Imaging forms the backbone of living subjects research. Living subjects research is both essential to the progress of translational medicine and very expensive. The research community actively seeks to develop and validate new clinical endpoints to solve a range of etiology, natural history, diagnostic and prognostic problems. This project aims to develop and commercialize LATTICE, an Electronic Research Record, Image Management and Sharing Solution, and Deep Learning Platform. LATTICE is designed to increase the efficiency of imaging-driven biomedical research and clinical trials. This efficiency is accomplished first through a structured workflow that includes protocol management, subject scheduling, and records collection from multiple imaging modalities. Access to imaging and associated data within the same workflow simplifies the process for the research team. Structuring the data into a de-identified, privacy-managed Image Bank enables sharing for collaboration and re-use for retrospective research. Image processing algorithms connected to the Image Bank facilitate batch analysis, while the system also provides a platform for the development of new image-based outcome measures and clinical endpoints. A key objective of LATTICE is to enable investigators and collaborators to accelerate the translation of insights to the clinic with maximum efficiency. Successful translation requires structuring the workflow, record keeping, and protocols into a rigorous, transparent, reproducible and validated process. LATTICE is designed to reduce the friction in translating successful research projects to the clinic. Researchers in the Advanced Ocular Imaging Program (AOIP) at the Medical College of Wisconsin developed elements of LATTICE as separate technologies.
The Specific Aims of this proposal are directed to an integrated workflow addressing a broader set of objectives. The AOIP LATTICE Electronic Research Record will be translated into a commercially managed repository and brought under regulatory Design Control. The current AOIP Image Bank containing 3,000,000 de- identified retinal images will be integrated into the LATTICE workflow. Critically, this integration will allow the sharing of the Image Bank with external researchers. Three retinal image process algorithms that operate on retinal images will integrate into this workflow. These algorithms include analysis of adaptive optics images of the fundus, analysis of the foveal avascular zone from optical coherence tomography angiography (OCTA), and model-based analysis of the fovea imaged with OCT. A computational deep learning workflow will also be prototyped using a cloud-based architecture. This final workflow will be constructed to demonstrate the feasibility of deploying a collaborative deep learning environment for the development of new clinical endpoints using shared, de-identified images. LATTICE will be a unique system for both prospective and retrospective translational research. LATTICE will make a profound impact on the cost of managing image-based research and add leverage to translational research expenditures for moving insights into the clinic.
LATTICE is an innovative electronic research record and development platform for image-based ophthalmic research. LATTICE is designed to reduce the cost of translational research, promote the re- use of images, and simplify the development and application of new techniques to analyze medical images. LATTICE will integrate research workflow tools with a database of 3,000,000 retinal images and advanced image processing software to accelerate the process of translating eye research insights from the lab to the clinic.
