Automatic detection of age-related macular degeneration pathologies in retinal fundus images

FundusQ-Net: A regression quality assessment deep learning algorithm for fundus images quality grading

OBJECTIVE

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.

METHODS

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).

RESULTS

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%.

SIGNIFICANCE

the proposed algorithm provides a new robust tool for automated quality grading of fundus images.

GLA-Net: A global-local attention network for automatic cataract classification

Cataracts are the most crucial cause of blindness among all ophthalmic diseases. Convenient and cost-effective early cataract screening is urgently needed to reduce the risks of visual loss. To date, many studies have investigated automatic cataract classification based on fundus images. However, existing methods mainly rely on global image information while ignoring various local and subtle features. Notably, these local features are highly helpful for the identification of cataracts with different severities. To avoid this disadvantage, we introduce a deep learning technique to learn multilevel feature representations of the fundus image simultaneously. Specifically, a global-local attention network (GLA-Net) is proposed to handle the cataract classification task, which consists of two levels of subnets: the global-level attention subnet pays attention to the global structure information of the fundus image, while the local-level attention subnet focuses on the local discriminative features of the specific regions. These two types of subnets extract retinal features at different attention levels, which are then combined for final cataract classification. Our GLA-Net achieves the best performance in all metrics (90.65% detection accuracy, 83.47% grading accuracy, and 81.11% classification accuracy of grades 1 and 2). The experimental results on a real clinical dataset show that the combination of global-level and local-level attention models is effective for cataract screening and provides significant potential for other medical tasks.

Automatic glaucoma detection based on transfer induced attention network

BACKGROUND

Glaucoma is one of the causes that leads to irreversible vision loss. Automatic glaucoma detection based on fundus images has been widely studied in recent years. However, existing methods mainly depend on a considerable amount of labeled data to train the model, which is a serious constraint for real-world glaucoma detection.

METHODS

In this paper, we introduce a transfer learning technique that leverages the fundus feature learned from similar ophthalmic data to facilitate diagnosing glaucoma. Specifically, a Transfer Induced Attention Network (TIA-Net) for automatic glaucoma detection is proposed, which extracts the discriminative features that fully characterize the glaucoma-related deep patterns under limited supervision. By integrating the channel-wise attention and maximum mean discrepancy, our proposed method can achieve a smooth transition between general and specific features, thus enhancing the feature transferability.

RESULTS

To delimit the boundary between general and specific features precisely, we first investigate how many layers should be transferred during training with the source dataset network. Next, we compare our proposed model to previously mentioned methods and analyze their performance. Finally, with the advantages of the model design, we provide a transparent and interpretable transferring visualization by highlighting the key specific features in each fundus image. We evaluate the effectiveness of TIA-Net on two real clinical datasets and achieve an accuracy of 85.7%/76.6%, sensitivity of 84.9%/75.3%, specificity of 86.9%/77.2%, and AUC of 0.929 and 0.835, far better than other state-of-the-art methods.

CONCLUSION

Different from previous studies applied classic CNN models to transfer features from the non-medical dataset, we leverage knowledge from the similar ophthalmic dataset and propose an attention-based deep transfer learning model for the glaucoma diagnosis task. Extensive experiments on two real clinical datasets show that our TIA-Net outperforms other state-of-the-art methods, and meanwhile, it has certain medical value and significance for the early diagnosis of other medical tasks.

Structure-Preserving Guided Retinal Image Filtering and Its Application for Optic Disk Analysis

Retinal fundus photographs have been used in the diagnosis of many ocular diseases such as glaucoma, pathological myopia, age-related macular degeneration, and diabetic retinopathy. With the development of computer science, computer aided diagnosis has been developed to process and analyze the retinal images automatically. One of the challenges in the analysis is that the quality of the retinal image is often degraded. For example, a cataract in human lens will attenuate the retinal image, just as a cloudy camera lens which reduces the quality of a photograph. It often obscures the details in the retinal images and posts challenges in retinal image processing and analyzing tasks. In this paper, we approximate the degradation of the retinal images as a combination of human-lens attenuation and scattering. A novel structure-preserving guided retinal image filtering (SGRIF) is then proposed to restore images based on the attenuation and scattering model. The proposed SGRIF consists of a step of global structure transferring and a step of global edge-preserving smoothing. Our results show that the proposed SGRIF method is able to improve the contrast of retinal images, measured by histogram flatness measure, histogram spread, and variability of local luminosity. In addition, we further explored the benefits of SGRIF for subsequent retinal image processing and analyzing tasks. In the two applications of deep learning-based optic cup segmentation and sparse learning-based cup-to-disk ratio (CDR) computation, our results show that we are able to achieve more accurate optic cup segmentation and CDR measurements from images processed by SGRIF.

3D Curvelet-Based Segmentation and Quantification of Drusen in Optical Coherence Tomography Images

Spectral-Domain Optical Coherence Tomography (SD-OCT) is a widely used interferometric diagnostic technique in ophthalmology that provides novel in vivo information of depth-resolved inner and outer retinal structures. This imaging modality can assist clinicians in monitoring the progression of Age-related Macular Degeneration (AMD) by providing high-resolution visualization of drusen. Quantitative tools for assessing drusen volume that are indicative of AMD progression may lead to appropriate metrics for selecting treatment protocols. To address this need, a fully automated algorithm was developed to segment drusen area and volume from SD-OCT images. The proposed algorithm consists of three parts: (1) preprocessing, which includes creating binary mask and removing possible highly reflective posterior hyaloid that is used in accurate detection of inner segment/outer segment (IS/OS) junction layer and Bruch’s membrane (BM) retinal layers; (2) coarse segmentation, in which 3D curvelet transform and graph theory are employed to get the possible candidate drusenoid regions; (3) fine segmentation, in which morphological operators are used to remove falsely extracted elongated structures and get the refined segmentation results. The proposed method was evaluated in 20 publically available volumetric scans acquired by using Bioptigen spectral-domain ophthalmic imaging system. The average true positive and false positive volume fractions (TPVF and FPVF) for the segmentation of drusenoid regions were found to be 89.15% ± 3.76 and 0.17% ± .18%, respectively.

Exploiting ensemble learning for automatic cataract detection and grading

Cataract is defined as a lenticular opacity presenting usually with poor visual acuity. It is one of the most common causes of visual impairment worldwide. Early diagnosis demands the expertise of trained healthcare professionals, which may present a barrier to early intervention due to underlying costs. To date, studies reported in the literature utilize a single learning model for retinal image classification in grading cataract severity. We present an ensemble learning based approach as a means to improving diagnostic accuracy. Three independent feature sets, i.e., wavelet-, sketch-, and texture-based features, are extracted from each fundus image. For each feature set, two base learning models, i.e., Support Vector Machine and Back Propagation Neural Network, are built. Then, the ensemble methods, majority voting and stacking, are investigated to combine the multiple base learning models for final fundus image classification. Empirical experiments are conducted for cataract detection (two-class task, i.e., cataract or non-cataractous) and cataract grading (four-class task, i.e., non-cataractous, mild, moderate or severe) tasks. The best performance of the ensemble classifier is 93.2% and 84.5% in terms of the correct classification rates for cataract detection and grading tasks, respectively. The results demonstrate that the ensemble classifier outperforms the single learning model significantly, which also illustrates the effectiveness of the proposed approach.