Week 1: Starting Up

After much deliberation and with the help of Dr. Asghari, we have chosen an optical coherence tomography and image processing based project that will be called Ultrafast Imaging as our senior capstone. With a plethora of resources from past contributors as well as Dr. Asghari and graduate students, we believe we can accomplish great things with this project. The overall objective of this project is to enhance the existing optical coherence tomography (OCT) system to produce renderings at faster rates and in greater volumes. The PC system at hand runs on the integrated graphics of the CPU. We firmly believe that a discrete GPU will relieve much of the rendering load off the CPU and increase the overall performance of the system.

Our general goal for the Fall 2019 semester is to understand, maintain, and alleviate the current bottlenecks of the established system while learning and developing a foundation for the implementation of a discrete graphics card. After a quick walkthrough of the system, we brainstormed potential areas of improvement. Currently the oscilloscope and ethernet cable being used are rated for speeds in the gigabit range, however they are both running in the megabit range. This is one of the first and main bottlenecks that need to be alleviated before the implementation of any graphics card. Although a GPU would be helpful to the system in its current state, it will not be anywhere near reaching its full potential if the data being streamed in is not fast enough to utilize the card. The system as it stands now can produce an image at four frames per second and can create a volumetric image of 100 pixels cubed. Another potential bottleneck of the system is the software that communicates with the oscilloscope and displays an image. This past summer, a team of students developed Python code to interface with the oscilloscope and render the image on the computer. We plan to assess the runtime and determine which sections could be optimized.

The first step to understanding the system begins with the laser itself. As per our understanding, the laser essentially produces a pulse train of delta functions with a period of ten nanoseconds and a width in the femtoseconds range. The laser then goes through a multi-stage system. The following stages we are still in the process of researching and plan to come with questions to our next lab walkthrough with Dr. Asghari.

The ultimate goal of this project is to create the fastest real-time image processing system to date. The current world record for this type of system is twenty frames per second. With a dedicated and passionate team, along with the support of Dr. Asghari and the LMU faculty, we believe that this goal is within reach. We look forward to the growth and knowledge we will gain throughout this process.