xini666666/OLIVES_Dataset

---
license: mit
size_categories:
- 10K<n<100K
pretty_name: 'OLIVES Dataset: Ophthalmic Labels for Investigating Visual Eye Semantics'
tags:
- medical
dataset_info:
- config_name: biomarker_detection
features:
- name: Image
dtype: image
- name: B1
dtype: float64
- name: B2
dtype: float64
- name: B3
dtype: float64
- name: B4
dtype: float64
- name: B5
dtype: float64
- name: B6
dtype: float64
- name: BCVA
dtype: float64
- name: CST
dtype: float64
- name: Eye_ID
dtype: float64
- name: Patient_ID
dtype: float64
splits:
- name: train
num_bytes: 15852565958.136
num_examples: 78822
- name: test
num_bytes: 968486633.741
num_examples: 3871
download_size: 15923453393
dataset_size: 16821052591.876999
- config_name: disease_classification
features:
- name: Image
dtype: image
- name: Scan (n/49)
dtype: float64
- name: Atrophy / thinning of retinal layers
dtype: float64
- name: Disruption of EZ
dtype: float64
- name: DRIL
dtype: float64
- name: IR hemorrhages
dtype: float64
- name: IR HRF
dtype: float64
- name: Partially attached vitreous face
dtype: float64
- name: Fully attached vitreous face
dtype: float64
- name: Preretinal tissue/hemorrhage
dtype: float64
- name: Vitreous debris
dtype: float64
- name: VMT
dtype: float64
- name: DRT/ME
dtype: float64
- name: Fluid (IRF)
dtype: float64
- name: Fluid (SRF)
dtype: float64
- name: Disruption of RPE
dtype: float64
- name: PED (serous)
dtype: float64
- name: SHRM
dtype: float64
- name: Eye_ID
dtype: float64
- name: BCVA
dtype: float64
- name: CST
dtype: float64
- name: Patient_ID
dtype: int64
- name: Disease Label
dtype: float64
splits:
- name: train
num_bytes: 15860241253.136
num_examples: 78822
download_size: 15061835755
dataset_size: 15860241253.136
configs:
- config_name: biomarker_detection
data_files:
- split: train
path: biomarker_detection/train-*
- split: test
path: biomarker_detection/test-*
- config_name: disease_classification
data_files:
- split: train
path: disease_classification/train-*
---

# OLIVES_Dataset

## Abstract
Clinical diagnosis of the eye is performed over multifarious data modalities including scalar clinical labels, vectorized biomarkers, two-dimensional fundus images, and three-dimensional Optical Coherence Tomography (OCT) scans. While the clinical labels, fundus images and OCT scans are instrumental measurements, the vectorized biomarkers are interpreted attributes from the other measurements. Clinical practitioners use all these data modalities for diagnosing and treating eye diseases like Diabetic Retinopathy (DR) or Diabetic Macular Edema (DME). Enabling usage of machine learning algorithms within the ophthalmic medical domain requires research into the relationships and interactions between these relevant data modalities. Existing datasets are limited in that: ($i$) they view the problem as disease prediction without assessing biomarkers, and ($ii$) they do not consider the explicit relationship among all four data modalities over the treatment period. In this paper, we introduce the Ophthalmic Labels for Investigating Visual Eye Semantics (OLIVES) dataset that addresses the above limitations. This is the first OCT and fundus dataset that includes clinical labels, biomarker labels, and time-series patient treatment information from associated clinical trials. The dataset consists of $1268$ fundus eye images each with $49$ OCT scans, and $16$ biomarkers, along with $3$ clinical labels and a disease diagnosis of DR or DME. In total, there are 96 eyes' data averaged over a period of at least two years with each eye treated for an average of 66 weeks and 7 injections. OLIVES dataset has advantages in other fields of machine learning research including self-supervised learning as it provides alternate augmentation schemes that are medically grounded.

## Subsets
There are 2 subsets included in this dataset:
- Disease Classification (`disease_classification`)
- Biomarker Detection ('biomarker_detection')

Disease classification provides the full dataset while the biomarker detection subset provides a curated train-test split.

### Disease Classification
This subset contains information regarding 78,000+ OCT scans obtained from a series of visits patients performed. In terms of labels, there are:

- `Image`: An image of the OCT scan
- `BCVA`: Best Central Visual Acuity
- `CST`: Central Subfield Thickness
- `Patient ID`: A value to help distinguish different patients
- `Disease Label`: A value of `0` for DR (Diabetic Retinopathy) and `1` for DME (Diabetic Macular Edema)

This information can be used to classify the disease. In addition, the first and last visit of a patient included extra biomarker information. This can be summarized into these 16 mostly-boolean labels:

- `Scan (n/49)`: The scan number out of the 49 scans taken for each patient
- `Atrophy / thinning of retinal layer`
- `Disruption of EZ`: Disruption of Ellipsoid Zone
- `DRIL`: Disruption of Retinal Inner Layers
- `IR hemorrhages`: Intraretinal hemorrhages
- `IR HRF`: Intraretinal Hyperreflective Foci
- `Partially attached vitreous face`
- `Fully attached vitreous face`
- `Preretinal tissue/hemorrhage`
- `Vitreous debris`
- `VMT`: Vitreomacular Traction
- `DRT/ME`: Diffuse Retinal Thickening or Macular Edema
- `Fluid (IRF)`: Intraretinal Fluid
- `Fluid (SRF)`: Subretinal Fluid
- `Disruption of RPE`: Disruption of Retinal Pigment Epithelium
- `PED (serous)`: Pigment Epithelial Detachment
- `SHRM`: Subretinal Hyperreflective Material
- `Eye_ID`: A value to help distinguish different eye scans

### Biomarker Detection
This subset was used to host the 2023 Video and Image Processing Cup (VIP) challenge hosted by IEEE Signal Processing Society. The goal is to detect 6 biomarkers given image and clinical label data. For additional information reagarding the challenge, the metrics, the train-test data splits, please visit: https://alregib.ece.gatech.edu/competitions/2023-vip-cup/
The 6 biomarkers and their associated interpretations:

- **B1 (Intraretinal Hyperreflective Foci (IRHRF)):** were indicated as present with the appearance of intraretinal, highly reflective spots, which correspond pathologically to microaneurysms or hard exudates, with or without shadowing of the more posterior retinal layers.

- **B2 (A Partially Attached Vitreous Face (PAVF)):** was indicated as present with evidence of perifoveal detachment of the vitreous from the internal limiting membrane (ILM) with a macular attachment point within a 3-mm radius of the fovea.

- **B3 (A Fully Attached Vitreous Face (FAVF)):** was indicated as present with no evidence of perifoveal or macular detachment from the ILM.

- **B4 (Intraretinal Fluid (IRF)):** was indicated as present when intraretinal hyporeflective areas or cysts had a minimum fluid height of 20 µm

- **B5 (Diffuse Retinal Thickening or Diabetic Macular Edema (DRT/ME)):** was indicated as present when there was increased retinal thickness of 50 µm above the otherwise flat retina surface with associated reduced reflectivity in the intraretinal tissues

- **B6 (Vitreous Debris (VD)):** was indicated as present with evidence of hyperreflective foci in the vitreous or shadowing of the retinal layers in the absence of an intraretinal hemorrhage.

The 6 biomarkers are chosen to have balanced train-test splits.

## Data Download
Sample code to download the disease classification dataset:

```python
from datasets import load_dataset
from torch.utils.data import DataLoader

olives = load_dataset('gOLIVES/OLIVES_Dataset', 'disease_classification', split = 'train')

# Covert into a Format Usable by Pytorch
olives = olives.with_format("torch")

dataloader = DataLoader(olives, batch_size=4)
for batch in dataloader:
print(batch)

# Example to get the VMT Biomarker of the first image in the dataset.
print(olives[0]['VMT'])

```

## Known Issues
- Patient ID #79 has missing `BCVA` and `CST` for most visits except the first and last visit as the biomarker information is present
- Certain visits for patients seem to have the exact same scans as a previous visit. For instance Patient ID #61 has identical images in W8 and their next visit in W12.

## Links

**Associated Website**: https://alregib.ece.gatech.edu/olives-dataset/

**More work from our Lab**: https://alregib.ece.gatech.edu/

**Paper**:

1. OLIVES Dataset: https://arxiv.org/pdf/2209.11195
2. VIP Cup Paper: https://arxiv.org/pdf/2408.11170

## Citations

If you find the work useful, please include the following citations in your work:

> @inproceedings{prabhushankarolives2022,\
> title={OLIVES Dataset: Ophthalmic Labels for Investigating Visual Eye Semantics},\
> author={Prabhushankar, Mohit and Kokilepersaud, Kiran and Logan, Yash-yee and Trejo Corona, Stephanie and AlRegib, Ghassan and Wykoff, Charles},\
> booktitle={Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks 2 (NeurIPS Datasets and Benchmarks 2022)},\
> year={2022}\
> }
>
> @article{alregib2024ophthalmic,
title={Ophthalmic Biomarker Detection: Highlights From the IEEE Video and Image Processing Cup 2023 Student Competition [SP Competitions]},
author={AlRegib, Ghassan and Prabhushankar, Mohit and Kokilepersaud, Kiran and Chowdhury, Prithwijit and Fowler, Zoe and Corona, Stephanie Trejo and Thomaz, Lucas A and Majumdar, Angshul},
journal={IEEE Signal Processing Magazine},
volume={41},
number={4},
pages={96--104},
year={2024},
publisher={IEEE}
}