PeopleofSuperintelligence

Researcher

Torfinn Taxt

Torfinn Taxt is a Norwegian researcher whose work develops classifiers and blind-deconvolution methods for medical ultrasound and MRI. He earned an M.D. in 1976, an M.S. in computer science in 1978 and a Ph.D. in developmental neuroscience in 1983 at the University of Oslo, then held professorships at Bergen and Oslo by 1996. The International Association for Pattern Recognition named him a Fellow in 2000 for multispectral image analysis; with Radovan Jiřík, he published blind-deconvolution methods for ultrasound images between 2004 and 2008. In 2009, he coauthored multispectral software for segmenting brain MR images, and in 2010 he coauthored a study of voxel-specific arterial input functions for brain perfusion imaging. Between 2014 and 2019, he worked on blind arterial-input-function estimation in DCE-MRI, and he coauthored a further DCE-MRI paper with a Bergen affiliation in 2024.

Timeline

  1. 1976–1996

    Medical training and professorships

    At the University of Oslo, he earns an M.D. in 1976, an M.S. in computer science in 1978 and a Ph.D. in developmental neuroscience in 1983. By 1996, he is professor in biomedical imaging at the University of Bergen, Norway, and professor in image processing at the University of Oslo.

  2. 2000

    An IAPR fellowship

    The International Association for Pattern Recognition names him a Fellow from Norway in 2000 for contributions to multispectral image analysis.

  3. 2004–2008

    Blind deconvolution for ultrasound

    In the Proceedings of the 2004 IEEE Ultrasonics Symposium, he and Radovan Jiřík present two-dimensional blind iterative deconvolution for medical ultrasound. Their 2005 EMBEC paper uses maximum-likelihood blind deconvolution; tests on a tissue-mimicking phantom and clinical data show increased spatial resolution. In 2008, they publish a two-dimensional blind Bayesian deconvolution method in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control; tests on phantom and clinical images show higher spatial resolution than the inputs.

  4. 2009

    Multispectral brain-image software

    In 2009, he coauthors “Multispectral analysis of multimodal images” with Yngve Kvinnsland, Njål Brekke and Renate Grüner in Acta Oncologica. Their software combines unsupervised EM and k-means classifiers with supervised Bayesian and kNN classifiers to segment T1- and T2-weighted MR images of brain tissue and tumors.

  5. 2010

    Voxel-specific arterial input functions

    At ISMRM 2010, he coauthors “Voxel-Specific Brain Arterial Input Functions from DSC-MRI and Blind Deconvolution in a Group of Healthy Males.” The study estimates voxel-specific arterial input functions for 44 healthy men with blind deconvolution and finds differences in delay and dispersion within brain regions and across participants.

  6. 2014–2016

    Parametric AIF models in DCE-MRI

    In the 2014 Proceedings of the International Society for Magnetic Resonance in Medicine, he coauthors an abstract with Radovan Jiřík, Karel Souček, Zenon Starčuk and others on blind multichannel deconvolution for estimating a parametric arterial input function in mouse DCE-MRI. A six-author Magma abstract in 2016 places him last after Jiří Kratochvíla, Jiřík, Starčuk, Michal Bartoš and Michal Standara; it uses a parametric AIF with blind deconvolution and the DCATH model to increase the DCE-MRI sampling interval while preserving perfusion-analysis accuracy.

  7. 2019

    Small-animal DCE-MRI input functions

    In 2019, he is second author after Jiřík on “Blind deconvolution estimation of an arterial input function for small animal DCE-MRI” in Magnetic Resonance Imaging. The paper proposes an AIF model for small-animal physiology and estimates its parameters with multichannel blind deconvolution.

  8. 2024

    Blind deconvolution in DCE-MRI

    In 2024, he coauthors “Blind deconvolution decreases requirements on temporal resolution of DCE-MRI: Application to 2nd generation pharmacokinetic modeling” in Magnetic Resonance Imaging while affiliated with the University of Bergen.

Career

  1. University of BergenProfessor in biomedical imaging
  2. University of OsloProfessor in image processing

Education

  1. 1983University of OsloPh.D. in developmental neuroscience
  2. 1978University of OsloM.S. in computer science
  3. 1976University of OsloM.D.

From the biography.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes University of Oslo. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes University of Oslo. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. The United States employs an estimated 48% of people shaping AIWe estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.

Sources and credits

Photo: We have no published photo of Torfinn Taxt yet. Send us a free-licensed photo.

  1. CVPR '96 Tutorials pages.cs.wisc.edu. Checked 21 October 2026.
  2. Chronological List of IAPR Fellows – International Association for Pattern Recognition iapr.org. Checked 21 October 2026.
  3. Two-dimensional blind iterative deconvolution of medical ultrasound images – R&D Result Detail, Brno University of Technology vut.cz. Checked 21 October 2026.
  4. Blind iterative deconvolution of ultrasonic images – R&D Result Detail, Brno University of Technology vut.cz. Checked 21 October 2026.
  5. Two-Dimmensional Blind Bayesian Deconvolution of Medical Ultrasound Images – R&D Result Detail, Brno University of Technology vut.cz. Checked 21 October 2026.
  6. Multispectral analysis of multimodal images (Journal Article) | OSTI.GOV osti.gov. Checked 21 October 2026.
  7. (ISMRM 2010) Voxel-Specific Brain Arterial Input Functions from DSC-MRI and Blind Deconvolution in a Group of Healthy Males archive.ismrm.org. Checked 21 October 2026.
  8. Blind Mulitchannel Deconvolution for Estimation of a Parametric AIF in DCE-MRI of Mice – R&D Result Detail, Brno University of Technology vut.cz. Checked 21 October 2026.
  9. Using multichannel blind AIF estimation with the DCATH model to allow increased temporal sampling interval in DCE-MRI – R&D Result Detail, Brno University of Technology vut.cz. Checked 21 October 2026.
  10. Blind deconvolution estimation of an arterial input function for small animal DCE-MRI | Institute of Computer Science MU | MUNI ICS ics.muni.cz. Checked 21 October 2026.
  11. Blind deconvolution decreases requirements on temporal resolution of DCE-MRI: Application to 2nd generation pharmacokinetic modeling | OpenAlex openalex.org. Checked 21 October 2026.