Celebrating the defence
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.
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.
.jpg)
Evaluation Committee
Supervisors
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.
Centre director Robert Jenssen recently gave a talk about the use and benefits of AI technologies during Pasientsikkerhetskonferansen 2024
Congratulations to Kristoffer Wickstrøm for receiving the Best Paper Award 2024 from the journal Pattern Recognition Letters!
.jpg)
Professor Michael Kampffmeyer gave a presentation titled "Kunstig intelligens som forklarer hva den tenker" as part of a Norwegian Centre for E-Health Research Webinar (Norwegian dialogue).
I teorien ja, men i praksis fungerer det ikke slik.
EUGLOH students from partner institutions under the EUGLOH alliance can now apply for an exclusive mobility scholarship for the NLDL 2025 Winter School which covers travel, accommodation, and sustenance for successful applicants.
Visual Intelligence researchers represented the centre during Inspirasjonsdagen 2024. The event aimed to stimulate high school students' interest and curiosity for STEM and healthcare-related study programmes at UiT The Arctic University of Norway.
Our summer students have worked on projects related to fatigue and stress recognition with machine learning as their first research experience. Their results were presented at Georg Sverdrups hus at University of Oslo on October 16th 2024.
PhD candidate Rwiddhi Chakraborty recently gave an invited research talk titled "Perspectives on Multimodal Reasoning" at the Pioneer Centre for AI at University of Copenhagen.
Visual Intelligence was represented at Frampeik 2024 by associate professor Elisabeth Wetzer. The event gathered student researchers at Verdensteatret in Tromsø for discussions around AI-related topics.
Registration for the Northern Lights Deep Learning Conference 2025 is now open. The general deadline for registration is January 1st 2025.
We happily welcome Jan Zavadil, who recently started as a doctoral research fellow at Visual Intelligence in Tromsø in September 2024.
Visual Intelligence researchers participated in various dissemination activities throughout Forskningsdagene 2024. The activities aimed to disseminate general knowledge about deep learning and Visual Intelligence's research activities to the general public.
93 people from across the Visual Intelligence (VI) consortium gathered at Quality Airport Hotel Gardermoen for Visual Intelligence Days 2024, 24th to 25th of September.
Congratulations to Ghadi Al Hajj for successfully defending his PhD thesis and achieving the degree of Philosophiae Doctor at the University of Oslo on September 16th 2024!
Kommende søndag er det åpen dag på Nordnorsk vitensenter. Der lanseres vitensenteret siste nyvinning, en interaktiv utstilling basert på kunstig intelligens (KI).
Durgesh Kumar Singh is a PhD candidate at Visual Intelligence (VI) in Tromsø. He visited the University of Oslo and GE Vingmed Ultrasound from May to June 2024.
Maskinlæringsgruppen på UiT og SFI Visual Intelligence har arbeidet med å lage en utstilling som skal øke barns interesse, nysgjerrighet, og forståelse for kunstig intelligens. Nå vil den være utstillingsklar under et åpent familiearrangement på Nordnorsk vitensenter 22 september
Forskere har trent opp kunstig intelligens til å finne og kategorisere mikrofossiler. De tror arbeidet kan bli til stor nytte for geologer verden over (Norwegian news story at forskning.no).
Associate professor Kristoffer Wickstrøm represented Visual Intelligence at Nordnorsk ingeniørkonferanse 2024.
We happily welcome Christian Salomonsen, who recently started as a doctoral research fellow at Visual Intelligence in Tromsø.