Datasets:

MaybeRichard
/

zju-eye-pretrain

---
license: other
license_name: ophthalmology-mixed
license_link: https://github.com/mayberichard/zju-eye-pretrain/blob/main/LICENSE
task_categories:
- image-classification
- image-segmentation
- image-to-image
- text-to-image
- unconditional-image-generation
language:
- en
- zh
tags:
- ophthalmology
- retina
- oct
- fundus
- slo
- medical-imaging
- segmentation
- pretraining
size_categories:
- 1M<n<10M
pretty_name: ZJU Eye-Pretrain (Private Shanghai Topcon + 25 public cohorts)
configs:
- config_name: private_topcon
  data_files:
  - split: train
    path: data/private_topcon/*.parquet
- config_name: public_fundus
  data_files:
  - split: train
    path: data/public_fundus/*.parquet
- config_name: public_oct
  data_files:
  - split: train
    path: data/public_oct/*.parquet
- config_name: drive_vessel
  data_files:
  - split: train
    path: data/public_fundus/public_drive_vessel-*.parquet
- config_name: eyepacs_combo_dr_aug
  data_files:
  - split: train
    path: data/public_fundus/public_eyepacs_combo_dr_aug-*.parquet
- config_name: gamma_multimodal
  data_files:
  - split: train
    path: data/public_fundus/public_gamma_multimodal-*.parquet
- config_name: idrid
  data_files:
  - split: train
    path: data/public_fundus/public_idrid-*.parquet
- config_name: messidor2_dr
  data_files:
  - split: train
    path: data/public_fundus/public_messidor2_dr-*.parquet
- config_name: amd_sd
  data_files:
  - split: train
    path: data/public_oct/public_oct_amd_sd-*.parquet
- config_name: areds2
  data_files:
  - split: train
    path: data/public_oct/public_oct_areds2-*.parquet
- config_name: aroi
  data_files:
  - split: train
    path: data/public_oct/public_oct_aroi-*.parquet
- config_name: c8
  data_files:
  - split: train
    path: data/public_oct/public_oct_c8-*.parquet
- config_name: chiu_dme_2015
  data_files:
  - split: train
    path: data/public_oct/public_oct_chiu_dme_2015-*.parquet
- config_name: glaucoma
  data_files:
  - split: train
    path: data/public_oct/public_oct_glaucoma-*.parquet
- config_name: kermany
  data_files:
  - split: train
    path: data/public_oct/public_oct_kermany-*.parquet
- config_name: neh_ut_2021
  data_files:
  - split: train
    path: data/public_oct/public_oct_neh_ut_2021-*.parquet
- config_name: nyu_poag
  data_files:
  - split: train
    path: data/public_oct/public_oct_nyu_poag-*.parquet
- config_name: octa500
  data_files:
  - split: train
    path: data/public_oct/public_oct_octa500-*.parquet
- config_name: octdl
  data_files:
  - split: train
    path: data/public_oct/public_oct_octdl-*.parquet
- config_name: octid
  data_files:
  - split: train
    path: data/public_oct/public_oct_octid-*.parquet
- config_name: oimhs
  data_files:
  - split: train
    path: data/public_oct/public_oct_oimhs-*.parquet
- config_name: olives
  data_files:
  - split: train
    path: data/public_oct/public_oct_olives-*.parquet
- config_name: retouch
  data_files:
  - split: train
    path: data/public_oct/public_oct_retouch-*.parquet
- config_name: sparsity_sdoct_2012
  data_files:
  - split: train
    path: data/public_oct/public_oct_sparsity_sdoct_2012-*.parquet
- config_name: srinivasan_2014
  data_files:
  - split: train
    path: data/public_oct/public_oct_srinivasan_2014-*.parquet
- config_name: thoct1800
  data_files:
  - split: train
    path: data/public_oct/public_oct_thoct1800-*.parquet
- config_name: uestc
  data_files:
  - split: train
    path: data/public_oct/public_oct_uestc-*.parquet
- config_name: refuge2_disc_cup
  data_files:
  - split: train
    path: data/public_fundus/public_refuge2_disc_cup-*.parquet
- config_name: private
  data_files:
  - split: train
    path: data/private_topcon/shanghai_drioct_triton-*.parquet
---

# ZJU Eye-Pretrain Dataset

> Unified multi-source ophthalmological imaging dataset for foundation model pretraining and downstream tasks.

**1.1M images** spanning **26 cohorts** with a **strict 41-column unified manifest schema**.

## Composition

| Source | Images | Modalities | Cohorts |
|---|---:|---|---|
| Private Shanghai DRI OCT Triton (SS-OCT) | 419,042 | oct_bscan + fundus_color + slo_gray | 1 |
| Public Fundus | 198,629 | fundus_color (+ GAMMA OCT) | 6 |
| Public OCT | 488,705 | oct_bscan | 19 |
| **Total** | **1,106,376** | | **26** |

See [DATASET_OVERVIEW.md](DATASET_OVERVIEW.md) for full details per cohort (devices, regions, masks, demographics).

## Quick Start

```python
from datasets import load_dataset, concatenate_datasets, Image

# === Load by batch ===
ds_priv = load_dataset("mayberichard/zju-eye-pretrain", "private_topcon")
ds_fun  = load_dataset("mayberichard/zju-eye-pretrain", "public_fundus")
ds_oct  = load_dataset("mayberichard/zju-eye-pretrain", "public_oct")

# === Load by single cohort (28 available, see configs in YAML above) ===
ds = load_dataset("mayberichard/zju-eye-pretrain", "kermany")    # 109k
ds = load_dataset("mayberichard/zju-eye-pretrain", "octa500")    # 120k

# === IMPORTANT: cast binary columns to Image for auto-decode ===
ds = ds.cast_column("image", Image())
# For cohorts with masks (DRIVE/IDRiD/REFUGE2/AROI/OIMHS/AMD-SD/Chiu/Glaucoma/OCTA500/RETOUCH):
for col in ds["train"].features:
    if col.endswith("_mask") and str(ds["train"].features[col]) == "Value('binary')":
        ds = ds.cast_column(col, Image())

# Each row after cast:
#   image: PIL.Image                                  (auto-decoded)
#   {vessel|fov|layer|lesion|disc_cup|...}_mask: PIL.Image or None
#   image_id, study_id, patient_hash, modality, anatomy, severity, diagnosis_group, ...

# === Concat 3 batches manually if needed ===
# Note: schemas differ across batches (mask column sets), so use only shared cols:
shared = ["image_id", "cohort", "study_id", "patient_hash", "modality",
          "anatomy", "device_vendor", "device_model", "severity",
          "diagnosis_group", "image", "bscan_index"]
all_ds = concatenate_datasets([
    ds_priv["train"].select_columns(shared).cast_column("image", Image()),
    ds_fun["train"].select_columns(shared).cast_column("image", Image()),
    ds_oct["train"].select_columns(shared).cast_column("image", Image()),
])
# 1.1M images total, mask cols dropped (use per-batch load if you need masks)

# === Streaming for big training runs (avoids downloading all 287 GB) ===
ds = load_dataset("mayberichard/zju-eye-pretrain", "public_oct", streaming=True)
ds = ds.cast_column("image", Image())
for row in ds["train"]:
    img = row["image"]   # PIL Image, lazy-decoded
    ...
```

## Schema (41-column manifest, identical across all batches)

```
cohort, study_id, patient_hash, visit_date, eye,
device_vendor, device_model, device_serial_hash, device_software_version,
hospital_domain, ethnicity,
image_quality_score, image_quality_band,
diagnosis_group, lesion_tags, lesion_location, layer_involvement, severity,
diagnosis_source, label_confidence, schema_version,
image_id, file_path, file_format,
modality, anatomy, device_technology, scan_protocol,
scan_x_mm, bscan_index,
image_height_px, image_width_px, axial_resolution_um,
has_segmentation, n_layers_visible,
fovea_x_norm, crt_um, choroid_thickness_um,
oct_footprint_bbox_fundus, oct_footprint_bbox_slo,
is_valid
```

Plus per-image `image` bytes and per-cohort mask columns.

## Captions

Each image has 5 captions (4 L1 variants + 1 L3 derived). Total 5.5M captions in `captions/`.

```python
from datasets import load_dataset
caps = load_dataset("mayberichard/zju-eye-pretrain", "captions_oct")
# join on image_id with the images config
```

## Licensing

This dataset aggregates multiple sources with mixed licenses. See [LICENSE](LICENSE) for per-cohort license terms. Users are responsible for compliance with the original license of each cohort.

**Private Shanghai Topcon data is included for research convenience.** Commercial use is prohibited.

## Citation

If you use this dataset, please cite the original source for each cohort used (see DATASET_OVERVIEW.md).

## Versioning & Updates

This dataset supports incremental updates. New cohorts can be added without touching existing data via additional shards in `data/<batch>/`. Schema migrations preserve old `*_v1.parquet` alongside new versions.