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

Image Segmentation Using Active Contours with Hessian-Based Gradient Vector Flow External Force

The gradient vector flow (GVF) model has been widely used in the field of computer image segmentation. In order to achieve better results in image processing, there are many research papers based on the GVF model. However, few models include image structure. In this paper, the smoothness constraint formula of the GVF model is re-expressed in matrix form, and the image knot represented by the Hessian matrix is included in the GVF model. Through the processing of this process, the relevant diffusion partial differential equation has anisotropy. The GVF model based on the Hessian matrix (HBGVF) has many advantages over other relevant GVF methods, such as accurate convergence to various concave surfaces, excellent weak edge retention ability, and so on. The following will prove the advantages of our proposed model through theoretical analysis and various comparative experiments.

Image Segmentation-Based Cervical Spine MRI Images to Evaluate the Treatment of Patients with Chronic Pain

The objective of this research was to investigate the application effect of cervical spine magnetic resonance imaging (MRI) image segmentation algorithm guidance in the treatment of chronic pain with cervical epidural puncture. A total of 72 patients with chronic pain were selected and divided into a cervical spine MRI image-guided group (group A) and a blind puncture group with traditional experience (group B). The results showed that the puncture time of group A was $9.9\pm 8.2$ (min), while that of group B was $15.2\pm 8.9$ (min), so the puncture time of patients in group A was significantly shorter than that of group B ( $P<0.05$ ). The incidences of pain at the puncture site of patients in group A and group B were 6% and 10%, respectively. The incidence of pain at the puncture site in group A was significantly lower than that in group B ( $P<0.05$ ). The success rate of the first puncture in group A was 78%, and that in group B was 54%. The success rate of the first puncture in group A was significantly higher than that in group B ( $P<0.05$ ). The complication rate of group A was 22.22%, and that of group B was 80.56%. The incidence of complications in group A was significantly lower than that in group B ( $P<0.05$ ). In addition, there was no significant difference in the puncture depth between the two groups ( $P>0.05$ ). In summary, the guidance of cervical spine MRI image segmentation algorithm can reduce the time and times of puncture and improve the success rate of puncture, thereby reducing the incidence of postoperative complications.

Macular Hole Detection Using a New Hybrid Method: Using Multilevel Thresholding and Derivation on Optical Coherence Tomographic Images

Optical coherence tomography (OCT) is a noninvasive imaging test. OCT imaging is analogous to ultrasound imaging, except that it uses light instead of sound. In this type of image, microscopic quality intratissue images are provided. In addition, fast and direct imaging of tissue morphology and reproducibility of results are the advantages of this imaging. Macular holes are a common eye disease that leads to visual impairment. The macular perforation is a rupture in the central part of the retina that, if left untreated, can lead to vision loss. A novel method for detecting macular holes using OCT images based on multilevel thresholding and derivation is proposed in this paper. This is a multistep method, which consists of segmentation, feature extraction, and feature selection. A combination of thresholding and derivation is used to diagnose the macular hole. After feature extraction, the features with useful information are selected and finally the output image of the macular hole is obtained. An open-access data set of 200 images with the size of 224 × 224 pixels from Sankara Nethralaya (SN) Eye Hospital, Chennai, India, is used in the experiments. Experimental results show better-diagnosing results than some recent diagnosing methods.

Active contour method for ILM segmentation in ONH volume scans in retinal OCT

The optic nerve head (ONH) is affected by many neurodegenerative and autoimmune inflammatory conditions. Optical coherence tomography can acquire high-resolution 3D ONH scans. However, the ONH's complex anatomy and pathology make image segmentation challenging. This paper proposes a robust approach to segment the inner limiting membrane (ILM) in ONH volume scans based on an active contour method of Chan-Vese type, which can work in challenging topological structures. A local intensity fitting energy is added in order to handle very inhomogeneous image intensities. A suitable boundary potential is introduced to avoid structures belonging to outer retinal layers being detected as part of the segmentation. The average intensities in the inner and outer region are then rescaled locally to account for different brightness values occurring among the ONH center. The appropriate values for the parameters used in the complex computational model are found using an optimization based on the differential evolution algorithm. The evaluation of results showed that the proposed framework significantly improved segmentation results compared to the commercial solution.

An automated method for choroidal thickness measurement from Enhanced Depth Imaging Optical Coherence Tomography images

The choroid is vascular tissue located underneath the retina and supplies oxygen to the outer retina; any damage to this tissue can be a precursor to retinal diseases. This paper presents an automated method of choroidal segmentation from Enhanced Depth Imaging Optical Coherence Tomography (EDI-OCT) images. The Dijkstra shortest path algorithm is used to segment the choroid-sclera interface (CSI), the outermost border of the choroid. A novel intensity-normalisation technique that is based on the depth of the choroid is used to equalise the intensity of all non-vessel pixels in the choroid region. The outer boundary of choroidal vessel and CSI are determined approximately and incorporated to the edge weight of the CSI segmentation to choose optimal edge weights. This method is tested on 190 B-scans of 10 subjects against choroid thickness (CTh) results produced manually by two graders. For comparison, results obtained by two state-of-the-art automated methods and our proposed method are compared against the manual grading, and our proposed method performed the best. The mean root-mean-square error (RMSE) for finding the CSI boundary by our method is 7.71±6.29 pixels, which is significantly lower than the RMSE for the two other state-of-the-art methods (36.17±11.97 pixels and 44.19±19.51 pixels). The correlation coefficient for our method is 0.76, and 0.51 and 0.66 for the other two state-of-the-art methods. The interclass correlation coefficients are 0.72, 0.43 and 0.56 respectively. Our method is highly accurate, robust, reliable and consistent. This identification can enable to quantify the biomarkers of the choroidin large scale study for assessing, monitoring disease progression as well as early detection of retinal diseases. Identification of the boundary can help to determine the loss or change of choroid, which can be used as features for the automatic determination of the stages of retinal diseases.

Structure-Function Relationships in Perimetric Glaucoma: Comparison of Minimum-Rim Width and Retinal Nerve Fiber Layer Parameters

Purpose

To test the hypotheses that: (1) structure–function (SF) relationships between visual fields (VF) and Bruch's membrane opening-based minimum rim width (BMO-MRW) measurements are superior to those for peripapillary retinal nerve fiber layer (pRNFL) in perimetric glaucoma; (2) BMO-MRW measurements may extend the utility of structural measurement across the range of glaucoma severity; and (3) to estimate the influence of Bruch's membrane opening (BMO) size on BMO-MRW measurements.

Methods

One hundred eight perimetric glaucoma eyes (68 patients) with good quality spectral-domain optical coherence tomography images of the optic disc and pRNFL, and reliable VF within 6 months were recruited. Relationship of global and sectoral BMO-MRW and pRNFL thickness with corresponding VF parameters and the influence of normalizing BMO-MRW (on BMO circumference, nBMO-MRW) on SF relationships were investigated. Broken-stick models were used to compare the point at which pRNFL and BMO-MRW parameters reached their measurement floor.

Results

The median (interquartile range) of VF mean deviation was −5.9 (−12.6 to −3.6) dB. Spearman correlation coefficients between pRNFL, BMO-MRW, and nBMO-MRW measures and corresponding VF cluster average deviations ranged between 0.55 to 0.80, 0.35 to 0.66, and 0.38 to 0.65, respectively. Bruch's membrane opening–MRW parameters demonstrated weaker SF relationships compared with pRNFL globally and in temporal, temporal-superior, and nasal-inferior sectors (P < 0.03). Normalization of BMO-MRW did not significantly influence SF relationships.

Conclusions

Structure–function relationships were somewhat weaker with BMO-MRW parameters compared with pRNFL in eyes with perimetric glaucoma. Bruch's membrane opening–MRW normalization did not significantly change SF relationships in this group of eyes with mild departures from average BMO size.

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).