Final Report Summary - ENDO-VV (eVV - endoVascularVision: new endoscopic tools for real time vascular assisted vision) The main goal of the project was to develop advanced endoscopic systems for the assisted visualization of vascular structures, in particular for the visualization of submucosal vessels. The industrial partner, TMB, had proof of principle and a patent on endoscopic assisted visualization based on the use of infrared light and spectral segmentation. The project aimed at bringing preliminary evidence to pre-clinical solutions, by evaluating the feasibility and developing a series of assisted endoscopic vision modalities, all based on the infrared illumination principle and spectral segmentation algorithms.The project was originally structured in four work packages (WP). WP 1 to 3 were designed in a sequential fashion in order to first develop the basis of a robust system for submucosal vessel detection (WP1), and later develop further refinements and vision modalities using this basis, which included vessel functional assessment (WP2), and vascular map generation (WP3). The goal of WP4 was to incorporate the developments in the previous work packages in a single clinical tool.The work started as planned, and several important steps were achieved, including a robust segmentation algorithm for superficial vessels, which used both visible and infrared lights. Likewise, a fully functioning hardware, capable of working in human and experimental surgery was developed. Inictially the system worked by recording images and post-processing, but later in the project, CUDA software engineers developed a fully functioning real-time system., which was refined through a series of experiments by the surgical groups. The main difficulties appeared with the robust detection of submucosal vessels. The technique proposed was found to be more difficult than expected when dealing with conditions of hidden vessels and under real clinical situation (movement, changes is depth and inclination...). This forced to reconsider the technique used and to a significant delay of the expected duration of WP1. At about mid project, the work revealed that the solution for submucosal identification of vessels was too unstable to be used in experimental pre-clinical conditions. Other aspects of the prototype progressed satisfactorily, but not having a robust submucosal vessel detection system represented an impediment to progress satisfactorily in WP2 and for this reason the partners decided to continue the second half of the project focusing efforts in the research aspects of WP1, in order to experiment with new illumination modalities and camera systems to achieve reproducible performance under clinically relevant conditions. This required further research efforts that were not scheduled in the beginning, and in summary a whole restructuring of the project as it was originally designed.The main changes introduced were a redesign of the research plan, and the introduction of a new partner (TNO), which should provide extra insight required to overcome the limitations encountered. Eventually, administrative delays prevented TNO from effectively joining the work done within the project, but fruitful discussions were held and it is expected that a collaboration will be established beyond this project. The research conducted henceforth was mainly focused along two lines: (1) Improving the illumination-camera system. Using phantoms and animal experiments, we tried to better understand the requirements in terms of wavelength (tested from 808 to 1510nm) and camera that affect the quality of that signal under clinical conditions. (2) Improving signal processing. Engineers at the industrial partner tried to improve mathematical processing algorithms. Important part of the work was dedicated to particularly exploring how texture analysis could improve vessel segmentation using tools transported from other solutions developed by the company for the segmentation and quantitative analysis of images. The project had to finalize to comply with the duration set forth in the contract, and consequently some of the important research that is required to solve the submucosal identification of vessel will have to continue through partners’ collaborations in the future. In summary the project has developed a fully functioning endoscopic instrumentation system prototype that identifies vascular structures by spectral segmentation using a combined signal of the visible and infrared light. The system is integrated in a endoscopic tower which allows providing a double screen signal to the surgeon, the normal and the “vascular” vision, which can be seen in parallel screens or in a superimposed manner. The system is also capable of creating a third information screen with all relevant illumination and camera technical data for the engineers working and fine-tuning the system in real time and under real operating conditions in the experimental setting. Finally, the systems works fully in real time. This prototype is expected to be used for future experiments to continue until a clinical device can be developed. Pending work for the immediate future is solving the challenge of identifying submucosal, sub-fat, non-visible vessels. This aim will require very dedicated research and we expect the partners of this project to continue working in collaboration in the near future.Eduard Gratacos http://endo-vv.com/Sabino Arana 1Hospital Clínic-Medicina Materno Fetal08028 - Barcelona0034 93 227 9333
