{"success":true,"data":{"author":"US FDA Division of Imaging, Diagnostics, and Software Reliability (Wiedeman, Sarmakeeva, Sizikova, Filienko, Lago, Delfino, Badano)","bucketAccessType":"public","bucketName":"didsr/tsynth","bucketPrefix":"","category":"Health & Biomedical","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:11.960Z","description":"T-SYNTH is a large open dataset of synthetic breast images from the US FDA's Division of Imaging, Diagnostics, and Software Reliability, generated with physics simulation using the VICTRE (Virtual Imaging Clinical Trial for Regulatory Evaluation) toolkit and MC-GPU Monte Carlo x-ray transport, replicating a Siemens detector. It pairs 2D digital mammography with 3D digital breast tomosynthesis (DBT), with DBT also projected to synthetic C-View images. The card describes 9,000 samples, half lesion-free and half with masses, across four breast density categories (fatty, scattered, heterogeneously dense, dense), mass radii of 5, 7 and 9 mm and three relative mass densities. Lesion masks and bounding boxes are generated automatically from the phantoms, so labels are exact. It complements M-SYNTH, the companion synthetic mammography release. As measured 2026-11-14 the repository uses about 64 GB, with images stored as .raw arrays under data/ (including cview and metadata folders); read them with numpy using the documented dimensions. Used for lesion detection and segmentation in DM and DBT, regulatory-science studies and augmentation of small patient datasets. Released by the FDA authors under CC0 1.0; cite Wiedeman et al. 2025 (arXiv:2507.04038).","displayTitle":"T-SYNTH: synthetic digital breast tomosynthesis and mammography with lesion masks (FDA)","id":"13e209f5-3987-58f9-a877-be70b2eb0a52","license":"CC0-1.0","licenseEvidenceUrl":"https://huggingface.co/datasets/didsr/tsynth","objectCountEstimate":601,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/2507.04038","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"didsr/tsynth\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"didsr/tsynth\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=didsr, license:cc0-1.0, size_categories:1K<n<10K, downloads=210"],"format":"mixed","generatedAt":"2026-11-14T14:15:16.078Z","layout":{"example":"data/embed_metadata/EMBED_split_dbt.csv","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":",Unnamed: 0,empi_anon,total_num_roi,entry_count,breast_density,group\n0,0,10000879,0,4,3.0,train\n1,1,10015693,0,8,2.0,train\n2,2,10029585,0,12,3.0,train\n3,3,10033806,0,8,3.0,train\n4,4,10042753,0,18,2.0,train\n5,5,10043955,0,8,3.0,train\n6,6,10064524,0,16,3.0,train\n7,7,10065686,0,12,3.0,train\n8,8,10093833,0,17,3.0,train\n9,9,10098626,0,8,3.0,train\n10,10,10138685,0,20,1.0,train\n11,11,10141610,0,25,2.0,train\n12,12,10144276,0,4,3.0,train\n13,13,10147772,0,8,2.0,train\n14,14,10156863,0,4,4.0,train\n15,15,10173804,0,20,3","encoding":"utf8","key":"data/embed_metadata/EMBED_split_dbt.csv","truncatedAt":512}],"sampleObjects":["data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/dense/2/5.0/SIM.zip","data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/dense/2/7.0/SIM.zip","data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/fatty/2/5.0/SIM.zip","data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/fatty/2/7.0/SIM.zip","data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/fatty/2/9.0/SIM.zip"],"sidecars":[{"bytes":"---\nlicense: cc0-1.0\ntask_categories:\n- object-detection\n- image-segmentation\ntags:\n- medical\npretty_name: T-SYNTH\nsize_categories:\n- 1K<n<10K\n---\n# T-SYNTH\nPaper: [T-SYNTH: A Knowledge-Based Dataset of Synthetic Breast Images](https://huggingface.co/papers/2507.04038)\n\n\n<!-- Provide a quick summary of the dataset. -->\n\nT-SYNTH is a synthetic digital breast tomosynthesis (DBT) dataset with four breast fibroglandular density distributions imaged using Monte Carlo x-ray simulations with the publicly available [Virtual Imaging Clinical Trial for Regulatory Evaluation (VICTRE)](https://github.com/DIDSR/VICTRE) toolkit.\n\n## Dataset Details\n\nThe dataset has the following characteristics:\n\n* Breast density: dense, heterogeneously dense, scattered, fatty\n* Mass radius (mm): 5.00, 7.00, 9.00\n* Mass density: 1.0, 1.06, 1.1 (ratio of mass radiodensity to that of fibroglandular tissue)\n\n\n### Dataset Description\n\n<!-- Provide a longer summary of what this dataset is. -->\n\n- **Curated by:** [Christopher Wiedeman](https://www.linkedin.com/in/christopher-wiedeman-a0b01014b), [Anastasiia Sarmakeeva](https://www.linkedin.com/in/anastasiia-sarmakeeva/), [Elena Sizikova](https://elenasizikova.github.io/), [Daniil Filienko](https://www.linkedin.com/in/daniil-filienko-800160215/), [Miguel Lago](https://www.linkedin.com/in/milaan/), [Jana Gut Delfino](https://www.linkedin.com/in/janadelfino/), [Aldo Badano](https://www.linkedin.com/in/aldobadano/)\n- **License:** Creative Commons 1.0 Universal License (CC0)\n\n## Data Acquisition Details\n\n**Imaging Modality:** Paired 2D digital mammography (DM) and 3D digital breast tomosynthesis (DBT) images. The DBT images are projected into C-VIEW via the method of (Klein, 2023). \n\n**Manufacturer and Model:** Replica of the Siemens detector based on MC-GPU (Badal and Badano, 2009).\n\n**Demographics:** All breast phantoms are synthetic and do not represent real human subjects.\n\n**Cohort Description:** 9,000 synthetic digital breast tomosynthesis (DBT) samples, distributed across four breast density categories:\n\n| Breast Density        | Fatty     | Scattered | Hetero  | Dense   | **Total** |\n| --------- | --------- | --------- | ------- | ------- | --------- |\n| **Les.-free / Les.-present** | 1350/1350 | 1350/1350 | 900/900 | 900/900 | 4500/4500 |\n\n\n**Annotation Protocols:** Lesion masks and bounding boxes were generated automatically from the phantom.\n\n**Data Format and Structure:** Image files are in .raw format.\n\n### Dataset Sources\n\n<!-- Provide the basic links for the dataset. -->\n\n- **Code:** [https://github.com/DIDSR/tsynth-release](https://github.com/DIDSR/tsynth-release)\n- **Arxiv:** [https://arxiv.org/abs/2507.04038](https://arxiv.org/abs/2507.04038)\n- **Poster:** [https://github.com/DIDSR/tsynth-release/blob/main/images/poster.pdf](https://github.com/DIDSR/tsynth-release/blob/main/images/poster.pdf)\n\n## Intended Use\n\n<!-- Address questions around how the dataset is intended to be used. -->\n\nT-SYNTH is intended to facilitate testing of AI with pre-computed synthetic digital breast tomosynthesis (DBT) data, complementing the M-SYNTH synthetic mammography dataset.\n\n\n## Ethical Considerations\nThis work is using synthetically generated data and does not include any real patient-identifiable information. Publication of synthetic data aims to promote transparency, \nreproducibility, and fairness in medical AI research. We recommend avoiding training models only on synthetic data without appropriate validation.\n\n\n## Dataset Structure\n\n<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->\n\nDirectory layout:\n```\nT-SYNTH/data/\n├── cview\n├── embed_metadata\n├── pretrained_models\n├── results\n└── volumes_subset\n\n```\nDescriptions:\n\n* **`cview/`** -- contains T-SYNTH C-VIEW images.\n\n* **`embed_metadata/`** -- Configuration files needed to reproduce experiments.\n\n* **`pretrained_models/`**  -- Pretrained models for ```DBT```, ```DM``` and `","encoding":"utf8","key":"README.md","name":"README.md","size":6529,"truncated":true}],"sourceObjectKey":"data/embed_metadata/EMBED_split_dbt.csv","starterCell":"import pandas as pd\ndf = pd.read_csv(\"hf://datasets/didsr/tsynth/data/embed_metadata/EMBED_split_dbt.csv\")\ndf.head()","starterSamplePath":"data/cview/output_cview_det_Victre/device_data_VICTREPhantoms_spic_1.0/dense/2/5.0/SIM.zip","totalObjects":601},"summary":"T-SYNTH is a large open dataset of synthetic breast images from the US FDA's Division of Imaging, Diagnostics, and Software Reliability, generated with physics simulation using…","title":"T-SYNTH: synthetic digital breast tomosynthesis and mammography with lesion masks (FDA)","updatedAt":"2026-11-14T17:52:54.141Z","visibility":"public","websiteUrl":"https://github.com/DIDSR/tsynth-release"}}