Search

Purpose: To analyze the use of deviation maps (DevMs) to understand thickness changes of inner retinal layers in optical coherence tomography (OCT) data. To test a new visual analytics (VA) method with reduced complexity of OCT data analysis by comparing the layer thickness of children with type 1 diabetes mellitus (T1DM) to matched controls. Methods: OCT was performed on unilateral eyes of 26 children with T1DM without diabetic retinopathy and 29 healthy children to obtain macular volume scans. Subsequently, segmented inner retinal layers were analyzed using VA. Deviation maps were generated to readily visualize thickness differences between both groups and to investigate thickness changes of individual patients in relation to the control group. Results: In DevMs of the patient group, the total retina (TR) demonstrated localized, irregular areas of thinning (mean ± standard deviation) involving foveal center, inner macula, and inferior-nasal outer macula (−9.31 ± 1.73 µm; p < 0.05). Similarly, retinal nerve fiber layer showed continuous and localized areas of thinning in both inner and outer macula, extending nasally (−5.45 ± 4.31 µm; p < 0.05). In DevMs of individual patients, the TR and inner retinal layers revealed remarkable changes in thickness that were present between patients at both late and early stages of diabetes. Conclusion: The VA method simplifies the in-depth analysis of OCT volume data from different groups and is effective in detecting retinal thickness changes in children with diabetes. It can be easily adopted in a clinical set-up and intuitively used in complex multidisciplinary studies. [ABSTRACT FROM AUTHOR]

Optical coherence tomography (OCT) enables noninvasive high-resolution 3D imaging of the human retina, and thus plays a fundamental role in detecting a wide range of ocular diseases. Despite the diagnostic value of OCT, managing and analyzing resulting data is challenging. We apply two visual analysis strategies for supporting retinal assessment in practice. First, we provide an interface for unifying and structuring data from different sources into a common basis. Fusing that basis with medical records and augmenting it with analytically derived information facilitates thorough investigations. Second, we present a tailored visual analysis tool for presenting, emphasizing, selecting, and comparing different aspects of the attributed data. This enables free exploration, reducing the data to relevant subsets, and focusing on details. By applying both strategies, we effectively enhance the management and the analysis of retinal OCT data for assisting medical diagnoses. Domain experts applied our solution successfully to study early retinal changes in patients suffering from type 1 diabetes mellitus. [ABSTRACT FROM AUTHOR]

Visual analysis of unknown data requires the combined use of various functions that are often part of standalone visual analytics (VA) tools. Performing cross-tool visual analysis with standalone VA tools, however, is a challenging and cumbersome endeavor. Some dedicated frameworks address this issue, yet in order to utilize any of them, a visual analytics tool needs to support their required API or architecture. Contrary to most existing frameworks, we present an approach that does not rely on a single predefined interchange mechanism for the entire ensemble of VA tools. Instead, we propose using any available channel for data exchange between two consecutive VA tools. This allows mixing and matching of different data exchange strategies over the course of a cross-tool analysis. In this paper, we identify the challenges associated with establishing such tool chaining platform for data-driven coordination. We further describe the structure and capabilities of data exchange and explain various functionalities of our platform in detail. Based on a demonstrating example, we discuss the limitations of our approach and elaborate new insight for the coordination of the visual output of multiple VA tools. [Display omitted] • We provide an in-depth investigation on different aspects for the coordination of independent Visual Analytics (VA) tools that lead to the development of the the Analytical Process Constructor (AnyProc). • We address issues from related work regarding data exchange, UI integration and analytic process support and propose first solution ideas based on our conceptual preliminary work. • We discuss the idea of pairwise data exchange mechanisms for VA tool coordination in domain workflows and present design principles to deal with persisting challenges. • Our main contribution is the introduction of our platform for the coordination of independent VA tools. We describe its features in detail especially regarding different solutions to the implementation of data exchange mechanisms. • We demonstrate the use of our platform for an example scenario in the healthcare sector and link its video presentation from our original submission at the EuroVA 2021. • We discuss our recent advances regarding UI layouts and conclude by summarizing our findings and suggesting entry points for future research. [ABSTRACT FROM AUTHOR]

Optical coherence tomography (OCT) supports the detection of thickness changes in intraretinal layers at an early stage of diabetes mellitus. However, the analysis of OCT data in cross-sectional studies is complex and time-consuming. We introduce an enhanced deviation map-based analysis (MA) and demonstrate its effectiveness in detecting early changes in intraretinal layer thickness in adults with type 2 diabetes mellitus (T2DM) compared to common early treatment diabetic retinopathy study (ETDRS) grid-based analysis (GA). To this end, we obtained OCT scans of unilateral eyes from 33 T2DM patients without diabetic retinopathy and 40 healthy controls. The patients were categorized according to concomitant diabetic peripheral neuropathy (DN). The results of MA and GA demonstrated statistically significant differences in retinal thickness between patients and controls. Thinning was most pronounced in total retinal thickness and the thickness of the inner retinal layers in areas of the inner macular ring, selectively extending into areas of the outer macular ring and foveal center. Patients with clinically proven DN showed the strongest thinning of the inner retinal layers. MA showed additional areas of thinning whereas GA tended to underestimate thickness changes, especially in areas with localized thinning. We conclude that MA enables a precise analysis of retinal thickness data and contributes to the understanding of localized changes in intraretinal layers in adults with T2DM. [ABSTRACT FROM AUTHOR]