Table 1 Summarization of publicly available fundus image quality assessment datasets.
From: MSHF: A Multi-Source Heterogeneous Fundus (MSHF) Dataset for Image Quality Assessment
Dataset | Year | Number | Quality label | Disease | Annotators |
|---|---|---|---|---|---|
DRIMDB | 2014 | 216 CFP images | ‘good’, ‘bad’ or ‘outlier’ | DR, non-DR | NA |
Kaggle DR Image Quality | 2018 | 88702 CFP images | ‘good’ or ‘poor’ for overall quality | DR, non-DR | NA |
EyeQ Assessment | 2019 | 28792 CFP images | ‘good’, ‘usable’ or ‘reject’ for overall quality | DR, non-DR | 2 |
DeepDRiD | 2022 | 2000 CFP images and 256 UWF images | score criteria 0~10 for artifact, clarity, field definition and 0/1 for overall quality | DR, non-DR | 3 |
MSHF (proposed) | 2022 | 500 CFP images, 500 UWF images and 302 portable camera images | 0/1 for illumination, clarity, contrast and overall quality | DR, healthy, glaucoma | 3 |
