Ultrafast optical coherence tomography algorithms for real-time processing**********

现代计算机中出现了更快的通信协议，导致快速数据采集的应用越来越多，从而提供了新的见解。随着傅立叶域锁模(FDML)扫描源激光器的出现，光学相干层析成像(OCT)正在经历一场变革，它可以与平衡检测干涉仪相结合，以A线速率实现高达MHz频率的OCT采集。这些系统预计将产生视频率的体积OCT图像数据。AlazarTech最近开发了一种专用采集卡，可以采集数据并以高达4 GHz的速率直接传输到GPU。该卡目前正在集成到Optores开发的OCT系统中。然而，基于FDML激光器的OCT快速傅里叶变换(FFT)所需的波数(即k空间)线性化所需的模数转换器在非均匀时钟下工作的能力不能被转换到更高的硬件频率上。本项目将开发基于在GPU上实现的非均匀FFT的OCT图像形成算法，以提供视频率体积OCT图像。我们将在需要获取速度的环境中进一步演示：大规模组织学，与当前的工作系统相比，我们预计加速20倍。*

The emergence of faster communication protocols in modern computers has led to increasing applications of fast data acquisition delivering novel insights. Optical Coherence tomography (OCT) is now undergoing a transition with the emergence of Fourier Domain Mode Locking (FDML) swept source lasers which can be combined with a balanced detection interferometer to achieve OCT acquisitions at A-line rates reaching MHz frequencies. These systems are anticipated to yield video-rate volumetric OCT image data. AlazarTech has recently developed a dedicated acquisition card that can acquire and stream data directly to GPUs at rates reaching 4GHz. The card is currently being integrated in OCT systems developed by Optores. However, the previous capability of analog-to-digital converters to be operated on a non-uniform clock, necessary to linearize the wave-numbers (i.e. k-space) for Fast Fourier Transform (FFT) in OCT based on FDML lasers, cannot be translated to higher hardware frequencies.This project will develop OCT image formation algorithms, based on a non-uniform FFT implemented on the GPU, to deliver video-rate volumetric OCT images. We will further perform a demonstration in a context where acquisition speed is required: large scale histology where we anticipate an acceleration by a factor of 20 compared to current working systems.****