Purchase this article with an account.
A. R. Fuller, R. J. Zawadzki, B. Hamann, J. S. Werner; Real-Time Interactive Software for Analysis of 3D OCT Retinal Data. Invest. Ophthalmol. Vis. Sci. 2008;49(13):4012. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Visualization and interactive exploration of volumetric data is increasingly available for retinal and optic nerve data acquired through optical coherence tomography (OCT), but there are several barriers to interrogation of these volumes for clinical and scientific analysis. The narrow field of view of the scanning system limits the scope of the structures that can be examined while the large size of the data can obscure specific characteristics of interest. Additionally, the relationships between different aspects of the volume both internally and with respect to clinical photos are often unclear. This work addresses these issues while maintaining a clinically applicable interface.
To increase the field of view of the volumes acquired through OCT we have developed an intuitive, user-friendly interface to stitch together multiple volumes by simply ‘drag and dropping’ them into a single visualization. By combining partially overlapping volumes, this method is able to reconstruct larger regions of the retina than the OCT system is normally capable of capturing. Additionally, we present two different types of volumetric annotation to help clarify the relationships of various aspects of the volume. The "image annotation" allows the user to overlay a two-dimensional fundus image on the volume. This defines a global relationship between the volume and this well-understood imaging modality. The "paint annotation" allows the user to mark specific points in the volume by painting on slices of the volume. These points can then be projected onto a fundus image or viewed in a volume rendering. This proves useful when correlating corresponding points on different images or views. Paint annotations can also be used to classify and isolate retinal structures of interest through a machine learning algorithm. This allows the user to restrict the amount of data being viewed to more manageable amounts. This classification can also be used to derive quantitative information or thickness maps.
These methods have been tested by clinicians for use in analyzing volumetric OCT data of the macula and optic disc. They have proven to be clinically useful and require little user training.
To fully realize the potential of three-dimensional OCT it is essential to develop methods that simplify the extraction of relevant information and relate it to more familiar clinical images. The generalized nature of our methods allows them to be useful in many types of scientific inquiry.
This PDF is available to Subscribers Only