Embedded Files
SCIA 2027
SCIA 2027
SCIA 2027 (Scandinavian Conference on Image Analysis) will be held at NTNU Gjøvik **June 9–11, 2027**. The conference is one of the most important gatherings for researchers and professionals in image analysis in the Nordic region, featuring expert keynotes, high-quality scientific presentations, and networking opportunities with leading figures from both academia and industry. Every two years, the **Best Nordic Thesis Prize** is also awarded to recognize outstanding research by emerging scientists.
Check out the website for information and registration: https://scia2027.org
Welcome to New Board Members
Welcome to New Board Members
NOBIM is delighted to welcome two new board members!
Annette Stahl is a Professor (Onsager Fellow) at the Department of Engineering Cybernetics at NTNU, and head of the Robotic Vision Group. Her research spans a broad range of topics within robotic and computer vision, including motion estimation, visual SLAM, underwater perception, and autonomous systems. She is the principal investigator of the AILARON project (funded by the Research Council of Norway) and affiliated with NTNU AMOS and SFI AutoShip. Annette holds a PhD in Applied Mathematics with a specialization in Computer Vision from Heidelberg University, Germany.
Xue-Cheng Tai is Chief Scientist at NORCE Research AS in Bergen, with research expertise in numerical methods for partial differential equations, inverse problems, image processing, and medical image analysis. He is internationally recognized as one of the inventors of the AOS (additive operator splitting) algorithm, a widely used method in PDE-based image processing. With over 250 publications in leading international journals and conferences, and membership on the editorial boards of SIAM Journal on Numerical Analysis and SIAM Journal on Imaging Science, Xue-Cheng brings exceptional academic breadth to the board.
We look forward to working with both of them!
Succesful Student Event at UiT
Succesful Student Event at UiT
On November 25, 2025, the **Student Association for Artificial Intelligence** at UiT – The Arctic University of Norway hosted a professional event with support from NOBIM. Robert Jenssen from Visual Intelligence was invited as a speaker and presented the research environment, ongoing projects, and concrete opportunities for students who wish to engage with artificial intelligence and machine learning.
A total of **18 students** attended the event, and the feedback was overwhelmingly positive. Many expressed that they gained valuable insight into how they can become part of the research community and bridge the gap between academia and industry — which is at the very core of NOBIM's mission to foster professional community across institutions and career stages.
NOBIM is proud to support initiatives that engage the next generation of researchers and professionals, and we encourage students and student associations at other institutions to reach out if they wish to organize similar events.
Featured Article: AI-Assisted Bone Marrow Image Analaysis
Featured Article: AI-Assisted Bone Marrow Image Analaysis
By Antony Gitau
Accurate and timely evaluation of bone marrow smears is critical for the diagnosis and treatment of hematological malignancies, such as multiple myeloma. These conditions are routinely identified based on the morphology of bone marrow cells. However, conventional manual assessment using a microscope is time-consuming and requires substantial specialist expertise. Recent efforts to digitize bone marrow smears — including at Vestfold Hospital Trust in Norway — create opportunities for AI-assisted morphological assessment.
The goal of Antony Gitau's doctoral work is to develop computer vision methods for analyzing digitized bone marrow whole-slide images (WSIs), enabling accurate, efficient, and explainable support for diagnosis. A key challenge in this setting is learning robust and generalizable cell-level representations that enable downstream tasks such as plasma cell quantification in WSIs.
As an initial step, the work focused on white blood cell (WBC) classification as part of the WBCBench 2026 challenge at IEEE ISBI 2026. This task served as a controlled setting to study how pathology foundation models (PFMs) learn representations that capture differences between morphologically similar cell types. In an oral presentation, a multi-stage end-to-end fine-tuning strategy using the DinoBloom-B foundation model was demonstrated [1]. This involved training multiple classifier heads with different geometries (cosine, linear, and multilayer perceptron), followed by ensembling their predictions to improve classification robustness [2]. This work provides insight into how PFM representations can be adapted for hematological cell classification and demonstrates their potential for robust cell-level modeling in bone marrow analysis.
[1] Koch, V., Wagner, S. J., Kazeminia, S., Sancar, E., Hehr, M., Schnabel, J. A., Peng, T., & Marr, C., "DinoBloom: A Foundation Model for Generalizable Cell Embeddings in Hematology," in *Proceedings of MICCAI 2024*, LNCS 15012, Springer, 2024, pp. 520–530.
[2] A. Gitau, M. Paulson, B.-J. Singstad, K. T. Hjelmervik, O. M. Lysaker, and V. G. Sanchez, "Multi-stage fine-tuning of pathology foundation models with head-diverse ensembling for white blood cell classification," arXiv:2603.20383, 2026.
Page updated
Google Sites
Report abuse