Optimal Transport-based Graph Matching for 3D retinal OCT image registration

Abstract

Registration of longitudinal optical coherence tomography (OCT) images assists disease monitoring and is essential in image fusion applications. Mouse retinal OCT images are often collected for longitudinal study of eye disease models such as uveitis, but their quality is often poor compared with human imaging. This paper presents a novel but efficient framework involving an optimal transport based graph matching (OT-GM) method for 3D mouse OCT image registration. We first perform registration of fundus-like images obtained by projecting all b-scans of a volume on a plane orthogonal to them, hereafter referred to as the x-y plane. We introduce Adaptive Weighted Vessel Graph Descriptors (AWVGD) and 3D Cube Descriptors (CD) to identify the correspondence between nodes of graphs extracted from segmented vessels within the OCT projection images. The AWVGD comprises scaling, translation and rotation, which are computationally efficient, whereas CD exploits 3D spatial and frequency domain information. The OT-GM method subsequently performs the correct alignment in the x-y plane. Finally, registration along the direction orthogonal to the x-y plane (the z-direction) is guided by the segmentation of two important anatomical features peculiar to mouse b-scans, the Internal Limiting Membrane (ILM) and the hyaloid remnant (HR). Both subjective and objective evaluation results demonstrate that our framework outperforms other well-established methods on mouse OCT images within a reasonable execution time.Retinal image registration aims to spatially align two or more retinal images captured at different times and from different viewpoints for clinical review of disease progression. Nowadays, non-invasive Optical Coherence Tomography (OCT), which enables 3-D retinal imaging, is widely used for in-vivo assessment of eye disease [1,2]. Data is generated by acquiring a series of A-scans that are combined to a 2D cross-sectional slice called a B scan. The summation of B-scans creates a fundus like projection image.An efficient longitudinal mouse OCT registration method has the potential to impact both early studies of disease mechanisms and treatment. It also helps generating strategies for analysis that are translatable to human imaging. However, imaging the mouse retina is more challenging and less well studied than is the case in humans. High-quality mouse OCT is usually expensive and hard to acquire. Low quality datasets suffer from major vascular dissimilarities,

CellStitch: 3D Cellular Anisotropic Image Segmentation via Optimal Transport

“…In particular, PASTE [24] performs pairwise spot alignment across adjacent Visium layers [25], and SCOTT [26] designs a shape-location combined system for cell tracking in 2D microscopy videos. Other applications of optimal transport have addressed 2D and 3D image registrations such as retinal fundus alignment [27,28]. However, those methods cannot be easily applied to fluorescent images with dense, cluttered cells, because they mainly focus on spot or pixellevel alignment with single or at most a few objects of interest.…”

Cellstitch: 3D cellular anisotropic image segmentation via optimal transport

“…In particular, PASTE [24] performs pairwise spot alignment across adjacent Visium layers [25], and SCOTT [26] designs a shapelocation combined system for cell tracking in 2D microscopy videos. Other applications of optimal transport have addressed 2D and 3D image registrations such as retinal fundus alignment [27,28]. However, those methods cannot be easily applied to fluorescent images with dense, cluttered cells, because they mainly focus on spot or pixel-level alignment with single or at most a few objects of interest.…”