated with negative LV remodelling after myocardial infarction, sometimes resulting in impaired LV function and dilation (iDCM). 4D Flow CMR has been recently exploited to assess intracardiac hemodynamic changes in presence of LV remodelling. Purpose To quantify 4D Flow intracardiac kinetic energy (KE) and viscous energy loss (EL) and investigate their relation with LV dysfunction and remodelling. Methods Patients with prior anterior myocardial infarction underwent a CMR study with 4D Flow sequences acquisition; they were divided into ICM (n=10) and iDCM (n=10, EDV>208 ml and EF<40%). 10 controls were used for comparison. LV was semi-automatically segmented using short axis CMR stacks and co-registered with 4D Flow. Global KE and EL were computed over the cardiac cycle. NT-proBNP measurements were correlated with average and peak values, during systole and diastole. Results Both LV volume and EF significantly differ (P<0.0001) between iDCM (EDV=294±56 ml, EF=24±8%), ICM (EDV=181±32 ml, EF=34±6%) and controls (EDV=124±29 ml, EF=72±5%). If compared to controls, both ICM and iDCM showed significantly lower KE (P≤0.0008); though lower than controls, EL was higher in iDCM than ICM. Within the iDCM subgroup, diastolic mean KE and peak EL reported good inverse correlation with NT-proBNP (r=−0.75 and r=−0.69, respectively). EL indexed (ELI) to average KE during systole was higher in the entire ischemic group as compared to controls (ELI(ischemic) = 0.17 vs. ELI(controls) = 0.10, P=0.0054). Conclusions 4D Flow analyses effectively mapped post-ischemic LV energetic changes, highlighting the disproportionate intraventricular EL relative to produced KE; preliminary good correlation between LV energetic changes and NT-proBNP will deserve further investigation in order to contribute to early detection of heart failure. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): Italian Ministry of Health


Author(s):  
Jens Sörensen ◽  
Jonny Nordström ◽  
Tomasz Baron ◽  
Stellan Mörner ◽  
Sven-Olof Granstam ◽  
...  

Abstract Aim To develop a method for diagnosing left ventricular (LV) hypertrophy from cardiac perfusion 15O-water positron emission tomography (PET). Methods We retrospectively pooled data from 139 subjects in four research cohorts. LV remodeling patterns ranged from normal to severe eccentric and concentric hypertrophy. 15O-water PET scans (n = 197) were performed with three different PET devices. A low-end scanner (66 scans) was used for method development, and remaining scans with newer devices for a blinded evaluation. Dynamic data were converted into parametric images of perfusable tissue fraction for semi-automatic delineation of the LV wall and calculation of LV mass (LVM) and septal wall thickness (WT). LVM and WT from PET were compared to cardiac magnetic resonance (CMR, n = 47) and WT to 2D-echocardiography (2DE, n = 36). PET accuracy was tested using linear regression, Bland–Altman plots, and ROC curves. Observer reproducibility were evaluated using intraclass correlation coefficients. Results High correlations were found in the blinded analyses (r ≥ 0.87, P < 0.0001 for all). AUC for detecting increased LVM and WT (> 12 mm and > 15 mm) was ≥ 0.95 (P < 0.0001 for all). Reproducibility was excellent (ICC ≥ 0.93, P < 0.0001). Conclusion 15O-water PET might detect LV hypertrophy with high accuracy and precision.


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
K.V Bunting ◽  
S Gill ◽  
A Sitch ◽  
S Mehta ◽  
K O'Connor ◽  
...  

Abstract Introduction Echocardiography is essential for the management of patients with atrial fibrillation (AF), but current methods are time consuming and lack any evidence of reproducibility. Purpose To compare conventional averaging of consecutive beats with an index beat approach, where systolic and diastolic measurements are taken once after two prior beats with a similar RR interval (not more than 60 ms difference). Methods Transthoracic echocardiography was performed using a standardized and blinded protocol in patients enrolled into the RAte control Therapy Evaluation in permanent AF randomised controlled trial (RATE-AF; NCT02391337). AF was confirmed in all patients with a preceding 12-lead ECG. A minimum of 30-beat loops were recorded. Left ventricular function was determined using the recommended averaging of 5 and 10 beats and using the index beat method, with observers blinded to clinical details. Complete loops were used to calculate the within-beat coefficient of variation (CV) and intraclass correlation coefficient (ICC) for Simpson's biplane left ventricular ejection fraction (LVEF), global longitudinal strain (GLS) and filling pressure (E/e'). Results 160 patients (median age 75 years (IQR 69–82); 46% female) were included, with median heart rate 100 beats/min (IQR 86–112). For LVEF, the index beat had the lowest CV of 32% compared to 51% for 5 consecutive beats and 53% for 10 consecutive beats (p&lt;0.001). The index beat also had the lowest CV for GLS (26% versus 43% and 42%; p&lt;0.001) and E/e' (25% versus 41% and 41%; p&lt;0.001; see Figure for ICC comparison). Intra-operator reproducibility, assessed by the same operator from two different recordings in 50 patients, was superior for the index beat with GLS bias −0.5 and narrow limits of agreement (−3.6 to 2.6), compared to −1.0 for 10 consecutive beats (−4.0 to 2.0). For inter-operator variability, assessed in 18 random patients, the index beat also showed the smallest bias with narrow confidence intervals (CI). Using a single index beat did not impact on the validity of LVEF, GLS or E/e' measurement when correlated with natriuretic peptides. Index beat analysis substantially shortened analysis time; 35 seconds (95% CI 35 to 39 seconds) for measuring E/e' with the index beat versus 98 seconds (95% CI 92 to 104 seconds) for 10 consecutive beats (see Figure). Conclusion Index beat determination of left ventricular function improves reproducibility, saves time and does not compromise validity compared to conventional quantification in patients with heart failure and AF. After independent validation, the index beat method should be adopted into routine clinical practice. Comparison for measurement of E/e' Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): National Institute of Health Research UK


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
T Koopsen ◽  
N Van Osta ◽  
E Willemen ◽  
F.A Van Nieuwenhoven ◽  
J Gorcsan ◽  
...  

Abstract Background/Introduction The mechanical properties of infarcted myocardium are important determinants of cardiac pump function and risk of developing heart failure following myocardial infarction (MI). Purpose To better understand the effects of infarct stiffness on compensatory hypertrophy and dilation of non-infarcted tissue in the left (LV) and right ventricle (RV), by using a computational model. Methods The CircAdapt computational model of the human heart and circulation was applied to simulate an acute MI involving 20% of LV wall mass. The simulation was validated using previously published experimental data. Subsequently, two degrees of increased infarct stiffness were simulated. In all three simulations, a model of structural myocardial adaptation of the non-infarcted tissue was applied, based on sensing of mechanical loading of myocytes and extracellular matrix (ECM). Results Mild and severe stiffening of the infarct reduced the increase of LV end-diastolic volume (EDV) from +23 mL to +17 mL and +16 mL, respectively, and the increase of LV non-infarcted tissue mass from +31% to +21% and +18%. RV EDV decreased after adaptation, and mild and severe infarct stiffening reduced the decrease of RV EDV from −21 mL to −12 mL and −10 mL, respectively. Increase of RV tissue mass was reduced from +13% to +8% and +7% with mild and severe infarct stiffening. In the LV, reduced dilation and hypertrophy were driven mainly by a reduction of maximum stress in the ECM and a higher stress between the myocytes and ECM following infarct stiffening. The decreased RV hypertrophy, but not EDV reduction, was caused by a reduction of maximum RV ECM stress and maximum RV active myofiber stress. Conclusions Model simulations predicted that a stiffened LV infarct reduces both LV and RV non-infarcted tissue hypertrophy as well as LV dilation. In LV remodeling, maximum ECM stress and stress between myocyte and ECM played a more prominent role than in RV remodeling, while maximum active stress was more important in the RV. Overview of all model simulations Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): This work was funded by the Netherlands Organisation for Scientific Research and the Dutch Heart Foundation.


2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
M Bjorkavoll-Bergseth ◽  
B Auestad ◽  
O Kleiven ◽  
O Skadberg ◽  
T Eftestol ◽  
...  

Abstract Background/Introduction Following prolonged strenuous exercise there is an exercise-induced troponin (cTn) elevation in healthy individuals. The precise mechanisms and clinical consequence of this cTn elevation remain to be determined. It has recently been demonstrated that exercise intensity, exceeding a heart rate (HR) of 150 bpm, is correlated with exercise-induced cTn elevation. Purpose The present work aims to determine if there is a threshold for exercise duration with a HR exceeding 150 bpm associated with an excessive exercise-induced cTn elevation. Methods A total of 177 healthy subjects were included in the present analysis of HR data obtained from sport watches used during a 91-km recreational mountain bike cycle race. Clinical status, cTnI, ECGs, blood pressure and demographics were obtained 24 h prior to- and at 3 h and 24 h after the race. Results are reported as median and 25th and 75th percentile. We used Tree regression to determine the association between elevated cTnI and exercise duration exceeding a HR of 150 bpm. Results Subjects were 82% (n=146) males, 44 (39–51) years, with a race time of 3.5 (3.1–3.9) h. Baseline cTnI was 1.9 (1.6–3.3) ng/L. There was a cTnI elevation in all study participants at 3 h, cTnI: 60.0 (36.0–99.3) ng/L, with a significant (p&lt;0.001) reduction at 24 hours following exercise, cTnI: 10.9 (6.1–22.4) ng/L. Tree regression identified 168 min of exercise, with a HR exceeding 150 bpm, to be associated with an excessive increase in cTnI both at 3 h, and at 24 h following the race (figure). The median cTn values above and below the threshold are presented in the Table. Conclusion The present analysis suggests that exceeding a specific duration of high intensity exercise may be associated with excessive cTn elevation in susceptible individuals. Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): Western Norway Health authoritites.


Sign in / Sign up

Export Citation Format

Share Document