Mohammad Saleh Miri

Retinal Image Analysis Using Curvelet Transform and Multistructure Elements Morphology by Reconstruction

Retinal images can be used in several applications, such as ocular fundus operations as well as human recognition. Also, they play important roles in detection of some diseases in early stages, such as diabetes, which can be performed by comparison of the states of retinal blood vessels. Intrinsic characteristics of retinal images make the blood vessel detection process difficult. Here, we proposed a new algorithm to detect the retinal blood vessels effectively. Due to the high ability of the curvelet transform in representing the edges, modification of curvelet transform coefficients to enhance the retinal image edges better prepares the image for the segmentation part. The directionality feature of the multistructure elements method makes it an effective tool in edge detection. Hence, morphology operators using multistructure elements are applied to the enhanced image in order to find the retinal image ridges. Afterward, morphological operators by reconstruction eliminate the ridges not belonging to the vessel tree while trying to preserve the thin vessels unchanged. In order to increase the efficiency of the morphological operators by reconstruction, they were applied using multistructure elements. A simple thresholding method along with connected components analysis (CCA) indicates the remained ridges belonging to vessels. In order to utilize CCA more efficiently, we locally applied the CCA and length filtering instead of considering the whole image. Experimental results on a known database, DRIVE, and achieving to more than 94% accuracy in about 50 s for blood vessel detection, proved that the blood vessels can be effectively detected by applying our method on the retinal images.

Multimodal Segmentation of Optic Disc and Cup From SD-OCT and Color Fundus Photographs Using a Machine-Learning Graph-Based Approach

In this work, a multimodal approach is proposed to use the complementary information from fundus photographs and spectral domain optical coherence tomography (SD-OCT) volumes in order to segment the optic disc and cup boundaries. The problem is formulated as an optimization problem where the optimal solution is obtained using a machine-learning theoretical graph-based method. In particular, first the fundus photograph is registered to the 2D projection of the SD-OCT volume. Three in-region cost functions are designed using a random forest classifier corresponding to three regions of cup, rim, and background. Next, the volumes are resampled to create radial scans in which the Bruchs Membrane Opening (BMO) endpoints are easier to detect. Similar to in-region cost function design, the disc-boundary cost function is designed using a random forest classifier for which the features are created by applying the Haar Stationary Wavelet Transform (SWT) to the radial projection image. A multisurface graph-based approach utilizes the in-region and disc-boundary cost images to segment the boundaries of optic disc and cup under feasibility constraints. The approach is evaluated on 25 multimodal image pairs from 25 subjects in a leave-one-out fashion (by subject). The performances of the graph-theoretic approach using three sets of cost functions are compared: 1) using unimodal (OCT only) in-region costs, 2) using multimodal in-region costs, and 3) using multimodal in-region and disc-boundary costs. Results show that the multimodal approaches outperform the unimodal approach in segmenting the optic disc and cup.

Multimodal segmentation of optic disc and cup from stereo fundus and SD-OCT images

Glaucoma is one of the major causes of blindness worldwide. One important structural parameter for the diagnosis and management of glaucoma is the cup-to-disc ratio (CDR), which tends to become larger as glaucoma progresses. While approaches exist for segmenting the optic disc and cup within fundus photographs, and more recently, within spectral-domain optical coherence tomography (SD-OCT) volumes, no approaches have been reported for the simultaneous segmentation of these structures within both modalities combined. In this work, a multimodal pixel-classification approach for the segmentation of the optic disc and cup within fundus photographs and SD-OCT volumes is presented. In particular, after segmentation of other important structures (such as the retinal layers and retinal blood vessels) and fundus-to-SD-OCT image registration, features are extracted from both modalities and a k-nearest-neighbor classification approach is used to classify each pixel as cup, rim, or background. The approach is evaluated on 70 multimodal image pairs from 35 subjects in a leave-10%-out fashion (by subject). A significant improvement in classification accuracy is obtained using the multimodal approach over that obtained from the corresponding unimodal approach (97.8% versus 95.2%; p < 0:05; paired t-test).

A machine-learning graph-based approach for 3D segmentation of Bruch’s membrane opening from glaucomatous SD-OCT volumes

HighlightsWe present a method for segmenting Bruch’s membrane opening in 3D from SD‐OCT volumes.The problem is formulated as finding a closed‐loop minimum‐cost path in a graph.The associated cost function is computed using machine‐learning techniques.The possible presence of externally oblique border tissue is taken into consideration.The approach enables automated computation of minimum‐rim‐width‐based parameters. Graphical abstract Figure. No Caption available. ABSTRACT Bruch’s membrane opening‐minimum rim width (BMO‐MRW) is a recently proposed structural parameter which estimates the remaining nerve fiber bundles in the retina and is superior to other conventional structural parameters for diagnosing glaucoma. Measuring this structural parameter requires identification of BMO locations within spectral domain‐optical coherence tomography (SD‐OCT) volumes. While most automated approaches for segmentation of the BMO either segment the 2D projection of BMO points or identify BMO points in individual B‐scans, in this work, we propose a machine‐learning graph‐based approach for true 3D segmentation of BMO from glaucomatous SD‐OCT volumes. The problem is formulated as an optimization problem for finding a 3D path within the SD‐OCT volume. In particular, the SD‐OCT volumes are transferred to the radial domain where the closed loop BMO points in the original volume form a path within the radial volume. The estimated location of BMO points in 3D are identified by finding the projected location of BMO points using a graph‐theoretic approach and mapping the projected locations onto the Bruch’s membrane (BM) surface. Dynamic programming is employed in order to find the 3D BMO locations as the minimum‐cost path within the volume. In order to compute the cost function needed for finding the minimum‐cost path, a random forest classifier is utilized to learn a BMO model, obtained by extracting intensity features from the volumes in the training set, and computing the required 3D cost function. The proposed method is tested on 44 glaucoma patients and evaluated using manual delineations. Results show that the proposed method successfully identifies the 3D BMO locations and has significantly smaller errors compared to the existing 3D BMO identification approaches.

Graph-based optimal multi-surface segmentation with a star-shaped prior: Application to the segmentation of the optic disc and cup

A novel graph-based optimal segmentation method which can simultaneously segment multiple star-shaped surfaces is presented in this paper. Minimum and maximum surface distance constraints can be enforced between different surfaces. In addition, the segmented surfaces are ensured to be smooth by incorporating surface smoothness constraints which limit the variation between adjacent surface voxels. A consistent digital ray system is utilized to make sure the segmentation result is star-shaped and consistent, without interpolating image as required by other methods. To the best of our knowledge, the concept of consistent digital rays is for the first time introduced into the field of medical imaging. The problem is formulated as an MRF optimization problem which can be efficiently and exactly solved by computing a single min s-t cut in an appropriately constructed graph. The method is applied to the segmentation of the optic disc and cup on 70 registered fundus and SD-OCT images from glaucoma patients. The result shows improved accuracy by applying the proposed method (versus using a classification-based approach).

Incorporation of gradient vector flow field in a multimodal graph-theoretic approach for segmenting the internal limiting membrane from glaucomatous optic nerve head-centered SD-OCT volumes

The internal limiting membrane (ILM) separates the retina and optic nerve head (ONH) from the vitreous. In the optical coherence tomography volumes of glaucoma patients, while current approaches for the segmentation of the ILM in the peripapillary and macular regions are considered robust, current approaches commonly produce ILM segmentation errors at the ONH due to the presence of blood vessels and/or characteristic glaucomatous deep cupping. Because a precise segmentation of the ILM surface at the ONH is required for computing several newer structural measurements including Bruchs membrane opening-minimum rim width (BMO-MRW) and cup volume, in this study, we propose a multimodal multiresolution graph-based method to precisely segment the ILM surface within ONH-centered spectral-domain optical coherence tomography (SD-OCT) volumes. In particular, the gradient vector flow (GVF) field, which is computed from a multiresolution initial segmentation, is employed for calculating a set of non-overlapping GVF-based columns perpendicular to the initial segmentation. The GVF columns are utilized to resample the volume and also serve as the columns to the graph construction. The ILM surface in the resampled volume is fairly smooth and does not contain the steep slopes. This prior shape knowledge along with the blood vessel information, obtained from registered fundus photographs, are incorporated in a graph-theoretic approach in order to identify the location of the ILM surface. The proposed method is tested on the SD-OCT volumes of 44 subjects with various stages of glaucoma and significantly smaller segmentation errors were obtained than that of current approaches.

Automated 3D region-based volumetric estimation of optic disc swelling in papilledema using spectral-domain optical coherence tomography

The six-stage Frisén scale is a qualitative and subjective method for assessing papilledema (optic disc swelling due to raised intracranial pressure) using fundus photographs. The recent introduction of spectral-domain optical coherence tomography (SD-OCT) presents a promising alternative to enable the 3-D quantitative estimation of papilledema. In this work, we propose an automated region-based volumetric estimation of the degree of papilledema from SD-OCT. After using a custom graph-based approach to segment the surfaces of the swollen optic nerve head, the volumes of the nasal, superior, temporal, and inferior regions are computed. Using a dataset of 70 SD-OCT optic-nerve-head (ONH) SD-OCT scans the Spearman rank correlation coefficients between expert-defined Frisén scale grades and the total retinal (TR) volume, nasal, superior, temporal, inferior regional volumes were 0.737, 0.752, 0.747, 0.770 and 0.758, respectively. Also, a fuzzy k-nearest-neighbor (k-NN) algorithm was used to predict Frisén scale grades (in a leave-one-subject-out fashion). Using multiple features rather than just the TR volume made the resulting mean Frisén grade difference (MGD) between the expert-defined grades 0.386 (down from 0.629) and prediction accuracy 64.29% (up from 41.43%).

Movement of Retinal Vessels to Optic Nerve Head with Intraocular Pressure Elevation in a Child

Optic nerve neuroretinal rim thinning, retinal nerve fiber layer thinning, and peripapillary atrophy are commonly seen in glaucoma. Varma et al demonstrated nasal displacement of the retinal blood vessels (RBVs) of the optic nerve head. This was especially common in younger patients with higher intraocular pressures (IOP). In a recent article, Radcliffe et al described changes over time in the RBV position surrounding the optic nerve head in patients with chronic open angle glaucoma. They found that individuals who developed RBV positional shifts demonstrated more rapid visual field loss than those who did not have RBV shifts. This RBV shift may present problems for optic nerve head imaging devices that use the position of the retinal vasculature to register images from one examination to the next. The assumption of a static vasculature may also create difficulties in studies that use the retinal vessels as reference points to gauge optic nerve size. We describe a young patient with dramatic optic nerve and retinal vascular changes associated with a marked increase in IOP. This patient had movement of the RBVs from the retinal surface onto the optic nerve head, a change that has not been described previously. She also developed marked attenuation of her retinal vasculature. A 4-year-old girl with juvenile idiopathic arthritis was found to have bilateral posterior synechiae on a school screening examination. She had preexisting swelling of her left ankle and right elbow. Besides the synechiae, she was found to have cystoid macular edema in both eyes and was begun on prednisolone acetate drops every 2 hours while awake, atropine sulfate, and ketorolac. Three months after initial presentation she was found to have IOPs of 33 mmHg in her right eye and 27 mmHg in her left eye. Her cystoid macular edema had resolved. She was begun on a combination of timolol maleate and dorzolamide HCl in both eyes. This controlled her IOP and, on a tapering course of prednisolone acetate, her IOPs remained low. At 7 months after presentation, during a period of excellent IOP control, baseline optic nerve photographs were obtained (Fig 1A, B). At 24 months after presentation, she underwent cataract extraction and anterior vitrectomy in her left eye. The week after surgery, her IOPs rose in both eyes to 40 mmHg in her right eye and 33 mmHg in her left eye. She presented to the glaucoma service 25 months after initial presentation with IOPs of 32 mmHg in her right eye and 33 mmHg in her left eye on maximum topical therapy. Photographs of the optic nerves demonstrated a marked increase in cupping, particularly in the right eye (Fig 1C, D). She immediately underwent surgery; a trabeculotomy in the right eye and placement of an Ahmed drainage device (New World Medical, Rancho Cucamonga, CA) with corneal patch graft in the left eye. Both of these procedures were successful and, at 35 months after presentation (9 months postoperatively), her IOPs were 14 mmHg in both eyes. She was on prednisolone daily in both eyes and methotrexate 25 mg subcutaneously every week, but no glaucoma medications. There was some reversal of cupping of both optic nerves, especially the left nerve (Fig 1E, F). Because optic nerve photos were taken before her sustained elevated IOPs, this child gives us the opportunity to observe changes in the optic nerve head in young eyes with very high IOPs. One can see shifting of the retinal vessels (compare Fig 1A with Fig 1C and Fig 1B with Fig 1D) that is somewhat different from that described by Radcliffe et al. They reported shifts in vessel position on the retinal surface while this patient has movement of arterioles and venules from the retina onto the optic nerve head in both eyes. Although there has been some regression of cupping with the dramatic decrease in IOP in the right eye, her optic nerve looks paler than it did, even at the height of the IOP (compare Fig 1C with Fig 1E). Also striking is the attenuation of the retinal arterioles in the right eye that seems to be worse once the IOP was decreased than it was at the time that the IOP was elevated. The attenuation is especially severe in the areas of maximal retinal nerve fiber layer thinning on Cirrus spectraldomain optical coherence tomography (Carl Zeiss Meditec, Dublin, CA; Fig 2 available at www.aaojournal.org). Enhanced-depth optical coherence tomography imaging (Spectralis, Heidelberg Engineering, Heidelberg, Germany) performed before and after her surgery (Fig 3, available at www.aaojournal.org) demonstrates marked reversal of cupping, especially in her left eye. The width of the Bruch’s membrane opening does not seem to change, despite the marked change in optic nerve head cupping, but Bruch’s membrane does seem to bend posteriorly at high IOPs. This report supports the findings from Radcliffe et al that the retinal vasculature around the optic nerve head is not a static structure in patients with glaucoma, particularly in young patients with very high IOPs. The retinal vessels can move onto the optic nerve head and can show marked attenuation in response to glaucoma damage.

Comparative study of multimodal intra-subject image registration methods on a publicly available database

This work reports on a comparative study between five manual and automated methods for intra-subject pair-wise registration of images from different modalities. The study includes a variety of inter-modal image registrations (MR-CT, PET-CT, PET-MR) utilizing different methods including two manual point-based techniques using rigid and similarity transformations, one automated point-based approach based on Iterative Closest Point (ICP) algorithm, and two automated intensity-based methods using mutual information (MI) and normalized mutual information (NMI). These techniques were employed for inter-modal registration of brain images of 9 subjects from a publicly available dataset, and the results were evaluated qualitatively via checkerboard images and quantitatively using root mean square error and MI criteria. In addition, for each inter-modal registration, a paired t-test was performed on the quantitative results in order to find any significant difference between the results of the studied registration techniques.

Multimodal registration of SD-OCT volumes and fundus photographs using histograms of oriented gradients

With availability of different retinal imaging modalities such as fundus photography and spectral domain optical coherence tomography (SD-OCT), having a robust and accurate registration scheme to enable utilization of this complementary information is beneficial. The few existing fundus-OCT registration approaches contain a vessel segmentation step, as the retinal blood vessels are the most dominant structures that are in common between the pair of images. However, errors in the vessel segmentation from either modality may cause corresponding errors in the registration. In this paper, we propose a feature-based registration method for registering fundus photographs and SD-OCT projection images that benefits from vasculature structural information without requiring blood vessel segmentation. In particular, after a preprocessing step, a set of control points (CPs) are identified by looking for the corners in the images. Next, each CP is represented by a feature vector which encodes the local structural information via computing the histograms of oriented gradients (HOG) from the neighborhood of each CP. The best matching CPs are identified by calculating the distance of their corresponding feature vectors. After removing the incorrect matches the best affine transform that registers fundus photographs to SD-OCT projection images is computed using the random sample consensus (RANSAC) method. The proposed method was tested on 44 pairs of fundus and SD-OCT projection images of glaucoma patients and the result showed that the proposed method successfully registers the multimodal images and produced a registration error of 25.34 ± 12.34 μm (0.84 ± 0.41 pixels).