{"success":true,"data":{"author":"Wilfrido Gomez-Flores, Maria Julia Gregorio-Calas, Wagner Coelho de Albuquerque Pereira; repackaged by Chehab Lab","bucketAccessType":"public","bucketName":"chehablab/BUS-BRA","bucketPrefix":"","category":"Health & Biomedical","cloudProvider":"huggingface","createdAt":"2026-11-21T22:16:41.618Z","description":"BUS-BRA is a breast ultrasound (BUS) dataset of 1,875 anonymized images from 1,064 patients who underwent routine breast studies in Brazil. Every tumor is biopsy-proven: 1,268 images show benign lesions and 607 malignant ones, and each carries its BI-RADS category (2 to 5) as assigned by the radiologist. The repackaging keeps the original PNGs unchanged and adds patient ID, pathology, BI-RADS and an integer label (0 benign, 1 malignant); split by patient_id since many patients have two views. The original release also includes tumor outlines, which are not in this copy. One Parquet file under data/, about 130 MB (measured 2026-11-21); read with Hugging Face datasets. Use for benign versus malignant breast tumor classification and BI-RADS prediction from ultrasound. Original data on Zenodo (DOI 10.5281/zenodo.8231412), released under CC BY 4.0. Cite Gomez-Flores et al., Medical Physics 51(4):3110-3123 (2024).","displayTitle":"BUS-BRA: 1,875 breast ultrasound images from 1,064 patients with biopsy-proven benign or malignant labels and BI-RADS (Brazil)","doi":"10.5281/zenodo.8231412","huggingFaceUrl":"https://huggingface.co/datasets/chehablab/BUS-BRA","id":"b1d16a99-ab44-5e48-8f76-480672ba406f","license":"CC-BY-4.0","licenseEvidenceUrl":"https://zenodo.org/records/8231412","objectCountEstimate":3,"probeStatus":"ok","referenceUrl":"https://zenodo.org/records/8231412","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"chehablab/BUS-BRA\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"chehablab/BUS-BRA\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=chehablab, license:cc-by-4.0, size_categories:1K<n<10K, format:parquet, downloads=101"],"format":"mixed","generatedAt":"2026-11-21T22:32:07.224Z","layout":{"example":"data/train-00000-of-00001.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/train-00000-of-00001.parquet","truncatedAt":512}],"sampleObjects":["data/train-00000-of-00001.parquet"],"sidecars":[{"bytes":"---\ndataset_info:\n  features:\n  - name: image_id\n    dtype: string\n  - name: image\n    dtype: image\n  - name: patient_id\n    dtype: string\n  - name: pathology\n    dtype: string\n  - name: birads\n    dtype: string\n  - name: label\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 117109923\n    num_examples: 1875\n  download_size: 132131766\n  dataset_size: 117109923\nconfigs:\n- config_name: default\n  data_files:\n  - split: train\n    path: data/train-*\nlicense: cc-by-4.0\ntask_categories:\n- image-classification\nlanguage:\n- en\ntags:\n- ultrasound\n- breast\n- breast-cancer\n- bi-rads\n- medical\n- classification\npretty_name: BUS-BRA\nsize_categories:\n- 1K<n<10K\n---\n\n# BUS-BRA — Breast Ultrasound Images for Computer-Aided Diagnosis\n\n> [!NOTE]\n> This dataset card was generated with the help of generative AI and checked against the original sources. If you find an error or anything unclear, please open a discussion in the [Community tab](https://huggingface.co/datasets/chehablab/BUS-BRA/discussions) of this repository.\n\nThis dataset is derived from BUS-BRA (Gómez-Flores et al., 2024), a publicly available dataset of anonymized breast ultrasound (BUS) images of 1,064 patients who underwent routine breast studies, originally hosted on Zenodo. The dataset contains biopsy-proven tumor cases with BI-RADS annotations in categories 2, 3, 4 and 5.\nWe kept the original PNG images unchanged and added the patient ID, the pathology, the BI-RADS category and an integer label from the original metadata file.\nThe dataset is designed for benign vs. malignant breast tumor classification.\n\n\n## 📦 Dataset Structure\n\nThe dataset has a single `train` split, with each row corresponding to a single ultrasound image:\n\n| Field | Description |\n|---|---|\n| `image_id` | Original image ID (e.g., `bus_0001-l`) |\n| `image` | Breast ultrasound image (variable size, PNG) |\n| `patient_id` | Patient (case) number from the original metadata |\n| `pathology` | `benign` or `malignant` |\n| `birads` | BI-RADS category: `2`, `3`, `4` or `5` |\n| `label` | Integer pathology: `0` = benign, `1` = malignant |\n\nIn total, the dataset contains 1,875 images from 1,064 patients.\n\n**Class Distribution:**\n\n| Pathology | BI-RADS 2 | BI-RADS 3 | BI-RADS 4 | BI-RADS 5 | Total |\n|---|---|---|---|---|---|\n| Benign | 552 | 432 | 280 | 4 | 1,268 |\n| Malignant | 10 | 31 | 413 | 153 | 607 |\n| Total | 562 | 463 | 693 | 157 | 1,875 |\n\n\n## ⚙️ Preprocessing\n\n- `BUSBRA.zip` was downloaded from Zenodo (record `8231412`, version 1.0)\n- Images were kept as the original PNG files (no re-encoding, resizing or cropping)\n- Each image was matched to its row in the original metadata file by its ID\n- `patient_id`, `pathology` and `birads` were taken from the `Case`, `Pathology` and `BIRADS` columns\n- `label` is `1` for malignant and `0` for benign\n- The tumor delineation masks of the original release were not included\n- All images listed in the metadata file were preserved\n\n\n## 🚀 Usage\n\n```python\nfrom datasets import load_dataset\nimport matplotlib.pyplot as plt\nfrom sklearn.model_selection import GroupShuffleSplit\n\nds = load_dataset(\"chehablab/BUS-BRA\", split=\"train\")\n\n# One benign and one malignant example\nlabels = ds[\"label\"]\nfig, axes = plt.subplots(1, 2, figsize=(8, 4))\nfor ax, label in zip(axes, [0, 1]):\n    sample = ds[labels.index(label)]\n    ax.imshow(sample[\"image\"], cmap=\"gray\")\n    ax.set_title(f\"{sample['pathology']} | BI-RADS {sample['birads']}\")\n    ax.axis(\"off\")\nplt.tight_layout()\nplt.show()\n\n# No official split: split by patient so no patient appears in both sets\ngss = GroupShuffleSplit(n_splits=1, test_size=0.2, random_state=0)\ntrain_idx, test_idx = next(gss.split(range(len(ds)), groups=ds[\"patient_id\"]))\ntrain, test = ds.select(train_idx), ds.select(test_idx)\n```\n\n---\n\n## 🏷️ Feature Descriptions\n\n### Pathology\n- Benign (`label=0`): 1,268 images\n- Malignant (`label=1`): 607 images\n\n### BI-RADS\nThe Breast Imaging Reporting and Data System (BI-RADS) category assigned to each image:\n\n- BI-RADS 2 — Benign finding\n- BI-RADS 3 — Probably benign\n- BI-RADS 4 — Suspicious for ","encoding":"utf8","key":"README.md","name":"README.md","size":7496,"truncated":true}],"sourceObjectKey":"data/train-00000-of-00001.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"chehablab/BUS-BRA\")\nprint(ds)","starterSamplePath":"data/train-00000-of-00001.parquet","totalObjects":3},"summary":"BUS-BRA is a breast ultrasound (BUS) dataset of 1,875 anonymized images from 1,064 patients who underwent routine breast studies in Brazil.","title":"BUS-BRA: 1,875 breast ultrasound images from 1,064 patients with biopsy-proven benign or malignant labels and BI-RADS (Brazil)","updatedAt":"2026-11-22T00:52:54.397Z","visibility":"public","websiteUrl":"https://zenodo.org/records/8231412"}}