Curia-2: Scaling Self-Supervised Learning for Radiology Foundation Models

• URL: http://arxiv.org/abs/2604.01987v1
• Date: Thu, 02 Apr 2026 12:49:38 GMT
• Title: Curia-2: Scaling Self-Supervised Learning for Radiology Foundation Models
• Abstract summary: We introduce Curia-2, which improves the original pre-training strategy and representation quality to better capture the specificities of radiological data.<n>Curia-2 outperforms all FMs on vision-focused tasks and fairs competitively to vision-language models on clinically complex tasks such as finding detection.
• Score: 5.468585846077723
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: The rapid growth of medical imaging has fueled the development of Foundation Models (FMs) to reduce the growing, unsustainable workload on radiologists. While recent FMs have shown the power of large-scale pre-training to CT and MRI analysis, there remains significant room to optimize how these models learn from complex radiological volumes. Building upon the Curia framework, this work introduces Curia-2, which significantly improves the original pre-training strategy and representation quality to better capture the specificities of radiological data. The proposed methodology enables scaling the architecture up to billion-parameter Vision Transformers, marking a first for multi-modal CT and MRI FMs. Furthermore, we formalize the evaluation of these models by extending and restructuring CuriaBench into two distinct tracks: a 2D track tailored for slice-based vision models and a 3D track for volumetric benchmarking. Our results demonstrate that Curia-2 outperforms all FMs on vision-focused tasks and fairs competitively to vision-language models on clinically complex tasks such as finding detection. Weights will be made publicly available to foster further research.

Related papers

• GreenRFM: Toward a resource-efficient radiology foundation model [13.676606904401433]
  We present a resource-efficient pre-training framework, GreenRFM, that achieves state-of-the-art performance.<n>Our framework ensures robust generalization across diverse patient populations and imaging protocols.<n>We conduct extensive experiments using over 200,000 images from four institutions and of two modalities.
  arXiv  Detail & Related papers  (2026-03-06T16:51:42Z)
• Reason Like a Radiologist: Chain-of-Thought and Reinforcement Learning for Verifiable Report Generation [13.580272788409092]
  BoxMed-RL is a groundbreaking unified training framework for generating spatially verifiable and explainable radiology reports.<n>Built on a large vision-language model, BoxMed-RL revolutionizes report generation through two integrated phases.<n>BoxMed-RL achieves an average 7% improvement in both METEOR and ROUGE-L metrics compared to state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-04-25T16:05:06Z)
• Training state-of-the-art pathology foundation models with orders of magnitude less data [1.7005561101170015]
  We present three novel vision foundation models (FMs) trained on up to two orders of magnitude fewer WSIs than those used to train other state-of-the-art FMs.<n>Even the model trained on TCGA alone (12k WSIs) outperforms most existing FMs and, on average, matches Virchow2, the second-best FM published to date.
  arXiv  Detail & Related papers  (2025-04-07T15:38:12Z)
• Vision Foundation Models for Computed Tomography [0.5320113414681007]
  Foundation models (FMs) have shown transformative potential in radiology by performing diverse, complex tasks across imaging modalities.<n>Here, we developed CT-FM, a large-scale 3D image-based pre-trained model designed explicitly for various radiological tasks.<n>CT-FM was pre-trained using 148,000 computed tomography (CT) scans from the Imaging Data Commons through label-agnostic contrastive learning.
  arXiv  Detail & Related papers  (2025-01-15T18:30:58Z)
• 3D-CT-GPT: Generating 3D Radiology Reports through Integration of Large Vision-Language Models [51.855377054763345]
  This paper introduces 3D-CT-GPT, a Visual Question Answering (VQA)-based medical visual language model for generating radiology reports from 3D CT scans. Experiments on both public and private datasets demonstrate that 3D-CT-GPT significantly outperforms existing methods in terms of report accuracy and quality.
  arXiv  Detail & Related papers  (2024-09-28T12:31:07Z)
• MGH Radiology Llama: A Llama 3 70B Model for Radiology [50.42811030970618]
  This paper presents an advanced radiology-focused large language model: MGH Radiology Llama.<n>It is developed using the Llama 3 70B model, building upon previous domain-specific models like Radiology-GPT and Radiology-Llama2.<n>Our evaluation, incorporating both traditional metrics and a GPT-4-based assessment, highlights the enhanced performance of this work over general-purpose LLMs.
  arXiv  Detail & Related papers  (2024-08-13T01:30:03Z)
• Summarizing Radiology Reports Findings into Impressions [1.8964110318127383]
  We present a model with state-of-art radiology report summarization performance. We also provide an analysis of the model limitations and radiology knowledge gain. Our best performing model was a fine-tuned BERT-to-BERT encoder-decoder with 58.75/100 ROUGE-L F1.
  arXiv  Detail & Related papers  (2024-05-10T20:29:25Z)
• Evaluating General Purpose Vision Foundation Models for Medical Image Analysis: An Experimental Study of DINOv2 on Radiology Benchmarks [5.8941124219471055]
  DINOv2 is an open-source foundation model pre-trained with self-supervised learning on 142 million curated natural images. This study comprehensively evaluates the performance DINOv2 for radiology.
  arXiv  Detail & Related papers  (2023-12-04T21:47:10Z)
• Towards Generalist Foundation Model for Radiology by Leveraging Web-scale 2D&3D Medical Data [66.9359934608229]
  This study aims to initiate the development of Radiology Foundation Model, termed as RadFM. To the best of our knowledge, this is the first large-scale, high-quality, medical visual-language dataset, with both 2D and 3D scans. We propose a new evaluation benchmark, RadBench, that comprises five tasks, including modality recognition, disease diagnosis, visual question answering, report generation and rationale diagnosis.
  arXiv  Detail & Related papers  (2023-08-04T17:00:38Z)
• Customizing General-Purpose Foundation Models for Medical Report Generation [64.31265734687182]
  The scarcity of labelled medical image-report pairs presents great challenges in the development of deep and large-scale neural networks. We propose customizing off-the-shelf general-purpose large-scale pre-trained models, i.e., foundation models (FMs) in computer vision and natural language processing.
  arXiv  Detail & Related papers  (2023-06-09T03:02:36Z)
• Competence-based Multimodal Curriculum Learning for Medical Report Generation [98.10763792453925]
  We propose a Competence-based Multimodal Curriculum Learning framework ( CMCL) to alleviate the data bias and make best use of available data. Specifically, CMCL simulates the learning process of radiologists and optimize the model in a step by step manner. Experiments on the public IU-Xray and MIMIC-CXR datasets show that CMCL can be incorporated into existing models to improve their performance.
  arXiv  Detail & Related papers  (2022-06-24T08:16:01Z)
• Pre-training and Fine-tuning Transformers for fMRI Prediction Tasks [69.85819388753579]
  TFF employs a transformer-based architecture and a two-phase training approach. Self-supervised training is applied to a collection of fMRI scans, where the model is trained for the reconstruction of 3D volume data. Results show state-of-the-art performance on a variety of fMRI tasks, including age and gender prediction, as well as schizophrenia recognition.
  arXiv  Detail & Related papers  (2021-12-10T18:04:26Z)
• Modelling the Distribution of 3D Brain MRI using a 2D Slice VAE [66.63629641650572]
  We propose a method to model 3D MR brain volumes distribution by combining a 2D slice VAE with a Gaussian model that captures the relationships between slices. We also introduce a novel evaluation method for generated volumes that quantifies how well their segmentations match those of true brain anatomy.
  arXiv  Detail & Related papers  (2020-07-09T13:23:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.