Microfluidics has shown great promise for the sorting or separation of biological cells such as circulating cancer cells since the first studies came out a few decades ago. With recent advances in high-throughput microfluidics, analysis of massive amounts of data needs to be completed in an iterative, timely manner. However, the majority of analysis is either performed manually or through the use of superimposing multiple images to define the flow of the particles. These forms of analysis are limited by the resolution of the image acquisition system and the user’s abilities to discern information from the images. For high throughput microfluidic devices, analysis typically requires a camera that is capable a high frame rate, from a few hundred of frames per second to a few thousand frames per second, producing hundreds of gigabytes of data in a short time. To analyze this much data manually, it takes countless numbers of hours. 

 The objective of the work is to increase the efficiency and repeatability of particle detection in a high-throughput deterministic lateral displacement (DLD) device. DLD is a passive microfluidic technique applied to the separation of particles or biological cells with precisely designed pillar arrays. These pillar arrays are carefully designed to create separate flow laminae, which are known and directed by the arrangement of the posts. Particles below a defined critical diameter are contained within the original flow streamline, while those larger than the diameter move to another flow. For example, circulating cancer cells in blood will be separated from the red and white blood cells, thus indicating the presence of a cancerous tumor in the body, with only using a blood sample, rather than extensive amounts of expensive testing and analysis. In addition to the minimal cost, microfluidic techniques require a significantly smaller sample compared to traditional analysis. 

The ultimate goal is to increase the efficiency and effectiveness of the design of a DLD device and its analysis. With the proposed tool, experiments will have more consistent results and it will be possible to save time analyzing the results of the experiments run through a DLD device. An additional goal of the tool is to provide more near real-time data feedback, allowing the user to monitor the progress of the experiment as it is run. The stored information gained from the analysis tool will be then be implemented into machine learning or deep learning algorithms in order to optimize and predict the effectiveness DLD design given design input parameters.


Towards Deep Learning Approaches for Quantitative Analysis of High-Throughput DLD

Division / Sponsors: 
Aerospace, Bioengineering, Applied Mechanics, Design Engineering Division, Advanced Energy Systems, Manufacturing Engineering Division, Noise Control and Acoustics, Nondestructive Evaluation, Diagnosis, and Prognosis (NDPD), Nuclear Engineering Division, Process Industries Division, Fluids Engineering, Heat Transfer, Materials, Microelectromechanical Systems, Safety Engineering and Risk Analysis, Computers & Information in Engineering


Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2020/volume/84478

IMECE2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.