docs: update dataset inventory and benchmark evaluation tasks with complete cohort metadata
Browse files
README.md
CHANGED
|
@@ -105,37 +105,39 @@ The dataset captures full optical, acoustic, and angiography representations acr
|
|
| 105 |
|
| 106 |
## 4. Archive Directory & Shard Inventory
|
| 107 |
|
| 108 |
-
Archives are organized into self-contained sub-directories chunked at ~1024 MiB boundaries. Each shard is compressed using `zstd -
|
| 109 |
|
| 110 |
| Dataset Directory | Modality Focus | Shards | Total Size | Primary Source Center | Key Files Included |
|
| 111 |
| :--- | :--- | :--- | :--- | :--- | :--- |
|
| 112 |
-
|
|
| 113 |
-
|
|
| 114 |
-
| | Longitudinal Retinal Cohort | 2 | 133.0 GB | Beijing Tongren Hospital | Raw longitudinal image archives (, ) |
|
| 115 |
-
|
|
| 116 |
-
|
|
| 117 |
-
|
|
| 118 |
-
|
|
| 119 |
-
|
|
| 120 |
-
|
|
| 121 |
-
|
|
| 122 |
-
|
|
| 123 |
-
|
|
| 124 |
-
|
|
| 125 |
-
|
|
| 126 |
-
|
|
| 127 |
-
|
|
| 128 |
-
|
|
| 129 |
-
|
|
| 130 |
-
|
|
| 131 |
-
|
|
| 132 |
-
|
|
| 133 |
-
|
|
| 134 |
-
|
|
| 135 |
-
|
|
| 136 |
-
|
|
| 137 |
-
|
|
| 138 |
-
|
|
|
|
|
|
|
|
| 139 |
|
| 140 |
## 5. Master Metadata (`metadata/`) & Clinical Crosswalks
|
| 141 |
|
|
@@ -167,23 +169,25 @@ Complete breakdown of 1,817,159 instruction-tuned samples and 1,459,879 downstre
|
|
| 167 |
|
| 168 |
## 6. Benchmark Evaluation Tasks
|
| 169 |
|
| 170 |
-
1. **17-Class Retinal Disease Classification ()**:
|
| 171 |
Standardized 17-class classification for common and sight-threatening retinal conditions (Normal, DR, AMD, RVO, Pathological Myopia, Macular Hole, Epiretinal Membrane, Retinal Detachment, Retinitis Pigmentosa, etc.).
|
| 172 |
-
2. **66-Class Multi-Center Fine-Grained Diagnosis ()**:
|
| 173 |
Challenging 66-class classification covering rare and subtle ophthalmic pathologies validated across Beijing Tongren and Handan populations.
|
| 174 |
-
3. **95-Class Clinical Semantic Classification (, )**:
|
| 175 |
Full-spectrum hospital diagnostic classification aligning free-text clinical reports and fundus photographs across 16 major disease families and 95 clinical subcategories.
|
| 176 |
-
4. **Dynamic FFA Vessel & Microvascular Leakage Segmentation (, )**:
|
| 177 |
Temporal sequence segmentation across early, arteriovenous, and late phases.
|
| 178 |
5. **Quantitative Anterior Chamber Morphometry & Angle-Closure Risk (UBM)**:
|
| 179 |
-
Automated measurement of Anterior Chamber Depth (ACD in mm), Trabecular-Iris Angle (TIA), and angle-closure risk assessment across , , and .
|
| 180 |
-
6. **Ophthalmic Image Quality & Pre-Filtering Assessment ()**:
|
| 181 |
Automated detection of non-diagnostic fundus images, lens opacity, uneven illumination, and motion artifacts for reliable clinical pipeline pre-filtering.
|
| 182 |
-
7. **Dual-View Diabetic Retinopathy Grading ()**:
|
| 183 |
Multi-field consistency and lesion grading combining macular and optic disc perspectives.
|
| 184 |
-
8. **Multi-Center Glaucoma Screening & Cup-to-Disc Ratio Estimation (, )**:
|
| 185 |
Optic nerve head segmentation, vertical cup-to-disc ratio (vCDR) calculation, and multi-device glaucoma detection.
|
| 186 |
|
|
|
|
|
|
|
| 187 |
## 7. Quick Start: Extraction & Usage Guide
|
| 188 |
|
| 189 |
### A. Download & Extract Archives (Linux / macOS)
|
|
|
|
| 105 |
|
| 106 |
## 4. Archive Directory & Shard Inventory
|
| 107 |
|
| 108 |
+
Archives are organized into self-contained sub-directories chunked at ~1024 MiB boundaries. Each shard is compressed using `zstd -3` for fast decompressibility and high compression efficiency:
|
| 109 |
|
| 110 |
| Dataset Directory | Modality Focus | Shards | Total Size | Primary Source Center | Key Files Included |
|
| 111 |
| :--- | :--- | :--- | :--- | :--- | :--- |
|
| 112 |
+
| `archives/yinhai_uwf/` | Ultra-Widefield (UWF) | 55 | 274.6 GB | Chengdu Yinhai Eye Hospital | 200° Optos color scans (55 tar.zst shards) |
|
| 113 |
+
| `archives/data0410/` + `data0410_shard7/` | CFP, OCT Buckets, Comorbidity | 21 | 287.4 GB | Multi-center clinical | `fundus_hcw.zip`, `octBuckets.zip` (15 split parts), diagnosis Excel |
|
| 114 |
+
| `archives/tongren_archives/`| Longitudinal Retinal Cohort | 2 | 133.0 GB | Beijing Tongren Hospital | Raw longitudinal image archives (`external.zip`, `fundus.zip`) |
|
| 115 |
+
| `archives/zhanghuan_selected_public/` | Public Benchmark Suites | 109 | 106.8 GB | International public sets | 109 standardized shards (Messidor, ODIR, DDR, CHAKSU, etc.) |
|
| 116 |
+
| `archives/Final_eryuan_FFA/` | Dynamic FFA/ICGA Sequences | 19 | 76.4 GB | ZJU-2 Eye Center | Full dynamic FFA frames + `output2.jsonl`, `visionoph4.jsonl` |
|
| 117 |
+
| `archives/eryuanFundusT17RDCls/` | 17-Class Retinal Disease CFP | 74 | 47.4 GB | ZJU-2 Eye Center | 17-category labeled fundus images for benchmark training |
|
| 118 |
+
| `archives/ChengduShiyiUBM/` | 50MHz Anterior UBM | 255 | 39.3 GB | Chengdu 1st People's Hospital | 50MHz ultrasound sweeps + `ChengduShiyiUBM.json` (ACD mm) |
|
| 119 |
+
| `archives/OCTA-500/` | 3D OCTA + Projection Slabs | 44 | 37.0 GB | NJUST / Public benchmark | 6mm & 3mm OCTA volumes with vascular masks (<=1GB shards) |
|
| 120 |
+
| `archives/tongren16G95DFundus/` | 16-Class 95-Subtype Gold-Standard | 36 | 35.2 GB | Beijing Tongren Hospital | 16 primary diagnostic families & 95 clinical subtypes aligned with `metadata/tongren95_taxonomy.csv` |
|
| 121 |
+
| `archives/OCTA500_OCT/` | Structural OCT Volumes | 29 | 28.6 GB | NJUST / Public benchmark | Structural OCT volumes paired with OCTA-500 (29 split parts) |
|
| 122 |
+
| `archives/WLOA/` | Wide-Field OCT Angiography | 18 | 17.0 GB | Benchmark suite | Wide-field OCTA scans (18 split parts) |
|
| 123 |
+
| `archives/DDR/` | Diabetic Retinopathy Lesion Seg | 10 | 15.0 GB | DDR Benchmark | 10-part partitioned DDR images and lesion segmentation masks |
|
| 124 |
+
| `archives/FQ_Datasets/` | Fundus Quality & Artifact Grading | 14 | 13.2 GB | Multi-center QA team | Human-curated `validFundus` (12.6GB) and `invalidFundus` (0.6GB) |
|
| 125 |
+
| `archives/BUltrasound/` | Ocular B-Scan Ultrasound | 38 | 11.3 GB | Xiangya-2 & Yinhai | B-scan ultrasounds + `XiangyaBscan.json`, `YinhaiBscan.json` |
|
| 126 |
+
| `archives/kaggle_glaucoma_v4/` | Multi-Source Glaucoma Suite v4 | 10 | 9.4 GB | Multi-source benchmark | 21,566 fundus images covering glaucoma screening and CDR metrics |
|
| 127 |
+
| `archives/YinhaiUBM/` | 50MHz UBM + Clinical Diagnostics | 10 | 9.2 GB | Chengdu Yinhai Eye Hospital | 83,752 UBM frames + comprehensive clinical diagnostic JSON annotations |
|
| 128 |
+
| `archives/trhd3_FRD_Cls66_Train/` | 66-Class Fine-Grained CFP | 8 | 7.8 GB | Tongren & Handan Eye Hospital | 66-disease multiclass training set with splits |
|
| 129 |
+
| `archives/UBM_SW3200L/` | Noise-Reduced 50MHz UBM Suite | 8 | 7.7 GB | Multi-center UBM | Denoised UBM ciliary body scans (`SW3200L` transducer suites) |
|
| 130 |
+
| `archives/RFMiD1/` | Multi-Disease Identifier 1 | 8 | 7.5 GB | RFMiD benchmark suite | Multi-label 46-disease fundus images (Training, Evaluation, Test) |
|
| 131 |
+
| `archives/AIChallenger/` | Retinal Disease Classification | 5 | 4.5 GB | AI Challenger benchmark | Full partitioned retinal challenge images |
|
| 132 |
+
| `archives/handan_ffa/` | FFA Dynamic Series | 5 | 4.5 GB | Handan Eye Hospital | Handan clinical FFA angiograms |
|
| 133 |
+
| `archives/handan_ffa2/` | FFA Dynamic Series 2 | 3 | 2.7 GB | Handan Eye Hospital | Longitudinal follow-up FFA |
|
| 134 |
+
| `archives/deepdrid/` | Dual-View DR Grading Benchmark | 2 | 2.0 GB | DeepDRiD benchmark suite | 2,058 paired macula-centered and disc-centered fundus images |
|
| 135 |
+
| `archives/ODIR/` | Ocular Disease Intelligent 5K | 2 | 1.3 GB | Shanggong / Benchmark | Multi-disease paired left/right fundus images (`ODIR-5K.zip`) |
|
| 136 |
+
| `archives/chengdu_ffa/` | FFA Dynamic Series | 2 | 1.2 GB | Chengdu Yinhai Eye Hospital | Chengdu clinical FFA angiograms |
|
| 137 |
+
| `archives/DGOCF/` | Optic Disc & Glaucoma CFP | 1 | 0.8 GB | Benchmark suite | Optic nerve head center annotations |
|
| 138 |
+
| `archives/yinhai_ffa/` | FFA Dynamic Series | 2 | 0.5 GB | Chengdu Yinhai Eye Hospital | High-resolution FFA sequences |
|
| 139 |
+
|
| 140 |
+
---
|
| 141 |
|
| 142 |
## 5. Master Metadata (`metadata/`) & Clinical Crosswalks
|
| 143 |
|
|
|
|
| 169 |
|
| 170 |
## 6. Benchmark Evaluation Tasks
|
| 171 |
|
| 172 |
+
1. **17-Class Retinal Disease Classification (`eryuanFundusT17RDCls`)**:
|
| 173 |
Standardized 17-class classification for common and sight-threatening retinal conditions (Normal, DR, AMD, RVO, Pathological Myopia, Macular Hole, Epiretinal Membrane, Retinal Detachment, Retinitis Pigmentosa, etc.).
|
| 174 |
+
2. **66-Class Multi-Center Fine-Grained Diagnosis (`trhd3_FRD_Cls66_Train`)**:
|
| 175 |
Challenging 66-class classification covering rare and subtle ophthalmic pathologies validated across Beijing Tongren and Handan populations.
|
| 176 |
+
3. **95-Class Clinical Semantic Classification (`tongren16G95DFundus`, `tongren95_taxonomy`)**:
|
| 177 |
Full-spectrum hospital diagnostic classification aligning free-text clinical reports and fundus photographs across 16 major disease families and 95 clinical subcategories.
|
| 178 |
+
4. **Dynamic FFA Vessel & Microvascular Leakage Segmentation (`Final_eryuan_FFA`, `handan_ffa`)**:
|
| 179 |
Temporal sequence segmentation across early, arteriovenous, and late phases.
|
| 180 |
5. **Quantitative Anterior Chamber Morphometry & Angle-Closure Risk (UBM)**:
|
| 181 |
+
Automated measurement of Anterior Chamber Depth (ACD in mm), Trabecular-Iris Angle (TIA), and angle-closure risk assessment across `ChengduShiyiUBM`, `YinhaiUBM`, and `UBM_SW3200L`.
|
| 182 |
+
6. **Ophthalmic Image Quality & Pre-Filtering Assessment (`FQ_Datasets`)**:
|
| 183 |
Automated detection of non-diagnostic fundus images, lens opacity, uneven illumination, and motion artifacts for reliable clinical pipeline pre-filtering.
|
| 184 |
+
7. **Dual-View Diabetic Retinopathy Grading (`deepdrid`)**:
|
| 185 |
Multi-field consistency and lesion grading combining macular and optic disc perspectives.
|
| 186 |
+
8. **Multi-Center Glaucoma Screening & Cup-to-Disc Ratio Estimation (`kaggle_glaucoma_v4`, `DGOCF`)**:
|
| 187 |
Optic nerve head segmentation, vertical cup-to-disc ratio (vCDR) calculation, and multi-device glaucoma detection.
|
| 188 |
|
| 189 |
+
---
|
| 190 |
+
|
| 191 |
## 7. Quick Start: Extraction & Usage Guide
|
| 192 |
|
| 193 |
### A. Download & Extract Archives (Linux / macOS)
|