News

Latest updates from the AI for Oncology Lab

Image for Equivariant CBCT reconstruction preprint
Equivariant CBCT reconstruction preprint

Medical image reconstruction with learned iterative schemes is particularly compute-heavy for inherently three-dimensional modalities such as Conebeam CT, and in our latest preprint we propose a novel architecture aimed for fast and memory efficient CBCT reconstruction

Read more
Image for New preprint on self-supervised MRI reconstruction
New preprint on self-supervised MRI reconstruction

Struggling with SSL methods in MRI reconstruction when ground truth data is unavailable? Our new pre-print on Joint Supervised and Self-supervised Learning (JSSL) might be the solution for training DL nets in the absence of fully sampled MRI data.

Read more
Image for New preprint on calibration of image segmentation algorithms
New preprint on calibration of image segmentation algorithms

Deep learning model outputs are generally not guaranteed to be well-calibrated. We propose a new approach, called "Kandinsky conformal prediction" for efficiently calibrating image segmentation models. Our method is built upon the framework of inductive conformal prediction, and is especially geared towards low-data scenarios such as medical image segmentation.

Read more
Image for aiEMBRACE project started
aiEMBRACE project started

On the 14th of November, we had the kickoff meeting of our aiEMBRACE project (AI-Enhanced personalized image-based breast cancer management). In this project, we will develop and validate models to improve the risk prediction, therapy response, and early recurrence of breast cancer.

Read more
Image for Cardiac MRI Reconstruction Winners at CMRxRecon Challenge
Cardiac MRI Reconstruction Winners at CMRxRecon Challenge

The AI for Oncology group at The Netherlands Cancer Institute secured the 2nd position in both tasks of the CMRxRecon Challenge during MICCAI 2023, demonstrating a novel 4D method for high-speed cardiac MRI reconstruction.

Read more
Image for New preprint on Accelerated MRI Reconstruction
New preprint on Accelerated MRI Reconstruction

Our latest preprint in MRI Reconstruction presents "vSHARP: A DL-Based Approach for Accelerated Parallel MRI Reconstruction," a novel method for solving complex inverse problems in medical imaging. vSHARP combines mathematical techniques for state-of-the-art MRI reconstruction.

Read more
Image for New publication about breast cancer primary treatment response assessment and prediction
New publication about breast cancer primary treatment response assessment and prediction

Primary systemic therapy (PST) is the treatment of choice in patients with locally advanced breast cancer and is nowadays also often used in patients with early-stage breast cancer. In our review, we critically discuss the literature on AI-based PST response prediction.

Read more
Image for New preprint about Constrained Empirical Risk Minimization
New preprint about Constrained Empirical Risk Minimization

Our latest research provides an innovative solution to the challenge of enforcing constraints on deep neural networks. We reframe the constraints on a network as an ordinary risk minimization problem on a Riemannian manifold.

Read more
Image for STAPLER: a language model to predict TCR–pMHC reactivity
STAPLER: a language model to predict TCR–pMHC reactivity

Our latest paper introduces STAPLER, a cutting-edge language model that significantly enhances TCR-pMHC reactivity prediction, outperforming previous models in the field.

A deeper understanding of TCR-pMHC interactions is key to unlocking the potential of personalized immunotherapies and expanding our knowledge of the immune system.

Read more
Image for GPU cluster expanded with gaia and galileo
GPU cluster expanded with gaia and galileo

We have added another 8xA6000 server (galileo) and a CPU server (gaia) to our Kosmos cluster. Kosmos now consists out of 70 GPUs, more than 1100 CPU cores, 6TB RAM and 1PB NAS.

Read more
Image for Retrospective k-space Subsampling schemes For Deep MRI Reconstruction
Retrospective k-space Subsampling schemes For Deep MRI Reconstruction

In our new publication, we investigate and compare various retrospective k-space subsampling patterns and their effect on the quality of DL-based reconstructions. Our findings suggest that non-rectilinear and non-Cartesian subsampling patterns may be more suitable for DL-based reconstructions.

Read more