Beng Hai Lee

ORIGA -light : An online retinal fundus image database for glaucoma analysis and research

Retinal fundus image is an important modality to document the health of the retina and is widely used to diagnose ocular diseases such as glaucoma, diabetic retinopathy and age-related macular degeneration. However, the enormous amount of retinal data obtained nowadays mostly stored locally; and the valuable embedded clinical knowledge is not efficiently exploited. In this paper we present an online depository, ORIGA-light, which aims to share clinical groundtruth retinal images with the public; provide open access for researchers to benchmark their computer-aided segmentation algorithms. An in-house image segmentation and grading tool is developed to facilitate the construction of ORIGA-light. A quantified objective benchmarking method is proposed, focusing on optic disc and cup segmentation and Cup-to-Disc Ratio (CDR). Currently, ORIGA-light contains 650 retinal images annotated by trained professionals from Singapore Eye Research Institute. A wide collection of image signs, critical for glaucoma diagnosis, are annotated. We will update the system continuously with more clinical ground-truth images. ORIGA-light is available for online access upon request.

Optic disc region of interest localization in fundus image for Glaucoma detection in ARGALI

Glaucoma is the second leading cause of permanent blindness worldwide. Glaucoma can be diagnosed through measurement of neuro-retinal optic cup-to-disc ratio (CDR). Correctly determining the optic disc region of interest (ROI) will produce a smaller initial image which takes much lesser time taken to process compared to the entire image. The earlier ROI localization in the ARGALI system used a grid based method. The new algorithm adds a preprocessing step before analyzing the image. This step significantly improves the performance of the ROI detection. A batch of 1564 retinal images from the Singapore Eye Research Centre was used to compare the performance of the two methods. From the results, the earlier and new algorithm detects the ROI correctly for 88% and 96% of the images respectively. The results indicate potential applicability of the method for automated and objective mass screening for early detection of glaucoma.

Focal biologically inspired feature for glaucoma type classification

Glaucoma is an optic nerve disease resulting in loss of vision. There are two common types of glaucoma: open angle glaucoma and angle closure glaucoma. Glaucoma type classification is important in glaucoma diagnosis. Ophthalmologists examine the iridocorneal angle between iris and cornea to determine the glaucoma type. However, manual classification/grading of the iridocorneal angle images is subjective and time consuming. To save workload and facilitate large-scale clinical use, it is essential to determine glaucoma type automatically. In this paper, we propose to use focal biologically inspired feature for the classification. The iris surface is located to determine the focal region. The association between focal biologically inspired feature and angle grades is built. The experimental results show that the proposed method can correctly classify 85.2% images from open angle glaucoma and 84.3% images from angle closure glaucoma. The accuracy could be improved close to 90% with more images included in the training. The results show that the focal biologically inspired feature is effective for automatic glaucoma type classification. It can be used to reduce workload of ophthalmologists and diagnosis cost.

Detection of Pathological Myopia by PAMELA with Texture-Based Features through an SVM Approach

Pathological myopia is the seventh leading cause of blindness worldwide. Current methods for the detection of pathological myopia are manual and subjective. We have developed a system known as PAMELA (Pathological Myopia Detection Through Peripapillary Atrophy) to automatically assess a retinal fundus image for pathological myopia. This paper focuses on the texture analysis component of PAMELA which uses texture features, clinical image context and support vector machine-based classification to detect the presence of pathological myopia in a retinal fundus image. Results on a test image set from the Singapore Eye Research Institute show an accuracy of 87.5% and a sensitivity and specificity of 0.85 and 0.90 respectively. The results show good promise for PAMELA to be developed as an automatic tool for pathological myopia detection.

Superpixel classification for initialization in model based optic disc segmentation

Optic disc segmentation in retinal fundus image is important in ocular image analysis and computer aided diagnosis. Because of the presence of peripapillary atrophy which affects the deformation, it is important to have a good initialization in deformable model based optic disc segmentation. In this paper, a superpixel classification based method is proposed for the initialization. It uses histogram of superpixels from the contrast enhanced image as features. In the training, bootstrapping is adopted to handle the unbalanced cluster issue due to the presence of peripapillary atrophy. A self-assessment reliability score is computed to evaluate the quality of the initialization and the segmentation. The proposed method has been tested in a database of 650 images with optic disc boundaries marked by trained professionals manually. The experimental results show an mean overlapping error of 10.0% and standard deviation of 7.5% in the best scenario. The results also show an increase in overlapping error as the reliability score reduces, which justifies the effectiveness of the self-assessment. The method can be used for optic disc boundary initialization and segmentation in computer aided diagnosis system and the self-assessment can be used as an indicator of cases with large errors and thus enhance the usage of the automatic segmentation.

Closed angle glaucoma detection in RetCam images

Closed/Open angle glaucoma classification is important for glaucoma diagnosis. RetCam is a new imaging modality that captures the image of iridocorneal angle for the classification. However, manual grading and analysis of the RetCam image is subjective and time consuming. In this paper, we propose a system for intelligent analysis of iridocorneal angle images, which can differentiate closed angle glaucoma from open angle glaucoma automatically. Two approaches are proposed for the classification and their performances are compared. The experimental results show promising results.

Automated anterior chamber angle localization and glaucoma type classification in OCT images

To identify glaucoma type with OCT (optical coherence tomography) images, we present an image processing and machine learning based framework to localize and classify anterior chamber angle (ACA) accurately and efficiently. In digital OCT photographs, our method automatically localizes the ACA region, which is the primary structural image cue for clinically identifying glaucoma type. Next, visual features are extracted from this region to classify the angle as open angle (OA) or angle-closure (AC). This proposed method has three major contributions that differ from existing methods. First, the ACA localization from OCT images is fully automated and efficient for different ACA configurations. Second, it can directly classify ACA as OA/AC based on only visual features, which is different from previous work for ACA measurement that relies on clinical features. Third, it demonstrates that higher dimensional visual features outperform low dimensional clinical features in terms of angle closure classification accuracy. From tests on a clinical dataset comprising of 2048 images, the proposed method only requires 0.26s per image. The framework achieves a 0.921 ± 0.036 AUC (area under curve) value and 84.0% ± 5.7% balanced accuracy at a 85% specificity, which outperforms existing methods based on clinical features.

Anterior chamber angle classification using multiscale histograms of oriented gradients for glaucoma subtype identification

Glaucoma subtype can be identified according to the configuration of the anterior chamber angle(ACA). In this paper, we present an ACA classification approach based on histograms of oriented gradients at multiple scales. In digital optical coherence tomography (OCT) photographs, our method automatically localizes the ACA, and extracts histograms of oriented gradients (HOG) features from this region to classify the angle as an open angle (OA) or an angle-closure(AC). This proposed method has three major features that differs from existing methods. First, the ACA localization from OCT images is fully automated and efficient for different ACA configurations. Second, the ACA is directly classified as OA/AC by using multiscale HOG visual features only, which is different from previous ACA assessment approaches that on clinical features. Third, it demonstrates that visual features with higher dimensions outperform low dimensional clinical features in terms of angle closure classification accuracy. Testing was performed on a large clinical dataset, comprising of 2048 images. The proposed method achieves a 0.835±0.068 AUC value and 75.8% ± 6.4% balanced accuracy at a 85% specificity, which outperforms existing ACA classification approaches based on clinical features.

Automatic localization of retinal landmarks

Retinal landmark detection is a key step in retinal screening and computer-aided diagnosis for different types of eye diseases, such as glaucomma, age-related macular degeneration(AMD) and diabetic retinopathy. In this paper, we propose a semantic image transformation(SIT) approach for retinal representation and automatic landmark detection. The proposed SIT characterizes the local statistics of a fundus image and boosts the intrinsic retinal structures, such as optic disc(OD), macula. We propose our salient OD and macular models based on SIT for retinal landmark detection. Experiments on 5928 images show that our method achieves an accuracy of 99.44% in the detection of OD and an accuracy of 93.49% in the detection of macula, while having an accuracy of 97.33% for left and right eye classification. The proposed SIT can automatically detect the retinal landmarks and be useful for further eye-disease screening and diagnosis.

A low-dimensional step pattern analysis algorithm with application to multimodal retinal image registration

Existing feature descriptor-based methods on retinal image registration are mainly based on scale-invariant feature transform (SIFT) or partial intensity invariant feature descriptor (PIIFD). While these descriptors are often being exploited, they do not work very well upon unhealthy multimodal images with severe diseases. Additionally, the descriptors demand high dimensionality to adequately represent the features of interest. The higher the dimensionality, the greater the consumption of resources (e.g. memory space). To this end, this paper introduces a novel registration algorithm coined low-dimensional step pattern analysis (LoSPA), tailored to achieve low dimensionality while providing sufficient distinctiveness to effectively align unhealthy multimodal image pairs. The algorithm locates hypotheses of robust corner features based on connecting edges from the edge maps, mainly formed by vascular junctions. This method is insensitive to intensity changes, and produces uniformly distributed features and high repeatability across the image domain. The algorithm continues with describing the corner features in a rotation invariant manner using step patterns. These customized step patterns are robust to non-linear intensity changes, which are well-suited for multimodal retinal image registration. Apart from its low dimensionality, the LoSPA algorithm achieves about two-fold higher success rate in multimodal registration on the dataset of severe retinal diseases when compared to the top score among state-of-the-art algorithms.