The Faculty of Biomedical Engineering, Technion-IIT, Haifa, Israel.
Rambam Medical Center: Rambam Health Care Campus, Israel.
Comput Methods Programs Biomed. 2023 Sep;239:107522. doi: 10.1016/j.cmpb.2023.107522. Epub 2023 May 26.
Ophthalmological pathologies such as glaucoma, diabetic retinopathy and age-related macular degeneration are major causes of blindness and vision impairment. There is a need for novel decision support tools that can simplify and speed up the diagnosis of these pathologies. A key step in this process is to automatically estimate the quality of the fundus images to make sure these are interpretable by a human operator or a machine learning model. We present a novel fundus image quality scale and deep learning (DL) model that can estimate fundus image quality relative to this new scale.
A total of 1245 images were graded for quality by two ophthalmologists within the range 1-10, with a resolution of 0.5. A DL regression model was trained for fundus image quality assessment. The architecture used was Inception-V3. The model was developed using a total of 89,947 images from 6 databases, of which 1245 were labeled by the specialists and the remaining 88,702 images were used for pre-training and semi-supervised learning. The final DL model was evaluated on an internal test set (n=209) as well as an external test set (n=194).
The final DL model, denoted FundusQ-Net, achieved a mean absolute error of 0.61 (0.54-0.68) on the internal test set. When evaluated as a binary classification model on the public DRIMDB database as an external test set the model obtained an accuracy of 99%.
the proposed algorithm provides a new robust tool for automated quality grading of fundus images.
青光眼、糖尿病视网膜病变和年龄相关性黄斑变性等眼科病理学是导致失明和视力损害的主要原因。因此，需要新型的决策支持工具来简化和加快这些病变的诊断。该过程的关键步骤是自动评估眼底图像的质量，以确保这些图像可由人工操作员或机器学习模型进行解释。我们提出了一种新的眼底图像质量量表和深度学习（DL）模型，可根据该新量表相对评估眼底图像质量。
两名眼科医生在 1-10 的范围内对总共 1245 张图像的质量进行分级，分辨率为 0.5。针对眼底图像质量评估，我们训练了一个 DL 回归模型。所使用的架构为 Inception-V3。该模型是使用来自 6 个数据库的总共 89947 张图像开发的，其中 1245 张图像由专家标记，其余 88702 张图像用于预训练和半监督学习。最终的 DL 模型在内部测试集（n=209）和外部测试集（n=194）上进行了评估。
最终的 DL 模型，称为 FundusQ-Net，在内部测试集上的平均绝对误差为 0.61（0.54-0.68）。当作为公共 DRIMDB 数据库上的二进制分类模型进行评估作为外部测试集时，该模型的准确率为 99%。
该算法为眼底图像的自动质量分级提供了一种新的强大工具。
