Celebrating the defence
Photo: Changkyu Choi

Celebrating the defence

Excellent PhD defense in the Visual Intelligence research centre

Stine Hansen defended her PhD thesis “Leveraging Supervoxels for Medical Image Volume Segmentation With Limited Supervision” on Dec. 16th at UiT The Arctic University of Norway.

Excellent PhD defense in the Visual Intelligence research centre

The focus of Stine Hansens thesis is on the development of new ML algorithms for medical image segmentation with limited supervision. In particular, completely unsupervised solutions to lung tumor segmentation and solutions that only require a few labeled samples to perform organ segmentation (few-shot learning) are proposed. To exploit the data-specific opportunities of medical images, automatically generated supervoxels are exploited in various ways to solve key challenges related to the task.

Dr. Stine Hansen (Center) with family, committee and supervisor. Photo: Changkyu Choi

Evaluation Committee

  • Professor Alexandros Iosifidis, Department of Electrical and Computer Engineering, Aarhus University, Danmark (1. Opponent)
  • Professor Kjersti Engan, Institutt for data og teknologi, Universitetet i Stavanger (2. Opponent)
  • Associate professor Shujian Yu, IFT, UiT (internal member)


  • Professor Robert Jenssen, IFT, UiT (main supervisor)
  • Associate professor Stian Normann Anfinsen, IFT, UiT (co-supervisor)

Summary of Stine Hansen's thesis work:

The majority of existing methods for machine learning-based medical image segmentation are supervised models that require large amounts of fully annotated images. These types of datasets are typically not available in the medical domain and are difficult and expensive to generate. A wide-spread use of machine learning based models for medical image segmentation therefore requires the development of data-efficient algorithms that only require limited supervision. To address these challenges, this thesis presents new machine learning methodology for unsupervised lung tumor segmentation and few-shot learning based organ segmentation. When working in the limited supervision paradigm, exploiting the available information in the data is key. The methodology developed in this thesis leverages automatically generated supervoxels in various ways to exploit the structural information in the images. The work on unsupervised tumor segmentation explores the opportunity of performing clustering on a population-level in order to provide the algorithm with as much information as possible. To facilitate this population-level across-patient clustering, supervoxel representations are exploited to reduce the number of samples, and thereby the computational cost. In the work on few-shot learning-based organ segmentation, supervoxels are used to generate pseudo-labels for self-supervised training. Further, to obtain a model that is robust to the typically large and inhomogeneous background class, a novel anomaly detection-inspired classifier is proposed to ease the modelling of the background. To encourage the resulting segmentation maps to respect edges defined in the input space, a supervoxel-informed feature refinement module is proposed to refine the embedded feature vectors during inference. Finally, to improve trustworthiness, an architecture-agnostic mechanism to estimate model uncertainty in few-shot segmentation is developed. Results demonstrate that supervoxels are versatile tools for leveraging structural information in medical data when training segmentation models with limited supervision.

Full Thesis Here

Latest news

New principal investigators at Visual Intelligence

February 15, 2024

Visual Intelligence congratulates associate professors Elisabeth Wetzer, Ali Ramezani-Kebrya, and Kristoffer Wickstrøm with their recently promoted roles as principal investigators at the research centre.

Two Visual Intelligence papers accepted for prestigious AI conference

February 9, 2024

New information theories and divergences by Visual Intelligence have been developed and accepted in the prestigious International Conference on Learning Representations (ICLR) 2024. ICLR has an acceptance rate of approximately 30 percent.

Taught graph machine learning at the Geilo Winter School

February 8, 2024

Associate professor Benjamin Ricaud was one of three invited lecturers at the 24th Geilo Winter School. His presentations focused on graph spectral theory, graph signal processing and their connection to graph machine learning.

TV2: Facebook to tag AI-generated pictures: - Major news

February 8, 2024

In a Norwegian news story by TV2, associate professor Kristoffer Wickstrøm shares his views on Meta's recent statements saying they will tag all AI-generated pictures published on their platforms.

TV2: Experts warn: - Makes you a felon

February 5, 2024

In a news story by TV2, associate professor Elisabeth Wetzer at UiT Machine Learning Group addresses the accessibility of deepfake tools and the necessity of proper AI legislation to combat the production and spread of such materials.

Visual Intelligence represented at Arctic Frontiers 2024

January 30, 2024

Visual Intelligence (VI) director Robert Jenssen represented the research centre during the Arctic Frontiers panel discussion, "Tilpasning til fremtidens næringsliv: kunnskap og kunstig intelligens".

Meet Petter, our new science communicator

January 26, 2024

His core work tasks involve communicating research activity from Visual Intelligence and UiT Machine Learning Group, as well as being a facilitative resource for science communication. - I am very excited to get acquainted with our researchers and their projects, says Bjørklund.

Thank you for participating in NLDL 2024!

January 16, 2024

Thank you to our participants who traveled to Tromsø to attend this year’s Northern Lights Deep Learning Conference (NLDL) from the 9th to 11th of January!

Pratet om kunstig intelligens og NLDL-konferansen på NRK Radio

January 5, 2024

I morgen tidlig kunne du høre senterleder for Visual Intelligence, Robert Jenssen prate om kunstig intelligens og NLDL-konferansen på NRKs radiosending. Hør hele NRK-intervjuet med Jenssen her (tidsstempel 01:30:00):

Kunstig intelligens: Advarer mot diskriminering av minoriteter

December 19, 2023

Helsevesenet er om bord når KI-toget nå forlater stasjonen. Men kan vi stole på at kunstig intelligens driver likebehandling av pasienter? – Det finnes en risiko for diskriminering, og minoritetsgrupper er særlig utsatt, advarer Mathias Karlsen Hauglid.

Change of Chair of the Visual intelligence Board

December 11, 2023

Gudmundur Jøkulsson, Kongsberg Satellite Services succeeds Anita Schumacher, CEO University Hospital of North Norway as chair of the Visual intelligence Board from 2024.

Visit from The Computer Vision Lab, University of Victoria, Canada

November 25, 2023

The Visual Intelligence group at NR recently had a visit by from Mélissa Côté, Research Associate at the Computer Vision Lab to discuss deep learning and marine acoustic data.

Ja takk til «krysskulturelle» prosjekter drevet fram av teknologiutvikling

November 21, 2023

Forskningsrådet står ovenfor en krevende oppgave med å blidgjøre et stort og aktivt KI-miljø i Norge. Vi håper at eksemplene på vellykket samarbeid mellom humanister og teknologer kan være til inspirasjon i prosessen