Particle Quantification from a Smartphone-based Biosensor using Deep Convolutional Neural Network for Clinical Diagnosis

Abstract

Biological cell quantification is an important step in diagnosing and strategizing treatment for many infections, cardiovascular diseases, and biomarker discovery which in turn helps in understanding immunological and genetic disorders, cancers, etc. A point-of-care diagnostic device integrated with microfluidic systems can benefit such applications by accelerating the diagnosis procedures and making it accessible throughout the world. Here, we present a computer vision methodology to aid particle and cell counting from images acquired by the novel smartphone based microfluidic biosensor. We implement a convolutional neural network architecture to train, validate and test it with different experimental datasets. This method proved to obtain results faster and analogous to that of the benchmark techniques.