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.
I am eager to absorb knowledge from these experts and contribute to our collective understanding, says Filippo Bascetta, who is one of two new VI interns.
Listen to VI co-director Line Eikvil and research director Anders Løland (Norsk Regnesentral) talk about developments before and after the breakthrough of deep learning in the NR podcast, Sannsynligvis VIKTIG (Norwegian dialogue).
The delegation lasted from April 23rd to 27th and aimed to develop and strengthen existing and future collaborations between UiT and Indian universities and research institutions.
We congratulate Adín Ramírez Rivera (UiO), Michael Kampffmeyer (UiT) and Andrea Marinoni (UiT), who were recently promoted to professors at their respective universities.
Scientists from UiT and University of Bergen gathered to discuss gender imbalance and discrimination in academia, as well as initiatives which aim to ensure diversity and inclusion in research, academia and work life.
We congratulate Mathias K. Hauglid for successfully defending his PhD dissertation and achieving the PhD degree in Law at UiT The Arctic University of Norway on April 26th 2024.
I år samler SFI Visual Intelligence, som ledes av Forskningsgruppen for maskinlæring ved UiT, og Lie Størmer senter kreftene for å markere dagen ved Verdensteateret i Tromsø den 6. mai. Arrangementet er gratis og åpent for alle interesserte.
Heggeli's primary work tasks involve managing, maintaining and optimizing the GPU cluster at UiT Machine Learning (ML) Group in Tromsø, while serving as an important advising resource for our UiT-based VI researchers.
The company's interest in Visual Intelligence is to utilize new developments in artificial intelligence (AI) to improve and automate interpretations of data, while providing the centre with data, problem formulations, and different forms of guidance.
The presented research is part of a larger collaboration between researchers from UiT The Arctic University of Norway and the University Hospital of North Norway.
Visual Intelligence researchers from UiT Machine Learning Group and the University Hospital of North Norway (UNN) were recently featured in the "Abels Tårn" spin-off show "Den store serien om KI".
While visiting Tromsø, Fua held two talks on the topics of "Deep Uncertainty" and "Deep 3D Surface Meshes" for students and researchers at UiT The Arctic University of Norway.
The report showcases Visual Intelligence's activities, results, staff and funding throughout the past year, serving as a detailed compilation of our researchers' efforts towards further development within the centre's four main innovation areas and solving our four research chall
Kunstig intelligens treffer dårligere på kvinner enn menn. Skjevhet i dataene er hovedutfordringen, mener Suaiba Salahuddin, som peker på tre forbedringsmuligheter.
Visual Intelligence Director Robert Jenssen recently gave an invited talk at the CatchID conference organized in Tromsø from 19th-20th March 2024 by the Norwegian Directorate of Fisheries.
The exclusive Lindau Nobel Laureate Meetings gather 650 distinguished scientists and 30 Nobel Prize winners from across the world and places strict requirements in terms of academic and scientific quality. – A great honor, says associate professor Elisabeth Wetzer.
Director Robert Jenssen gave a talk on the uses and misuses of artificial intelligence for members of the seniors association at UiT The Arctic University of Norway. The UiT seniors association consists of former academic and administrative employees at the university.
Researcher Anders U. Waldeland from The Norwegian Computing Center gave a talk at The Biennial Geophysical Seminar in Kristiansand presenting work from Visual Intelligence on seismic structure search using a neural network trained with self-supervised learning.
State secretary Tomas Norvoll visited Visual Intelligence and was given a tour at the PET center at The University Hospital of North Norway by centre director Robert Jenssen and associate professor Samuel Kuttner to learn more about research and innovation within PET and AI.
We are thrilled to announce that the paper "Interrogating Sea Ice Predictability With Gradients", a collaborative paper between Visual Intelligence, The Alan Turing Institute and the British Antarctic Survey, has been accepted in IEEE Geoscience and Remote Sensing Letters.