CIUS: Centre for Innovative Ultrasound Solutions (completed)

• Talberg, Andreas Sørbrøden; Golfetto, Cristiana; Johansen, Tonni Franke; Avdal, Jørgen; Ekroll, Ingvild Kinn & Dong, Hefeng [Show all 7 contributors for this article] (2026). Enhanced Flow Detection through a Steel Barrier using Doppler Ultrasound: a Numerical and Experimental Study. Ultrasonics. ISSN 0041-624X. 162. doi: 10.1016/j.ultras.2025.107922. Full text in Research Archive Show summary

  Uncontrolled leakages of hydrocarbons from oil and gas wells can have vast environmental consequences and lead to huge costs for the companies responsible. Thus, preventing and detecting potential leakages in wells is of great importance. In this work, Doppler ultrasound was used for the detection of flow behind steel to help assess the integrity of oil and gas wells. The main goal was to enhance flow detection with Doppler ultrasound through steel by improving the beamforming of the ultrasonic pulse. A naive beamforming approach assuming the medium to be homogeneous was compared to a ray tracing based beamforming algorithm, both in a numerical study and in an experimental setup. Simulations showed a 5 dB increase of the peak amplitude when taking refraction in the steel into account by applying the ray tracing based beamforming. In addition, the targeted focus was hit with a narrower, more precisely controlled beam on the far side of the steel layer. These two beamforming techniques were then employed and compared for flow detection behind steel in an experimental setup. Two cases were investigated, low velocity and high velocity, using three different transmit voltages for each velocity. Pulsed wave Doppler spectra were displayed, and the ratio between the flow signal power averaged over a range of frequencies was calculated. An increase of 3-4 dB was observed using the ray tracing technique, showing enhanced flow detection compared to the naive technique.

• Chernyshov, Artem; Nyberg, John Anders Tomas; Holmstrøm, Vegard; Azad, Md Abulkalam; Grenne, Bjørnar Leangen & Dalen, Håvard [Show all 9 contributors for this article] (2025). Low Complexity Point Tracking of the Myocardium in 2D Echocardiography. IEEE Access. 13, p. 186992–187004. doi: 10.1109/ACCESS.2025.3627527. Full text in Research Archive Show summary

  Deep learning methods for point tracking are applicable in 2D echocardiography, but do not yet take advantage of domain specifics that enable extremely fast and efficient configurations. We developed MyoTracker, a low-complexity architecture (0.3M parameters) for point tracking in echocardiography. It builds on the CoTracker2 architecture by simplifying its components and extending the temporal context to provide point predictions for the entire sequence in a single step. We applied MyoTracker to the right ventricular (RV) myocardium in RV-focused recordings and compared the results with those of CoTracker2 and EchoTracker, another specialized point tracking architecture for echocardiography. MyoTracker achieved the lowest average point trajectory error at 2.00±0.53 mm. Calculating RV Free Wall Strain using MyoTracker’s point predictions resulted in a −0.3% bias with 95% limits of agreement from −6.1% to 5.4% compared to reference values from commercial software. This range falls within the interobserver variability reported in previous studies. The limits of agreement were wider for both CoTracker2 and EchoTracker, worse than the interobserver variability. At inference, MyoTracker used 67% less GPU memory than CoTracker2 and 84% less than EchoTracker on large sequences (100 frames). MyoTracker was 74 times faster during inference than CoTracker2 and 11 times faster than EchoTracker with our setup. Maintaining the entire sequence in the temporal context was the greatest contributor to MyoTracker’s accuracy. Slight additional gains can be made by re-enabling iterative refinement, at the cost of longer processing time. MyoTracker source code: https://github.com/artemcher/myotracker

• Pasdeloup, David; Østvik, Andreas; Olaisen, Sindre Hellum; Skogvoll, Eirik; Dalen, Havard & Løvstakken, Lasse (2025). Challenges and Strategies for Deep Learning in Cardiovascular Imaging. JACC Cardiovascular Imaging. ISSN 1936-878X. 18(7), p. 751–764. doi: 10.1016/j.jcmg.2025.02.011. Full text in Research Archive Show summary

  Background Automated measurements in cardiac imaging with the use of deep learning (DL) is a highly active area of research and innovation. However, some concerns challenge the translation of DL methods from research to clinical implementation. Objectives The authors evaluated 3 challenges for cardiac measurements by DL using left ventricular ejection fraction (LVEF) for management of heart failure and discuss mitigation strategies. Methods Using 3 different populations (N = 3,538), automated LVEF measurements were obtained with the use of supervised end-to-end learning and analyzed in terms of HF management. Three common challenges related to evaluation metrics, training data, and model generalization were studied. Results For the evaluation challenge, the authors identified significant unreliability of the AUC when applied to dichotomized heart failure diagnosis. Specifically, AUC varied from 0.71 to 0.98 owing solely to changes in population characteristics. For the training data challenge, model performance could be enhanced even after reducing the number of training subjects by 40%. For the generalization challenge, a performance degradation was observed compared with internal data when testing the model on external data. Integrating medical imaging domain knowledge in the DL framework effectively helped to recover performance and improve generalizability. Conclusions Both training data and generalization aspects challenge the performance of DL algorithms for automated cardiac measurements. In addition, evaluation metrics challenge the ability to detect underperforming algorithms. By considering evaluation metrics and training data distribution, and incorporating imaging domain knowledge, the design and evaluation of DL models can be improved, leading to more robust models, improved interpretation, and easier comparison across data sets. These findings may guide researchers and clinicians in implementing DL models for cardiovascular imaging.

• Le, Hoang Duy; Savoia, Alessandro Stuart; Manh, Tung; Nguyen, Hoang-Vu & Hoff, Lars (2025). Fabrication of a piezoelectric array for use in a dual frequency hybrid piezoelectric—CMUT probe. Microsystem Technologies : Micro- and Nanosystems Information Storage and Processing Systems. ISSN 0946-7076. 31, p. 2641–2656. doi: 10.1007/s00542-025-05892-3. Full text in Research Archive Show summary

  Wideband transducers are highly preferable for medical ultrasound imaging. Dual-frequency transducers have been developed to enlarge the operating bandwidth by separating transmission and reception. A novel dual-frequency transducer design, comprising a hybrid configuration of a partly epoxy-filled kerf piezoelectric array and a capacitive micromachined ultrasound transducer (CMUT) array, has been proposed. This work studied the fabrication feasibility such hybrid transducer, including both fabrication of the partly epoxy-filled piezoelectric array and lamination of inactive CMUTs to this array to mimic the hybrid transducer. The bonding properties of piezoelectric elements on a flexible printed circuit board were characterized through both mechanical and electrical methods. The adhesion strength of elements to the flexible printed circuit board demonstrates robustness, with a peel strength reaching up to 2.7 N/mm. Additionally, the contact resistance of an element to the board is negligibly small, approximately 0.11 Ω. Four prototype arrays, each consisting of 80 elements with air-filled kerfs, were successfully fabricated, achieving a yield ratio up to 100%. The lamination process was successfully performed, resulting in a shear strength above 4.2 MPa and maintaining an epoxy fill less than 30% in the kerfs.

• Diez, Anja; Viggen, Erlend Magnus & Johansen, Tonni Franke (2025). Ultrasonic pulse-echo dataset from numerical modelling for oil and gas well integrity investigations. Scientific Data. 12. doi: 10.1038/s41597-025-04851-x. Full text in Research Archive Show summary

  The ultrasonic pulse-echo (PE) measurement is a crucial measurement technique to determine the integrity of oil and gas wells. Oil companies use various analysis techniques and corrections to derive the pipe thickness and impedance of the material behind the pipe from PE measurements that are carried out inside the pipe. While some field measurements are publicly available, they have no corresponding ground truth. We therefore simulated a dataset of PE measurements with ground truth. The dataset was generated using axisymmetric models and 3D models in COMSOL Multiphysics. The base geometry was based on common parameters from the field: oil-based mud on the inside of a 9.625 in pipe and cement on the outside of the pipe. From this base geometry, variations in the model parameters were introduced, for example, plate/pipe wall thickness, different materials on both sides of the wall, different pipe diameter, different annulus thicknesses, eccentering. The generated dataset allows detailed investigations of existing PE analysis algorithms, comparison of those and development of new PE analysis techniques.

• Vyver, Gilles van de; Thomas, Sarina; Ben-Yosef, Guy; Olaisen, Sindre Hellum; Dalen, Håvard & Løvstakken, Lasse [Show all 7 contributors for this article] (2024). Toward Robust Cardiac Segmentation Using Graph Convolutional Networks. IEEE Access. 12, p. 33876–33888. doi: 10.1109/ACCESS.2024.3373046. Full text in Research Archive Show summary

  Fully automatic cardiac segmentation can be a fast and reproducible method to extract clinical measurements from an echocardiography examination. The U-Net architecture is the current state-of-the-art deep learning architecture for medical segmentation and can segment cardiac structures in real-time with average errors comparable to inter-observer variability. However, this architecture still generates large outliers that are often anatomically incorrect. This work uses the concept of graph convolutional neural networks that predict the contour points of the structures of interest instead of labeling each pixel. We propose a graph architecture that uses two convolutional rings based on cardiac anatomy. While this architecture does not improve performance on classical measures like Dice score and Hausdorff distance, it does eliminate anatomical incorrect segmentations. Additionally, we propose to use the inter-model agreement of the U-Net and the graph network as a predictor of both the input and segmentation quality in real-time. The results show that from the 100 high agreement samples, 93 were in distribution, while from the 100 low agreement samples, only 7 were in distribution. Finally, this work contributes with an ablation study of the graph convolutional architecture on the publicly available CAMUS dataset and an evaluation of clinical measurements on the clinical HUNT4 dataset. Source code is available online: https://github.com/gillesvntnu/GCN_multistructure

• Giri, Samita; Tronvik, Erling Andreas; Dalen, Håvard; Ellekjær, Hanne; Olsen, Alexander & Hagen, Knut (2024). Headache disorders and risk of stroke. A register-linked HUNT study. Cephalalgia Reports. 7. doi: 10.1177/25158163241295735. Full text in Research Archive Show summary

  Background: Many studies have found that migraine with aura (MA) is a risk factor of stroke, but the relationship between stroke and migraine without aura and tension-type headache is unclear. The aim of this study was to investigate whether primary headache disorders, including subtypes of migraine, increase the risk of ischemic or hemorrhagic stroke. Methods: This large population-based 13-year follow-up study used baseline headache data from the third Trøndelag Health Study (HUNT3) performed between 2006 and 2008. The HUNT3 headache data were linked to the Norwegian National Stroke Register that includes stroke diagnoses recorded from 2013 until December 2021. The association between stroke and headache status was investigated in individuals aged ≥20 years without stroke at baseline. Prospective associations were evaluated using multivariable Cox proportional hazard models with 95% confidence intervals. Separate sub-group analyses by age and sex were performed. Results: Among 37,364 included participants, 50% were younger than 55 years. A total of 1095 (2.9%) developed stroke after a median of 9 years. In the multi-adjusted model, reporting MA at baseline was associated with increased risk of stroke at follow-up (HR 1.59, 95% CI 1.14–2.21) compared with those without headache. The increased risk of stroke was most evident among individuals <55 years with MA (HR 2.18, 95% CI 1.24–3.82) and among women (HR 1.70, 95% CI 1.12–2.59). Migraine without aura (MO), tension-type headache (TTH) and unclassified headaches were not associated with increased risk of stroke. Conclusions: During 15 years of follow-up, individuals with MA were more likely to suffer from stroke compared to those without headache. The relationship with MA was even stronger in women, and for young individuals aged <55 years. Individuals with MO or TTH were not associated with increased risk of stroke.

• Gundersen, Erlend Løland; Smistad, Erik; Jahren, Tollef & Måsøy, Svein-Erik (2024). Hardware-Independent Deep Signal Processing: A Feasibility Study in Echocardiography. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control. ISSN 0885-3010. 71(11), p. 1491–1500. doi: 10.1109/TUFFC.2024.3404622. Full text in Research Archive Show summary

  Deep learning (DL) models have emerged as alternative methods to conventional ultrasound (US) signal processing, offering the potential to mimic signal processing chains, reduce inference time, and enable the portability of processing chains across hardware. This paper proposes a DL model that replicates the fine-tuned BMode signal processing chain of a high-end US system and explores the potential of using it with a different probe and a lower-end system. A deep neural network was trained in a supervised manner to map raw beamformed in-phase and quadrature component data into processed images. The dataset consisted of 30,000 cardiac image frames acquired using the GE HealthCare Vivid E95 system with the 4Vc-D matrix array probe. The signal processing chain includes depth-dependent bandpass filtering, elevation compounding, frequency compounding, and image compression and filtering. The results indicate that a lightweight DL model can accurately replicate the signal processing chain of a commercial scanner for a given application. Evaluation on a 15 patient test dataset of about three thousand image frames gave a structural similarity index measure of 98.56 ± 0.49. Applying the DL model to data from another probe showed equivalent or improved image quality. This indicates that a single DL model may be used for a set of probes on a given system that targets the same application, which could be a cost-effective tuning and implementation strategy for vendors. Further, the DL model enhanced image quality on a Verasonics dataset, suggesting the potential to port features from high-end US systems to lower-end counterparts.

• Shah, Amil M.; Myhre, Peder Langeland; Arthur, Victoria; Dorbala, Pranav; Rasheed, Humaira & Buckley, Leo F. [Show all 22 contributors for this article] (2024). Large scale plasma proteomics identifies novel proteins and protein networks associated with heart failure development. Nature Communications. 15(1). doi: 10.1038/s41467-023-44680-3. Full text in Research Archive Show summary

  Heart failure (HF) causes substantial morbidity and mortality but its pathobiology is incompletely understood. The proteome is a promising intermediate phenotype for discovery of novel mechanisms. We measured 4877 plasma proteins in 13,900 HF-free individuals across three analysis sets with diverse age, geography, and HF ascertainment to identify circulating proteins and protein networks associated with HF development. Parallel analyses in Atherosclerosis Risk in Communities study participants in mid-life and late-life and in Trøndelag Health Study participants identified 37 proteins consistently associated with incident HF independent of traditional risk factors. Mendelian randomization supported causal effects of 10 on HF, HF risk factors, or left ventricular size and function, including matricellular (e.g. SPON1, MFAP4), senescence-associated (FSTL3, IGFBP7), and inflammatory (SVEP1, CCL15, ITIH3) proteins. Protein co-regulation network analyses identified 5 modules associated with HF risk, two of which were influenced by genetic variants that implicated trans hotspots within the VTN and CFH genes.

• Dalen, Håvard; Letnes, Jon Magne; Høydal, Morten Andre & Wisløff, Ulrik (2024). Diastolic function and dysfunction in athletes. European Heart Journal-Cardiovascular Imaging. ISSN 2047-2404. 25(11), p. 1537–1545. doi: 10.1093/ehjci/jeae155. Full text in Research Archive
• Kavlie, Trym Løvseth; Kildahl, Henrik Agerup; Dalen, Håvard; Nordhaug, Dag Ole; Slagsvold, Katrine Hordnes & Grenne, Bjørnar Leangen [Show all 7 contributors for this article] (2024). Five-year outcomes of mitral valve repair for leaflet prolapse at a medium-sized Norwegian university hospital. Scandinavian Cardiovascular Journal. ISSN 1401-7431. 58(1). doi: 10.1080/14017431.2024.2379336. Full text in Research Archive
• Måsøy, Svein-Erik; Dénarié, Bastien Emmanuel; Sørnes, Anders Rasmus; Holte, Espen; Grenne, Bjørnar Leangen & Espeland, Torvald [Show all 10 contributors for this article] (2024). Aberration correction in 3D transthoracic echocardiography. WFUMB Ultrasound Open. 2(2). doi: 10.1016/j.wfumbo.2024.100062. Full text in Research Archive
• Wifstad, Sigurd Vangen; Kildahl, Henrik Agerup; Holte, Espen; Berg, Erik Andreas Rye; Grenne, Bjørnar Leangen & Salvesen, Øyvind Olav [Show all 8 contributors for this article] (2024). EasyPISA: Automatic Integrated PISA Measurements of Mitral Regurgitation From 2-D Color-Doppler Using Deep Learning. Ultrasound in Medicine and Biology. ISSN 0301-5629. doi: 10.1016/j.ultrasmedbio.2024.06.008. Full text in Research Archive
• Hall, Martina; Skogholt, Anne Heidi; Surakka, Ida; Dalen, Håvard & Almaas, Eivind (2024). Genome-wide association studies reveal differences in genetic susceptibility between single events vs. recurrent events of atrial fibrillation and myocardial infarction: the HUNT study. Frontiers in Cardiovascular Medicine. 11. doi: 10.3389/fcvm.2024.1372107. Full text in Research Archive Show summary

  Genetic research into atrial fibrillation (AF) and myocardial infarction (MI) has predominantly focused on comparing afflicted individuals with their healthy counterparts. However, this approach lacks granularity, thus overlooking subtleties within patient populations. In this study, we explore the distinction between AF and MI patients who experience only a single disease event and those experiencing recurrent events. Integrating hospital records, questionnaire data, clinical measurements, and genetic data from more than 500,000 HUNT and United Kingdom Biobank participants, we compare both clinical and genetic characteristics between the two groups using genome-wide association studies (GWAS) meta-analyses, phenome-wide association studies (PheWAS) analyses, and gene co-expression networks. We found that the two groups of patients differ in both clinical characteristics and genetic risks. More specifically, recurrent AF patients are significantly younger and have better baseline health, in terms of reduced cholesterol and blood pressure, than single AF patients. Also, the results of the GWAS meta-analysis indicate that recurrent AF patients seem to be at greater genetic risk for recurrent events. The PheWAS and gene co-expression network analyses highlight differences in the functions associated with the sets of single nucleotide polymorphisms (SNPs) and genes for the two groups. However, for MI patients, we found that those experiencing single events are significantly younger and have better baseline health than those with recurrent MI, yet they exhibit higher genetic risk. The GWAS meta-analysis mostly identifies genetic regions uniquely associated with single MI, and the PheWAS analysis and gene co-expression networks support the genetic differences between the single MI and recurrent MI groups. In conclusion, this work has identified novel genetic regions uniquely associated with single MI and related PheWAS analyses, as well as gene co-expression networks that support the genetic differences between the patient subgroups of single and recurrent occurrence for both MI and AF.

• Giri, Samita; Tronvik, Erling Andreas; Dalen, Håvard; Ellekjær, Hanne; Loennechen, Jan Pål & Olsen, Alexander [Show all 7 contributors for this article] (2024). Migraine and risk of atrial fibrillation: A 9-year follow-up based on the Trøndelag Health Study. Cephalalgia. ISSN 0333-1024. 44(5). doi: 10.1177/03331024241254517. Full text in Research Archive Show summary

  Background: Data from some population-based studies have indicated an increased risk of atrial fibrillation (AF) among patients with migraine, particularly among individuals with migraine with aura. The present study aimed to assess the association between primary headache disorders and AF. Methods: In a population-based 9-year follow-up design, we evaluated the questionnaire-based headache diagnosis, migraine and tension-type headache (TTH) included, collected in the Trøndelag Health Study (HUNT3) conducted in 2006–2008, and the subsequent risk of AF in the period until December 2015. The population at risk consisted of 39,340 individuals ≥20 years without AF at HUNT3 baseline who answered headache questionnaire during HUNT3. The prospective association was evaluated by multivariable Cox proportional hazard models with 95% confidence intervals (CIs). Results: Among the 39,340 participants, 1524 (3.8%) developed AF during the 9-year follow up, whereof 91% of these were ≥55 years. In the multivariable analyses, adjusting for known confounders, we did not find any association between migraine or TTH and risk of AF. The adjusted hazard ratios (HRs) were respectively 0.84 (95% CI = 0.64–1.11) for migraine, 1.16 (95% CI = 0.86–1.27) for TTH and 1.04 (95% CI = 0.86–1.27) for unclassified headache. However, in sensitivity analyses of individuals aged ≥55 years, a lower risk of AF was found for migraine (HR = 0.53; 95% CI = 0.39–0.73). Conclusions: In this large population-based study, no increased risk of AF was found among individuals with migraine or TTH at baseline. Indeed, among individuals aged ≥55 years, migraine was associated with a lower risk for AF.

• Viggen, Erlend Magnus; Diez, Anja & Johansen, Tonni Franke (2024). Pyintegrity: An Open-Source Toolbox for Processing Ultrasonic Pulse-Echo Well Integrity Log Data, SPE Norway Subsurface Conference, April 17, 2024, Bergen, Norway. Society of Petroleum Engineers. ISSN 9781959025344. doi: 10.2118/218476-MS. Full text in Research Archive Show summary

  Many companies offer similarly designed wireline tools using ultrasonic pulse-echo measurements toevaluate barrier integrity in cased-hole wells. While these tools provide very similar data, differentcompanies process their data using different algorithms, typically to estimate the pipe wall thickness andthe outer material's acoustic impedance. While the algorithms themselves are public, no openly availablesoftware implementations are available. Therefore, we have developed an open-source software toolboxcalled Pyintegrity implementing many of these algorithms. In this article, we demonstrate Pyintegrity byapplying its algorithm implementations to a well integrity log from the open Volve Data Village dataset. Ourresults demonstrate that it is quite possible to process data recorded by a particular tool using processingalgorithms developed for use with other similar tools, and we find a good correspondence between thedifferent processing algorithms. Comparing the results produced by the different processing algorithms letsus confidently identify certain features in some of the results as processing artifacts that do not reflect thephysical state of the well.

• Tasken, Anders Austlid; Yu, Jinyang; Berg, Erik Andreas Rye; Grenne, Bjørnar Leangen; Holte, Espen & Dalen, Håvard [Show all 10 contributors for this article] (2024). Automatic Detection and Tracking of Anatomical Landmarks in Transesophageal Echocardiography for Quantification of Left Ventricular Function. Ultrasound in Medicine and Biology. ISSN 0301-5629. doi: 10.1016/j.ultrasmedbio.2024.01.017. Full text in Research Archive
• Wangensteen, Magnus; Johansen, Tonni Franke; Fatemi, Ali & Viggen, Erlend Magnus (2024). Pipe wall thickness estimation by frequency–wavenumber analysis of circumferential guided waves. Mechanical systems and signal processing. ISSN 0888-3270. 215. doi: 10.1016/j.ymssp.2024.111369. Full text in Research Archive Show summary

  Ultrasonic guided waves in pipes propagating in the circumferential direction carry information about the thickness of the pipe wall. This study proposes a method for estimating the pipe wall thickness based on measurements from circumferentially distributed sensors and a set of pre-computed theoretical dispersion curves. The recorded data are Fourier transformed into a frequency–wavenumber representation. The wall thickness is obtained by identifying the best fitting dispersion curve to the frequency–wavenumber data by determining the maximum stacking power. The thickness estimation method has demonstrated robustness against noise. Given the presence of significant noise (0.0 dB signal-to-noise ratio) and simulations where the thickness ranges from 6.0 to 10.0 mm, using eight sensors or more attain a mean absolute percentage error of less than 1%. The method is also supported by real laboratory measurements using eight sensor positions, where the error is less than 0.7%. Experimental measurements near the pipe end demonstrate that with appropriate positioning, end-reflections should not have a detrimental effect on the accuracy. The method estimates the mean thickness in cases where the pipe wall has near-uniform thickness but may not be as reliable when there are large shallow defects caused by metal loss, which is typically the case when the inner wall has suffered from erosion. In this case, monitoring of the stacking power curves could detect regional metal loss due to erosion and analysis of the local maxima could uncover information about individual thickness segments around the pipe circumference.

• Viggen, Erlend Magnus & Arnestad, Håvard (2024). On the cutoff, reappearance, and persistence of subsonic leaky A0 Lamb waves. Journal of Physics: Conference Series (JPCS). ISSN 1742-6588. 2822. doi: 10.1088/1742-6596/2822/1/012168. Full text in Research Archive Show summary

  Subsonic leaky A0 Lamb waves, whose phase speeds are lower than the fluid's sound speed, exhibit several peculiar behaviours. In this work, we investigate three such behaviours: Cutoff, where the A0 branch cuts off at some point in the subsonic domain, reappearance, where the cut-off A0 branch reappears at lower frequencies, and persistence, where the A0 branch is never cut off at all. From the literature, we take the case of a 1 mm thick aluminium plate immersed in air of varying density. We then analyse its exact leaky A0 wave solutions by means of a sound radiation model that takes energy conservation into account. This analysis explains these peculiar behaviours by connecting them with the conditions that allow sound-radiating subsonic leaky A0 Lamb waves.

• Liu, Xueqing; Ytterdal, Trond & Shur, Michael S. (2024). Compact Spice Models for Terafets. International Journal of High Speed Electronics and Systems. ISSN 0129-1564. 33(1 ). doi: 10.1142/S0129156424500046. Full text in Research Archive
• Wangensteen, Magnus; Johansen, Tonni Franke; Fatemi, Ali; Viggen, Erlend Magnus & Haugan, Lars Eidissen (2024). Pitting Detection and Characterization From Ultrasound Timelapse Images Using Convolutional Neural Networks. IEEE Open Journal of Instrumentation and Measurement (OJIM). 3. doi: 10.1109/OJIM.2024.3396829. Full text in Research Archive Show summary

  Pitting corrosion, a localized form of corrosion leading to cavities and structural failure in metallic materials, requires early detection for effective mitigation. While ultrasonic inspection techniques can readily detect uniform wall thinning, they often struggle to identify pitting corrosion. This study proposes a time-lapse ultrasound inspection method to detect early-stage pitting using pulse-echo sensors. By recording multiple ultrasonic traces over time, 2-D timelapse images of ultrasonic reflectivity can be generated and fed into a trained neural network for pitting diagnostics. In general, training a machine-learning model requires a large training dataset. This work used data from a drilling experiment to generate a suitable dataset. Dataset construction by random time-ordered combinations of ultrasonic measurements was conducted to create a diverse set of time-lapse image samples to generalize the resulting machine-learning model adequately. A classification neural network was trained to detect the presence of drilled holes, and a separate regression network was trained to estimate the hole depth. Based on drilling data from an independently acquired test dataset, results demonstrate a mean absolute error of 0.163 mm for hole depth estimations. All holes are successfully detected when 0.1 mm deeper than the defined pitting threshold of 0.5 mm. This suggests that the proposed method generalizes well and can be deployed to any similar acquisition system.

• Wifstad, Sigurd Vangen; Kildahl, Henrik Agerup; Grenne, Bjørnar Leangen; Holte, Espen; Hauge, Ståle Wågen & Sæbø, Sigbjørn [Show all 12 contributors for this article] (2024). Mitral Valve Segmentation and Tracking from Transthoracic Echocardiography Using Deep Learning. Ultrasound in Medicine and Biology. ISSN 0301-5629. doi: 10.1016/j.ultrasmedbio.2023.12.023. Full text in Research Archive
• Huynh, Thong; Eggen, Trym Haakon & Hoff, Lars (2024). Investigation of Non-Linearities in Medical Ultrasound Imaging Probes by Characterizing Free and Clamped Capacitances. IEEE Open Journal of Ultrasonics, Ferroelectrics, and Frequency Control Society (OJUFFC). 4, p. 116–125. doi: 10.1109/OJUFFC.2024.3459858. Full text in Research Archive Show summary

  Tissue harmonic imaging requires good control of the nonlinearity in the ultrasound probe, as transmitted second harmonics from the probe may interfere with tissue harmonics and degrade image quality. We have studied the nonlinearity in four different medical ultrasound probes by measuring the capacitive part of their electrical impedances under varying electric fields, at frequencies well below and above the resonances. The probes were made with two different piezoelectric materials, piezoceramic PZT and single-crystal PMN-PT, with either soft backing operating at half-wavelength resonance or hard backing operating at quarter-wavelength resonance. When the applied electric field amplitude E was increased from 0.05 V/μm to 0.5 V/μm , we observed an increase in both the capacitance at high frequency, interpreted as clamped conditions, and at low frequency, interpreted as free conditions. This is a nonlinear phenomenon as these capacitances will not change in the linear regime. The increase in free capacitance was from 4 to 10 times larger than the increase in clamped capacitance for all the investigated probes. This indicates a stronger nonlinearity for the free capacitance. At the low-frequency excitation, we observed distortion in the current passing through the acoustic stack corresponding to a relative second harmonic level of −20 dB. We conclude that the nonlinear impedance of the acoustic stack in the investigated probes was primarily caused by nonlinearities in the mechanical coefficients, while contributions from dielectric nonlinearity were negligible.

• Letnes, Jon Magne; Nes, Bjarne Martens; Sandbakk, Øyvind Bucher; Tjønna, Arnt Erik; Fremo, Thomas & Moldjord, Christian [Show all 9 contributors for this article] (2024). Comparison of resting heart rate and left ventricular ejection fraction in elite endurance athletes and the general population. European Journal of Preventive Cardiology (EJPC). ISSN 2047-4873. doi: 10.1093/eurjpc/zwae294. Full text in Research Archive
• Le, Hoang Duy; Manh, Tung & Hoff, Lars (2024). Lamination of Capacitive Micromachined Ultrasonic Transducer on a Piezoelectric Array: Process and Evaluation, 24th European Microelectronics & Packaging Conference (EMPC 2023). IEEE (Institute of Electrical and Electronics Engineers). ISSN 9780956808691. doi: 10.23919/EMPC55870.2023.10418279. Full text in Research Archive Show summary

  Harmonic imaging plays a crucial role in medical ultrasound imaging, where the transducer probes are required to emit low-frequency ultrasonic waves into the tissues, and receive reflected signals at the higher, typically the second harmonic of the transmitted wave. We have designed a unique probe with a high-frequency capacitive micromachined ultrasonic transducer (CMUT) on top of a low-frequency piezoelectric array for harmonic imaging. This paper presents a lamination process between CMUT and piezoelectric array using epoxy. The bond-line between the two components was thoroughly investigated using mechanical and electrical methods. The resulting bond-line thickness was less than 2 um, and shear strength fell within the range 4.8 - 5.1 MPa. Additionally, electrical impedance measurement of all elements in the low-frequency array exhibited resonance shifts before and after bonding, providing further insight into the bonding process.

• Sæbø, Sigbjørn; Dalen, Håvard; Nyberg, John Anders Tomas; Grenne, Bjørnar Leangen; Jakobsen, Even Olav & Nes, Bjarne Martens [Show all 8 contributors for this article] (2024). Left atrial to ventricular volume ratio and relation to fitness, cardiovascular risk factors, and diastolic function in healthy individuals: the HUNT Study. European Heart Journal – Imaging Methods and Practice (EHJ-IMP). doi: 10.1093/ehjimp/qyae028. Full text in Research Archive
• Rye, Cathrine Sæthern; Ofstad, Anne Pernille; Åsvold, Bjørn Olav; Romundstad, Pål Richard; Horn, Julie & Dalen, Håvard (2024). The influence of diagnostic subgroups, patient- and hospital characteristics for the validity of cardiovascular diagnoses–Data from a Norwegian hospital trust. PLOS ONE. 19(4), p. 1–14. doi: 10.1371/journal.pone.0302181. Full text in Research Archive Show summary

  Background: Cardiovascular discharge diagnoses may serve as endpoints in epidemiological studies if they have a high validity. Aim was to study if diagnoses-specific characteristics like type, sub-categories, and position of cardiovascular diagnoses affected diagnostic accuracy. Methods: Patients (n = 7,164) with a discharge diagnosis of acute myocardial infarction, heart failure or cerebrovascular disease were included. Data were presented as positive predictive values (PPV) and sensitivity. Results: PPV was high (≥88%) for acute myocardial infarction (n = 2,189) (except for outpatients). For heart failure (n = 4,026) PPV was 67% overall, but higher (>99%) when etiology or echocardiography was included. For hemorrhagic (n = 257) and ischemic (n = 1,034) strokes PPVs were 87% and 80%, respectively, with sensitivity of 79% and 75%. Transient ischemic attacks (n = 926) had PPV 56%, but sensitivity 86%. Primary diagnoses showed higher validity than subsequent diagnoses and inpatient diagnoses were more valid than outpatient diagnoses (except for transient ischemic attack). The diagnoses of acute myocardial infarction and heart failure where most valid when placed at cardiology units, while ischemic stroke when discharged from an internal medicine unit. Conclusions: The diagnoses of acute myocardial infarction and stroke had excellent validity when placed during hospital stays. Similarly, heart failure diagnoses had excellent validity when echocardiography was performed before placing the diagnosis, while overall the diagnoses of heart failure and transient ischemic attack were less valid. In conclusion, the results indicate that cardiovascular diagnoses based on objective findings such as acute myocardial infarction and stroke have excellent validity and may be used as endpoints in clinical epidemiological studies with less rigid validation.

• Mansoori, Amirfereydoon; Salmani, Hamed; Hanke, Ulrik; Hoff, Lars & Halvorsen, Einar (2024). Electroacoustic Modeling of Patterned Piezoelectric Micromachined Ultrasonic Transducers. IEEE Sensors Journal. ISSN 1530-437X. 24(12), p. 18859–18871. doi: 10.1109/JSEN.2024.3391655. Full text in Research Archive
• Berg, Erik Andreas Rye; Tasken, Anders Austlid; Nordal, Trym; Grenne, Bjørnar Leangen; Espeland, Torvald & Garstad, Idar Kirkeby [Show all 11 contributors for this article] (2023). Fully automatic estimation of global left ventricular systolic function using deep learning in transesophageal echocardiography. European Heart Journal – Imaging Methods and Practice (EHJ-IMP). 1(1). doi: 10.1093/ehjimp/qyad007. Full text in Research Archive Show summary

  Aims To improve monitoring of cardiac function during major surgery and intensive care, we have developed a method for fully automatic estimation of mitral annular plane systolic excursion (auto-MAPSE) using deep learning in transoesophageal echocardiography (TOE). The aim of this study was a clinical validation of auto-MAPSE in patients with heart disease. Methods and results TOE recordings were collected from 185 consecutive patients without selection on image quality. Deep-learning-based auto-MAPSE was trained and optimized from 105 patient recordings. We assessed auto-MAPSE feasibility, and agreement and inter-rater reliability with manual reference in 80 patients with and without electrocardiogram (ECG) tracings. Mean processing time for auto-MAPSE was 0.3 s per cardiac cycle/view. Overall feasibility was >90% for manual MAPSE and ECG-enabled auto-MAPSE and 82% for ECG-disabled auto-MAPSE. Feasibility in at least two walls was ≥95% for all methods. Compared with manual reference, bias [95% limits of agr

• Kildahl, Henrik Agerup; Brenne, Evelyn Merete Lauvstad; Dalen, Håvard & Wahba, Alexander (2023). Systemic embolization in infective endocarditis. Indian journal of thoracic and cardiovascular surgery. ISSN 0970-9134. doi: 10.1007/s12055-023-01616-2. Full text in Research Archive
• Tasken, Anders Austlid; Berg, Erik Andreas Rye; Grenne, Bjørnar Leangen; Holte, Espen; Dalen, Håvard & Stølen, Stian Bergseng [Show all 9 contributors for this article] (2023). Automated estimation of mitral annular plane systolic excursion by artificial intelligence from 3D ultrasound recordings. Artificial Intelligence in Medicine. ISSN 0933-3657. 144. doi: 10.1016/j.artmed.2023.102646. Full text in Research Archive Show summary

  Perioperative monitoring of cardiac function is beneficial for early detection of cardiovascular complications. The standard of care for cardiac monitoring performed by trained cardiologists and anesthesiologists involves a manual and qualitative evaluation of ultrasound imaging, which is a time-demanding and resource-intensive process with intraobserver- and interobserver variability. In practice, such measures can only be performed a limited number of times during the intervention. To overcome these difficulties, this study presents a robust method for automatic and quantitative monitoring of cardiac function based on 3D transesophageal echocardiography (TEE) B-mode ultrasound recordings of the left ventricle (LV). Such an assessment obtains consistent measurements and can produce a near real-time evaluation of ultrasound imagery. Hence, the presented method is time-saving and results in increased accessibility. The mitral annular plane systolic excursion (MAPSE), characterizing global LV function, is estimated by landmark detection and cardiac view classification of two-dimensional images extracted along the long-axis of the ultrasound volume. MAPSE estimation directly from 3D TEE recordings is beneficial since it removes the need for manual acquisition of cardiac views, hence decreasing the need for interference by physicians. Two convolutional neural networks (CNNs) were trained and tested on acquired ultrasound data of 107 patients, and MAPSE estimates were compared to clinically obtained references in a blinded study including 31 patients. The proposed method for automatic MAPSE estimation had low bias and low variability in comparison to clinical reference measures. The method accomplished a mean difference for MAPSE estimates of (-0.16 +- 1.06) mm. Thus, the results did not show significant systematic errors. The obtained bias and variance of the method were comparable to inter-observer variability of clinically obtained MAPSE measures on 2D TTE echocardiography. The novel pipeline proposed in this study has the potential to enhance cardiac monitoring in perioperative- and intensive care settings.

• Salte, Ivar Mjåland; Østvik, Andreas; Olaisen, Sindre Hellum; Karlsen, Sigve; Dahlslett, Thomas & Smistad, Erik [Show all 13 contributors for this article] (2023). Deep Learning for Improved Precision and Reproducibility of Left Ventricular Strain in Echocardiography: A Test-Retest Study. Journal of the American Society of Echocardiography. ISSN 0894-7317. 36(7), p. 788–799. doi: 10.1016/j.echo.2023.02.017. Full text in Research Archive Show summary

  Aims Assessment of left ventricular (LV) function by echocardiography is hampered by modest test-retest reproducibility. A novel artificial intelligence (AI) method based on deep learning provides fully automated measurements of LV global longitudinal strain (GLS) and may improve the clinical utility of echocardiography by reducing user-related variability. The aim of this study was to assess within-patient test-retest reproducibility of LV GLS measured by the novel AI method in repeated echocardiograms recorded by different echocardiographers and to compare the results to manual measurements. Methods Two test-retest data sets (n = 40 and n = 32) were obtained at separate centers. Repeated recordings were acquired in immediate succession by 2 different echocardiographers at each center. For each data set, 4 readers measured GLS in both recordings using a semiautomatic method to construct test-retest interreader and intrareader scenarios. Agreement, mean absolute difference, and minimal detectable change (MDC) were compared to analyses by AI. In a subset of 10 patients, beat-to-beat variability in 3 cardiac cycles was assessed by 2 readers and AI. Results Test-retest variability was lower with AI compared with interreader scenarios (data set I: MDC = 3.7 vs 5.5, mean absolute difference = 1.4 vs 2.1, respectively; data set II: MDC = 3.9 vs 5.2, mean absolute difference = 1.6 vs 1.9, respectively; all P < .05). There was bias in GLS measurements in 13 of 24 test-retest interreader scenarios (largest bias, 3.2 strain units). In contrast, there was no bias in measurements by AI. Beat-to-beat MDCs were 1,5, 2.1, and 2.3 for AI and the 2 readers, respectively. Processing time for analyses of GLS by the AI method was 7.9 ± 2.8 seconds. Conclusion A fast AI method for automated measurements of LV GLS reduced test-retest variability and removed bias between readers in both test-retest data sets. By improving the precision and reproducibility, AI may increase the clinical utility of echocardiography.

• Espeland, Torvald; Wigen, Morten Smedsrud; Dalen, Håvard; Berg, Erik Andreas Rye; Hammer, Tommy Arild & Salles, Sebastien Andre Roger [Show all 9 contributors for this article] (2023). Mechanical Wave Velocities in Left Ventricular Walls in Healthy Subjects and Patients With Aortic Stenosis. JACC Cardiovascular Imaging. ISSN 1936-878X. 17(2), p. 111–124. doi: 10.1016/j.jcmg.2023.07.009. Full text in Research Archive Show summary

  Background Mechanical wave velocity (MWV) measurement is a promising method for evaluating myocardial stiffness, because these velocities are higher in patients with myocardial disease. Objectives Using high frame rate echocardiography and a novel method for detection of myocardial mechanical waves, this study aimed to estimate the MWVs for different left ventricular walls and events in healthy subjects and patients with aortic stenosis (AS). Feasibility and reproducibility were evaluated. Methods This study included 63 healthy subjects and 13 patients with severe AS. All participants underwent echocardiographic examination including 2-dimensional high frame rate recordings using a clinical scanner. Cardiac magnetic resonance was performed in 42 subjects. The authors estimated the MWVs at atrial kick and aortic valve closure in different left ventricular walls using the clutter filter wave imaging method. Results Mechanical wave imaging in healthy subjects demonstrated the highest feasibility for the atrial kick wave reaching >93% for all 4 examined left ventricular walls. The MWVs were higher for the inferolateral and anterolateral walls (2.2 and 2.6 m/sec) compared with inferoseptal and anteroseptal walls (1.3 and 1.6 m/sec) (P < 0.05) among healthy subjects. The septal MWVs at aortic valve closure were significantly higher for patients with severe AS than for healthy subjects. Conclusions MWV estimation during atrial kick is feasible and demonstrates higher velocities in the lateral walls, compared with septal walls. The authors propose indicators for quality assessment of the mechanical wave slope as an aid for achieving consistent measurements. The discrimination between healthy subjects and patients with AS was best for the aortic valve closure mechanical waves. (Ultrasonic Markers for Myocardial Fibrosis and Prognosis in Aortic Stenosis; NCT03422770 )

• Rindal, Ole Marius Hoel; Bjåstad, Tore Grüner; Espeland, Torvald; Berg, Erik Andreas Rye & Måsøy, Svein-Erik (2023). A Very Large Cardiac Channel Data Database (VLCD) Used to Evaluate Global Image Coherence (GIC) as an In Vivo Image Quality Metric. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control. ISSN 0885-3010. 70(10), p. 1295–1307. doi: 10.1109/TUFFC.2023.3308034. Full text in Research Archive Show summary

  Ultrasound image quality is of utmost importance for a clinician to reach a correct diagnosis. Conventionally, image quality is evaluated using metrics to determine the contrast and resolution. These metrics requires localization of specific regions and targets in the image such as a region of interest (ROI), a background region, and or, a point scatterer. Such objects can all be difficult to identify in in-vivo images, especially for automatic evaluation of image quality in large amounts of data. Using a matrix array probe, we have recorded a Very Large cardiac Channel data Database (VLCD) to evaluate coherence as an in-vivo image quality metric. The VLCD consists of 33 280 individual image frames from 538 recordings of 106 patients. We also introduce a Global Imag Coherence (GIC), an in-vivo image quality metric that does not require any identified ROI since it is defined as an average coherence value calculated from all the data pixels used to form the image, below a pre-selected range. The GIC is shown to be a quantitative metric for in-vivo image quality when applied to the VLCD. We demonstrate, on a subset of the dataset, that the GIC correlates well with the conventional metrics contrast ratio (CR) and the generalized contrast-to-noise ratio (gCNR)with R=0.74 (p0.9, p

• Shur, M.S.; Liu, X. & Ytterdal, Trond (2023). Editors’ Choice—Thin Film Transistor Response in the THz Range. ECS Journal of Solid State Science and Technology. ISSN 2162-8769. 12(3). doi: 10.1149/2162-8777/acc20d. Full text in Research Archive
• Shur, Michael S.; Liu, Xueqing & Ytterdal, Trond (2023). Phase-sensitive Silicon CMOS TeraFETs. In Cooke, David (Eds.), 2023 48th International Conference on Infrared, Millimeter, and Terahertz Waves. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350336603. doi: 10.1109/IRMMW-THz57677.2023.10298984. Full text in Research Archive
• Sæbø, Sigbjørn; Pettersen, Håkon Neergaard; Smistad, Erik; Pasdeloup, David Francis Pierre; Stølen, Stian Bergseng & Grenne, Bjørnar Leangen [Show all 9 contributors for this article] (2023). Real-time guiding by deep learning during echocardiography to reduce left ventricular foreshortening and measurement variability. European Heart Journal – Imaging Methods and Practice (EHJ-IMP). 1. doi: 10.1093/ehjimp/qyad012. Full text in Research Archive
• Letnes, Jon Magne; Nes, Bjarne Martens; Langlo, Knut Asbjørn Rise; Aksetøy, Inger-Lise Aamot; Lundgren, Kari Margrethe & Skovereng, Knut [Show all 9 contributors for this article] (2023). Indexing cardiac volumes for peak oxygen uptake to improve differentiation of physiological and pathological remodeling: from elite athletes to heart failure patients. European Heart Journal-Cardiovascular Imaging. ISSN 2047-2404. 24(6), p. 721–729. doi: 10.1093/ehjci/jead034. Full text in Research Archive Show summary

  Aims Cardiovascular structures adapt to meet metabolic demands, but current methodology for indexing by body size does not accurately reflect such variations. Therefore, we aimed to investigate how left ventricular end-diastolic volume (LVEDV) and left atrial maximal volume (LAVmax) are associated with absolute (L/min) peak oxygen uptake (VO2peak) and fat-free mass (FFM) compared to body surface area (BSA). We subsequently assessed the impact of indexing by absolute VO2peak, FFM, and BSA to discriminate pathological from physiological remodeling. Methods and results We used data from 1190 healthy adults to explore relationships for BSA, FFM, and absolute VO2peak with LVEDV and LAVmax by regression and correlation analyses. We then compared these indexing methods for classification to normalcy/pathology in 61 heart failure patients and 71 endurance athletes using the chi-squared and Fisher exact tests and the net reclassification and integrated discrimination indices. Absolute VO2peak correlated strongly with LVEDV, explaining 52% of variance vs. 32% for BSA and 44% for FFM. Indexing LVEDV for VO2peak improved discrimination between heart failure patients and athletes on top of indexing to BSA. Seventeen out of 18 athletes classified to pathology by BSA were reclassified to normalcy by VO2peak indexing (P < 0.001), while heart failure patients were reclassified to pathology (39–95%, P < 0.001). All indexing methods explained below 20% of the variance in LAVmax in univariate models. Conclusions Indexing LVEDV to VO2peak improves the ability to differentiate physiological and pathological enlargement. The LVEDV to absolute VO2peak ratio may be a key index in diagnosing heart failure and evaluating the athlete’s heart.

• Feyling, Fredrik Esp; Malmberg, Hampus; Wulff, Carsten; Loeliger, Hans-Andrea & Ytterdal, Trond (2023). Design and Analysis of the Leapfrog Control-Bounded A/D Converter. IEEE Transactions on Very Large Scale Integration (vlsi) Systems. ISSN 1063-8210. 32(1), p. 79–88. doi: 10.1109/TVLSI.2023.3320279. Full text in Research Archive
• Balasubramanian, Linknath Surya; Rizkalla, Maher; Lee, John; Ytterdal, Trond & Kumar, Mukesh (2023). Towards No Penalty Control Hazard Handling. In Abshire, Pamela (Eds.), Proceedings of the 66th IEEE International Midwest Symposium on Circuits and Systems. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350302103. p. 1128–1131. doi: 10.1109/MWSCAS57524.2023.10405871. Full text in Research Archive
• Gopinath, Anoop; Ytterdal, Trond; Yadhav, Avinash; Lee, John; Rizkalla, Maher & Kumar, Mukesh (2023). SRAM Vmin Scaling Via Negative Wordline. In Abshire, Pamela (Eds.), Proceedings of the 66th IEEE International Midwest Symposium on Circuits and Systems. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350302103. p. 227–231. doi: 10.1109/MWSCAS57524.2023.10406117. Full text in Research Archive
• Bolstad, Per Kristian; Frijlink, Martijn & Hoff, Lars (2023). Evaluation of bonding techniques for ultrasound transducers. Microelectronics and reliability. ISSN 0026-2714. 151. doi: 10.1016/j.microrel.2023.115272. Full text in Research Archive
• Le, Hoang Duy; Manh, Tung & Hoff, Lars (2023). A novel wide-band design of dual-frequency transducer based on PZT-PVDF for super harmonic imaging. Proceedings - IEEE Ultrasonics Symposium. ISSN 1948-5719. doi: 10.1109/IUS51837.2023.10308044. Full text in Research Archive

View all works in NVA

• Viggen, Erlend Magnus (2024). Proceedings of the 47th Scandinavian Symposium on Physical Acoustics. Norsk Fysisk Selskap. ISBN 9788281230248. 18 p. Full text in Research Archive
• Viggen, Erlend Magnus (2023). Proceedings of the 46th Scandinavian Symposium on Physical Acoustics. Norsk Fysisk Selskap. ISBN 9788281230231. 87 p. doi: https:/www.researchgate.net/publication/370074370_Proceedings_of_the_46th_Scandinavian_Symposium_on_Physical_Acoustics. Full text in Research Archive

View all works in NVA

• Fadnes, Solveig (2025). Vector flow imaging: clinical utility of intracardiac flow parameters. Full text in Research Archive
• Dalen, Håvard (2025). Systolisk og diastolisk funksjon hos atleter. Full text in Research Archive
• Dalen, Håvard (2025). This year in structural heart disease. Full text in Research Archive
• Dalen, Håvard (2025). Kunstig intelligens i ekkokardiografi - nye metoder og veien videre. Full text in Research Archive
• Dalen, Håvard (2025). Early intervention in aortic stenosis - Valve, left ventricle or symptoms? Full text in Research Archive
• Dalen, Håvard (2025). Normalverdier for ekkokardiografi. Full text in Research Archive
• Dalen, Håvard (2025). Echocardiography for improved cardiac diagnostics. Full text in Research Archive
• Dalen, Håvard (2025). GLS - How to think in the multimodal/multivendor world. Full text in Research Archive
• Dalen, Håvard (2024). Sharing sensitive data. Full text in Research Archive
• Dalen, Håvard (2024). CT i kardiologi. Full text in Research Archive
• Dalen, Håvard (2024). Forbedret hjertediagnostikk ved kunstig intelligens. Full text in Research Archive
• volnes-eriksen, torfinn & Dalen, Håvard (2024). Body Surface Area Calculation and Its Impact on Indexed Echocardiographic Measurements. JACC Cardiovascular Imaging. ISSN 1936-878X. 17, p. 842–842. doi: 10.1016/j.jcmg.2024.04.011. Full text in Research Archive
• Måsøy, Svein-Erik (2024). The Centre for Innovative Ultrasound Solutions – examples of innovation from echocardiography. Full text in Research Archive
• Nancy, Bazilchuk; Pasdeloup, David Francis Pierre; Dalen, Håvard; Løvstakken, Lasse & Måsøy, Svein-Erik (2024). Peering into your heart – with the help of AI. Gemini (English edition). ISSN 0804-8568. Full text in Research Archive
• Pasdeloup, David Francis Pierre; Dalen, Håvard; Løvstakken, Lasse; Østvik, Andreas & Olaisen, Sindre Hellum (2024). Challenges and Opportunities in end-to-end Dep Learning for Echocardiography. Full text in Research Archive
• Holmstrøm, Vegard; Smistad, Erik; Nyberg, John; Østvik, Andreas; Holte, Espen & Løvstakken, Lasse [Show all 8 contributors for this article] (2024). Deep Learning for Fully Automated Echocardiographic Measurements of Left Ventricular Wall Thickness and Chamber Dimensions in the Parasternal Long-Axis View. Full text in Research Archive
• Holmstrøm, Vegard; Smistad, Erik; Østvik, Andreas; Holte, Espen; Løvstakken, Lasse & Dalen, Håvard [Show all 7 contributors for this article] (2024). Real-Time Analysis of Left Ventricular Global Longitudinal Strain Using Deep Learning. Full text in Research Archive
• Holmstrøm, Vegard; Smistad, Erik; Stølen, Stian Bergseng; Løvstakken, Lasse; Dalen, Håvard & Holte, Espen [Show all 8 contributors for this article] (2024). Improving Time-Efficiency and Accuracy in Echocardiography: Real-Time Automated Measurements of Left Ventricular Global Longitudinal Strain using Deep Learning. Full text in Research Archive
• Nancy, Bazilchuk; Pasdeloup, David Francis Pierre; Dalen, Håvard; Løvstakken, Lasse & Måsøy, Svein-Erik (2024). Kunstig intelligens hjelper til med å se hjertet ditt. Gemini. ISSN 0802-085X. Full text in Research Archive
• Dalen, Håvard (2024). Klaffesykdom i Etiopia. Full text in Research Archive
• Dalen, Håvard (2024). Improving cardiac diagnostic imaging by artificial intelligence. Full text in Research Archive
• Viggen, Erlend Magnus (2024). Planning a new introductory textbook on waves in solids. Full text in Research Archive
• Viggen, Erlend Magnus & Arnestad, Håvard (2024). On acoustic radiation from subsonic surface vibrations. Full text in Research Archive
• Wifstad, Sigurd Vangen; Kildahl, Henrik Agerup; Holte, Espen; Berg, Erik Andreas Rye; Grenne, Bjørnar Leangen & Dalen, Håvard [Show all 7 contributors for this article] (2024). Automated valve quantification using deep learning. Full text in Research Archive
• Wangensteen, Magnus; Johansen, Tonni Franke; Fatemi, Ali & Viggen, Erlend Magnus (2024). Researchers Use Machine Learning to Detect Pitting Corrosion. [Journal]. Materials Performance. Full text in Research Archive
• Dalen, Håvard (2024). Sharing of sensitive data. Full text in Research Archive
• Dalen, Håvard (2024). Hjertefysiologi og rehabilitering. Full text in Research Archive
• Dalen, Håvard (2024). Forskning som endrer klinisk praksis. Full text in Research Archive
• Rye, Cathrine Sæthern & Dalen, Håvard (2024). Kvalitet på kardiovaskulære diagnoser i sykehus. Full text in Research Archive
• Dalen, Håvard (2024). Kunstig intelligens ved ekkokardiografi. Full text in Research Archive
• Bryon, Vincent; Wifstad, Sigurd Vangen; Hu, Jieyu; Grønli, Thomas & Løvstakken, Lasse (2023). Automated Patient-Specific Left Ventricular Simulations for Cardiac Function Evaluation Using Image-Based Computational Fluid Dynamics and Color Flow Imaging. Full text in Research Archive
• Wifstad, Sigurd Vangen; Indimath, Shivanandan; Bryon, Vincent; Avdal, Jørgen; Løvstakken, Lasse & Måsøy, Svein-Erik [Show all 7 contributors for this article] (2023). Can we use Deep Learning to extract important information from poor resolution Doppler images? Full text in Research Archive
• Bryon, Vincent; Wifstad, Sigurd Vangen; Hu, Jieyu; Grønli, Thomas & Løvstakken, Lasse (2023). Simulating Blood Flow in the Heart: Enhancing Color Flow Imaging with Computational Fluid Dynamics. Full text in Research Archive
• Wifstad, Sigurd Vangen; Kildahl, Henrik Agerup; Holte, Espen; Grenne, Bjørnar Leangen; Dalen, Håvard & Løvstakken, Lasse (2023). Automatic Jet Flow Convergence Detection and Quantification from 2D Color Doppler Flow Using Deep Learning. Full text in Research Archive
• Letnes, Jon Magne; Wisløff, Ulrik & Dalen, Håvard (2023). Heart Failure With Preserved Ejection Fraction: Exercise Deficiency or Ventricular Maladaptation to Metabolic Demands? JACC Cardiovascular Imaging. ISSN 1936-878X. doi: 10.1016/j.jcmg.2023.05.026. Full text in Research Archive
• Salte, Ivar Mjåland; Østvik, Andreas; Olaisen, Sindre Hellum; Karlsen, Sigve; Dahlslett, Thomas & Smistad, Erik [Show all 13 contributors for this article] (2023). Response to “Minimal Detectable Change and Reproducibility of Echocardiographic Strain: Implications for Clinical Practice”. Journal of the American Society of Echocardiography. ISSN 0894-7317. doi: 10.1016/j.echo.2023.08.017. Full text in Research Archive
• Pettersen, Håkon Neergaard; Sæbø, Sigbjørn; Smistad, Erik; Pasdeloup, David Francis Pierre; Stølen, Stian Bergseng & Grenne, Bjørnar Leangen [Show all 9 contributors for this article] (2023). Real-time guiding by deep learning during echocardiography to reduce left ventricular foreshortening and measurement variability. Full text in Research Archive
• Fadnes, Solveig (2023). Vector Flow Imaging and extracted flow features. Full text in Research Archive
• Fadnes, Solveig (2023). High Frame Rate Imaging in Pediatric and Fetal Medicine - Opportunities and Challenges. Full text in Research Archive Show summary

  This was an invited talk to the IEEE International Ultrasonics Symposium 2023 in Montréal, Canada. The talk summarized the research possibilities in pediatric and fetal medicine given by high frame rate imaging with focus on the visualization and quantification of intracardiac blood flow. Current clinical state-of-the-art imaging technology was demonstrated, and both the opportunities of new pathophysiological knowledge and diagnostic tools, and the inherent challenges of high-frame-rate imaging was discussed.

• Rindal, Ole Marius Hoel; Fiorentini, Stefano; Avdal, Jørgen; Austeng, Andreas & Molares, Alfonso Rodriguez (2023). The Generalized Beamformer in the UltraSound ToolBox. Full text in Research Archive Show summary

  Access to channel data, and real-time processing of such, have left us a landscape filled with beamforming techniques often using specific wavefronts with specific reconstruction algorithms. This implies high costs for code maintenance and makes it difficult to compare different techniques. We demonstrate a generalized beamformer (GB) to reconstruct any transmit (tx) waveform from any probe into a defined grid of pixels. It covers linear and sector scans, both steered and stepped tx and even complex beamforming such as multiple line acquisition (MLA) and retrospective tx beamforming (RTB). Also, we show how the GB can be utilized to improve the channel data used for adaptive beamforming.

• Rindal, Ole Marius Hoel; Vrålstad, Anders Emil; Bjåstad, Tore Grüner; Sørnes, Anders Rasmus & Måsøy, Svein-Erik (2023). Improving Image Quality Degradation due to Blocked Array in Cardiac Ultrasound Imaging. Full text in Research Archive Show summary

  The footprint of a cardiac ultrasound (US) probe is designed to fit the size of the intercostal space between the ribs to avoid ribs blocking the US energy. Some patients will have an intercostal space narrower than the footprint, making array blockage inevitable. In some cases, the clinician may be forced to place the probe in a position subject to rib blockage to achieve the desired anatomical view. Firstly, we demonstrate that such rib blockages reduce the image quality and lead to depth-varying geometrical distortions, potentially tilting structures and reducing resolution. Secondly, we demonstrate that compensating for the blocked array in the retrospective transmit beamforming (RTB) corrects most of the unwanted effects and improves image quality.

• Shur, Michael S.; Liu, Xueqing & Ytterdal, Trond (2023). Phase-sensitive Silicon CMOS TeraFETs. Full text in Research Archive
• Olaisen, Sindre Hellum (2023). AI-based measurements in echocardiography, clinical results. Full text in Research Archive
• Viggen, Erlend Magnus & Arnestad, Håvard (2023). Explaining the cutoff, reappearance, and persistence of subsonic leaky A0 Lamb waves. Full text in Research Archive
• Shur, Michael S.; Liu, Xueqing & Ytterdal, Trond (2023). Switching Characteristics of GaN Power Transistors. Full text in Research Archive
• Nyberg, John; Jakobsen, Even Olav; Østvik, Andreas; Holte, Espen; Stølen, Stian & Løvstakken, Lasse [Show all 8 contributors for this article] (2023). Determinants of left atrial reservoir strain in a large healthy population. Full text in Research Archive
• Nyberg, John; Jakobsen, Even Olav; Østvik, Andreas; Holte, Espen; Stølen, Stian & Løvstakken, Lasse [Show all 8 contributors for this article] (2023). Left atrial strain in a large healthy population. Full text in Research Archive
• Kasongo, Muana; Ytterdal, Trond; Lee, John; Rizkalla, Maher & Kumar, Mukesh (2023). Evaluation of Leakage Currents in Memristor Crossbar Arrays. Full text in Research Archive
• Balasubramanian, Linknath Surya; Rizkalla, Maher; Lee, John; Ytterdal, Trond & Kumar, Mukesh (2023). Towards No Penalty Control Hazard Handling. Full text in Research Archive
• Gopinath, Anoop; Ytterdal, Trond; Yadhav, Avinash; Lee, John; Rizkalla, Maher & Kumar, Mukesh (2023). SRAM Vmin Scaling Via Negative Wordline. Full text in Research Archive
• Viggen, Erlend Magnus & Arnestad, Håvard (2023). Why leaky flexural plate waves misbehave at low frequencies. Full text in Research Archive
• Le, Hoang Duy; Manh, Tung & Hoff, Lars (2023). A novel wide-band design of dual-frequency transducer based on PZT-PVDF for super harmonic imaging. Full text in Research Archive
• Hau, Maïlys Françoise; Kiss, Gabriel Hanssen; Lindseth, Frank; Grenne, Bjørnar Leangen; Veronesi, Federico & Gérard, Olivier (2026). Spatio-temporal Deep Learning for Cardiac Structure and Motion Analysis in Transesophageal Echocardiography. Norges teknisk-naturvitenskapelige universitet. ISSN 9788235300683. Full text in Research Archive Show summary

  This thesis explores the application of convolutional neural network-based (CNN) architectures on transesophageal echocardiography (TEE) acquisitions. The objective is to achieve 3D segmentation and 2D landmark detection to provide clinicians with automated tools to support patient diagnosis and monitoring. Echocardiography is the most widely used method for cardiac screening, as it yields real-time visualisation of the heart. However, it is difficult to analyse quantitatively and quickly. This causes clinicians to often rely on qualitative measurements done on the fly. Albeit fast, these methods lack reproducibility. Moreover, when performed, manual annotations are often dependent on the annotators (inter-operator variability) and the time or day that the same labelling is done (intra-operator variability). This dissertation employs automatic image analysis methods on echocardiograms to provide clinicians with fast, consistent, and quantitative cardiac measurements to address issues of observer variability and the time-demanding annotation process. Mainly, the work conducted in this thesis takes advantage of deep learning techniques based on U-Net and ResNet to segment relevant parts of the mitral valve (MV) and detect keypoints in the tricuspid valve (TV) and right ventricle (RV). The study is specifically run on TEE, where the ultrasound probe is placed in the patient’s oesophagus, offering a clearer view of the valves, as the sonographer does not have to deal with the ribcage obstructing part of their field of view on the heart. The first part of this dissertation focuses on segmenting the MV from 3D TEE recordings. Using various training strategies, the MV’s shape is modelled directly from the echocardiograms. One of the approaches automatically creates a refined bounding box by leveraging existing clinical software. The second transfers knowledge from one ultrasound domain to another using semi-supervised learning strategies. Patient-specific valve-modelling can provide high accuracy in diagnosis and surgery planning (e.g. valve repair), in addition to erasing intra- and inter-operator variabilities, leading to more consistent follow-ups of the patients, and opening the way for more standardised patient examination. In the second part, this thesis investigates the feasibility of automatically computing clinical indices for patient monitoring. Three landmarks, two on the TV and one at the RV’s apex, are detected across the heart cycle on 2D TEE recordings by training a U-Net-based neural network. From these three points, it is possible to precisely compute a range of indices relating to the well-being of a patient in milliseconds (e.g. TAPSE and linear RVS). Automatic cardiovascular indices computation provides stability and reproducibility in patient measurements, along with precise monitoring when in the intensive care unit (ICU). These results could initiate real-time and automatic detection of heart failures in critical care patients.

• Bolstad, Per Kristian (2024). Evaluation of Metallurgical Bonding for Ultrasound Transducers. Universitetet i Sørøst-Norge/Universitetet i Søraust-Noreg. ISSN 9788272068522. Full text in Research Archive Show summary

  Common ultrasound transducers use a piezoelectric element to convert between electrical and mechanical energy. The piezoelectric layer is sandwiched between acoustic matching and backing layers to maximize energy transfer and bandwidth. The bonding layers joining these materials are conventionally made of polymers, and decades of experience verify that it functions excellently in many ultrasound transducer applications. These bonding layers should be very thin compared to the acoustic wavelength to avoid reverberations and reduced bandwidth. Transducers for operation in harsh environments rely on careful selection of materials, and polymer materials are known to degrade at high temperatures. Metallurgical bonding is proposed as an alternative to polymeric adhesives. Metals form strong, electrically- and thermally conductive bonds with a high characteristic acoustic impedance. Solid-liquid interdiffusion (SLID) bonding is a technique which relies on the formation of intermetallic compounds, which are stable at temperatures above the processing temperature. SLID bonds are not reflowable, as compared to bonds formed through soldering. This thesis explores the binary metal system of gold (Au) and Tin (Sn) for bonding essential layers of ultrasound transducers, such as piezoelectric ceramic materials. Metallurgical bonding techniques, Au-Sn SLID and Au-Sn soldering, are evaluated and compared to conventional polymeric adhesives. Testing of bonding performance during and after exposure to high temperature and high pressure indicated high-temperature stability with high mechanical strength. A common challenge related to metallurgical bonding is the formation of voids within the layer, which are gas or vacuum-filled pockets. Reliable estimation of the influence from such voids on the acoustic performance is important if metallurgical bonding is to be used in acoustic transducers. A finite element method modelling technique is presented in this thesis, which can be applied to accurately estimate effective medium parameters of layers containing voids of arbitrary size, concentration, and distribution.

• Mawad, Wadi & Nyrnes, Siri Ann (2024). Right ventricular energetics using blood speckle tracking in pediatric congenital heart disease. Norges teknisk-naturvitenskapelige universitet. ISSN 9788232679713. Full text in Research Archive Show summary

  This thesis is composed of three studies of right ventricular (RV) energetics in pediatric patients with congenital heart disease (CHD), using blood speckle tracking (BST). The objectives were to describe the RV energetics and flow dynamics in the context of various CHD resulting in adverse loading of the RV. We compared these patient groups with healthy controls. Paper 1: In this pilot study, we used BST to image RV flow patterns in children with large atrial septal defects (ASD) and those with repaired tetralogy of Fallot (rTOF) with severe residual pulmonary insufficiency (PI). This study included 57 participants: 21 with rTOF, 11 with large ASD, and 25 healthy individuals (CTL). Additionally, a flow phantom was utilized to evaluate the impact of imaging plane and smoothing adjustments on the parameter of energy loss (EL) obtained with BST. We observed that RV diastolic EL in ASD and rTOF was comparable, but both were higher than in CTL. High EL areas during systole were consistent across all groups, appearing in the RV outflow tract and near the tricuspid valve leaflets in early diastole. Notably, an extra area of early diastolic EL at the apex was identified in rTOF patients, related to the interaction of tricuspid inflow and PI. The flow phantom experiment indicated that while EL varied with imaging plane and smoothing settings, the overall EL trend remained stable if these settings were consistent. Paper 2: This study used BST to study of blood flow dynamics and patterns in the right ventricle and main pulmonary artery in pediatric pulmonary hypertension (PAH) compared to healthy controls. Eighteen subjects were included in each group. A diastolic vortex in the main pulmonary artery (MPA) was identified in 16 of the PAH patients, but not in controls. Significantly higher MPA systolic and diastolic EL, as well as increased vector complexity (VC) and diastolic vorticity were noted in PAH compared to controls. This study demonstrated abnormal flow patterns in the MPA with diastolic vortex formation in most patients with PAH. This diastolic vortex likely results from reflected waves from the distal pulmonary bed. Our data indicate that diastolic vortex could potentially be used in the diagnosis of PAH. Paper 3: This study included patients with univentricular hearts (UVH) after Fontan palliation and compared their intracardiac flow patterns to healthy controls. The UVH included 29 patients and 30 CTL. Qualitative vortex formation was similar between groups in systole and diastole except for double inlet left ventricle which had more complex inflow patterns consisting of colliding vortices. Quantitatively, UVH kinetic energy (KE), energy loss (EL) and vorticity (VO) were lower when compared to healthy controls. We also demonstrated a correlation between KE, EL and VO. We concluded that although similar vortex formation is preserved in well-functioning UVH after Fontan, the latter is an energy-deprived circulation where all mechanisms of energy conservation are important.

View all works in NVA