{"success":true,"data":{"author":"Julio Silva-Rodriguez, Adrian Colomer, Valery Naranjo et al. (Universitat Politecnica de Valencia); repackaged by Chehab Lab","bucketAccessType":"public","bucketName":"chehablab/SICAPv2","bucketPrefix":"","category":"Health & Biomedical","cloudProvider":"huggingface","createdAt":"2026-11-21T22:16:41.592Z","description":"SICAPv2 is a prostate cancer histopathology database built from H&E-stained whole slide images of prostate biopsies, annotated with a global Gleason score per slide and Gleason grades for regions within it. This repackaging provides the labeled image patches extracted from those slides, each tagged with its slide and patient, a patch-level grade (NC non-cancerous, G3, G4 or G5 Gleason pattern), a flag for the cribriform Gleason 4 pattern and an integer label. Patches are the original JPEGs, unchanged; the original cross-validation folds, test partition and segmentation masks are not included, so split by patient_id. Four Parquet files under data/, about 1.4 GB (measured 2026-11-21); read with Hugging Face datasets. Use for automated Gleason grading, prostate cancer detection and weakly supervised pathology models. Original data on Mendeley Data (DOI 10.17632/9xxm58dvs3.2), released under CC BY 4.0. Cite Silva-Rodriguez et al., Computer Methods and Programs in Biomedicine 195:105637 (2020).","displayTitle":"SICAPv2: prostate biopsy H&E patches with Gleason grade labels (NC, G3, G4, G5) and cribriform pattern flag","doi":"10.17632/9xxm58dvs3.2","huggingFaceUrl":"https://huggingface.co/datasets/chehablab/SICAPv2","id":"77993928-c4e8-57df-8329-8b61499a4295","license":"CC-BY-4.0","licenseEvidenceUrl":"https://data.mendeley.com/public-api/datasets/9xxm58dvs3?version=2","objectCountEstimate":6,"probeStatus":"ok","referenceUrl":"https://doi.org/10.1016/j.cmpb.2020.105637","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"chehablab/SICAPv2\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"chehablab/SICAPv2\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=chehablab, license:cc-by-4.0, size_categories:10K<n<100K, format:parquet, downloads=107"],"format":"mixed","generatedAt":"2026-11-21T22:32:06.825Z","layout":{"example":"data/train-00000-of-00004.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/train-00000-of-00004.parquet","truncatedAt":512}],"sampleObjects":["data/train-00000-of-00004.parquet","data/train-00001-of-00004.parquet","data/train-00002-of-00004.parquet","data/train-00003-of-00004.parquet"],"sidecars":[{"bytes":"---\ndataset_info:\n  features:\n  - name: image_id\n    dtype: string\n  - name: image\n    dtype: image\n  - name: slide_id\n    dtype: string\n  - name: patient_id\n    dtype: string\n  - name: grade\n    dtype: string\n  - name: cribriform\n    dtype: bool\n  - name: label\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 1528412239\n    num_examples: 12081\n  download_size: 1412503832\n  dataset_size: 1528412239\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- histopathology\n- prostate\n- prostate-cancer\n- gleason\n- h&e\n- medical\n- classification\npretty_name: SICAPv2\n---\n# SICAPv2 — Prostate Histology Patches with Gleason Grade Annotations\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/SICAPv2/discussions) of this repository.\n\nThis dataset is derived from SICAPv2 (Silva-Rodríguez et al., 2020), a database of prostate histology whole slide images with both global Gleason score annotations and patch-level Gleason grade annotations, originally hosted on Mendeley Data.\nHere, we provide the labelled image patches. We kept the original JPG patches unchanged and added the slide ID, the patient ID, the Gleason grade, a cribriform flag and an integer label from the original annotation files.\nThe dataset is designed for patch-level Gleason grading of prostate histology.\n\n\n## 📦 Dataset Structure\n\nThe dataset has a single `train` split, with each row corresponding to a single H&E image patch:\n\n| Field | Description |\n|---|---|\n| `image_id` | Original patch filename without extension |\n| `image` | RGB H&E-stained prostate histology patch (JPG) |\n| `slide_id` | Whole slide image the patch was extracted from (the prefix of the patch filename) |\n| `patient_id` | Patient ID from the original whole-slide label file |\n| `grade` | Patch-level Gleason grade: `NC`, `G3`, `G4` or `G5` |\n| `cribriform` | `True` if the patch is annotated with the cribriform Gleason 4 pattern |\n| `label` | Integer grade: `0` = NC, `1` = G3, `2` = G4, `3` = G5 |\n\n\n## ⚙️ Preprocessing\n\n- `SICAPv2.zip` was downloaded from Mendeley Data (`9xxm58dvs3`, version 2)\n- Patches were kept as the original JPG files (no re-encoding, resizing or color normalization)\n- Patch labels were taken from the original partition files, which give each patch a one-hot Gleason grade (`NC`, `G3`, `G4`, `G5`) and a cribriform flag (`G4C`)\n- `slide_id` was taken from the patch filename, and `patient_id` from the original whole-slide label file\n- `label` is the index of `grade` in the list `NC, G3, G4, G5`\n- The original partitions (cross-validation folds and test set) and the segmentation masks were not included\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/SICAPv2\", split=\"train\")\n\n# One example patch per Gleason grade\ngrades = [\"NC\", \"G3\", \"G4\", \"G5\"]\nlabels = ds[\"label\"]\nfig, axes = plt.subplots(1, len(grades), figsize=(12, 3))\nfor ax, label in zip(axes, range(len(grades))):\n    sample = ds[labels.index(label)]\n    ax.imshow(sample[\"image\"])\n    ax.set_title(f\"{grades[label]} | slide {sample['slide_id']}\")\n    ax.axis(\"off\")\nplt.tight_layout()\nplt.show()\n\n# 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### Gleason Grades\nPatch-level grades following the Gleason grading system for prostate adenocarcinoma:\n\n- NC (`label=0`) — Non-cancerous tissue\n- G3 (`label=1`) — Gleason pattern 3: well-formed, discrete glands\n- G4 (`label=2`) — Gleason pattern 4: fused, poorly formed or cribriform glands","encoding":"utf8","key":"README.md","name":"README.md","size":7335,"truncated":true}],"sourceObjectKey":"data/train-00000-of-00004.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"chehablab/SICAPv2\")\nprint(ds)","starterSamplePath":"data/train-00000-of-00004.parquet","totalObjects":6},"summary":"SICAPv2 is a prostate cancer histopathology database built from H&E-stained whole slide images of prostate biopsies, annotated with a global Gleason score per slide and Gleason…","title":"SICAPv2: prostate biopsy H&E patches with Gleason grade labels (NC, G3, G4, G5) and cribriform pattern flag","updatedAt":"2026-11-22T00:59:58.128Z","visibility":"public","websiteUrl":"https://data.mendeley.com/datasets/9xxm58dvs3/2"}}