CT-RATE

The CT-RATE Team organizes the VLM3D Challenge VLM3D 2026 (2nd Edition) → Challenge Finals at MICCAI 2026 VLM3D 2025 (1st Edition) → Challenge Finals at MICCAI 2025 • Workshop at ICCV 2025 The CT-RATE Team is developing the MR-RATE Dataset A large-scale brain MRI dataset with paired radiology reports for training 3D vision-language models. GitHub | Dataset | Metadata Dashboard Generalist Foundation Models from a Multimodal Dataset for 3D Computed Tomography Welcome to the official page for our paper, which introduces CT-RATE—a pioneering dataset in 3D medical imaging that uniquely pairs textual data with image data focused on chest CT volumes. Here, you will find the CT-RATE dataset, comprising chest CT volumes paired with corresponding radiology text reports, multi-abnormality labels, and metadata, all freely accessible to researchers. CT-RATE: A novel dataset of chest CT volumes with corresponding radiology text reports A major challenge in computational research in 3D medical imaging is the lack of comprehensive datasets. Addressing this issue, we present CT-RATE, the first 3D medical imaging dataset that pairs images with textual reports. CT-RATE consists of 25,692 non-contrast chest CT volumes, expanded to 50,188 through various reconstructions, from 21,304 unique patients, along with corresponding radiology text reports, multi-abnormality labels, and metadata. We divided the cohort into two groups: 20,000 patients were allocated to the training set and 1,304 to the validation set. Our folders are structured as split_patientID_scanID_reconstructionID. For instance, "valid_53_a_1" indicates that this is a CT volume from the validation set, scan "a" from patient 53, and reconstruction 1 of scan "a". This naming convention applies to all files. CT-CLIP: CT-focused contrastive language-image pre-training framework Leveraging CT-RATE, we developed CT-CLIP, a CT-focused contrastive language-image pre-training framework. As a versatile, self-supervised model, CT-CLIP