Complementary Performances of Convolutional and Capsule Neural Networks on Classifying Microfluidic Images of Dividing Yeast Cells

Abstract

Microfluidic-based assays have become effective high-throughput approaches to examining replicative aging of budding yeast cells. Deep learning may offer an efficient way to analyze a large number of images collected from microfluidic experiments. Here, we compare three deep learning architectures to classify microfluidic time-lapsed images of dividing yeast cells into categories that represent different stages in the yeast replicative aging process. We found that convolutional neural networks outperformed capsule networks in terms of accuracy, precision, and recall. The capsule networks had the most robust performance at detecting one specific category of cell images. An ensemble of three best-fitted single-architecture models achieves the highest overall accuracy, precision, and recall due to complementary performances. In addition, extending classification classes and augmentation of the training dataset can improve the predictions of the biological categories in our study. This work lays a useful framework for sophisticated deep-learning processing of microfluidics-based assays of yeast replicative aging. 4 lifetime. Microfluidics is a fast-growing tool for high-throughput applications in 5 chemical, biological, optics, and information technology, including single-cell imaging 6 analysis [3]. Typically, microfluidic images are taken in time intervals with relatively low 7 resolution compared to confocal microscopic images that are often of high resolution, 8 rendering unique challenges for microfluidic imaging [4]. Capturing the full progression 9of cellular replicative lifespans requires identifying both mother cells and daughter cells 10 in full cell cycles. However, the full automation of this process is often hindered by low 11 image resolution, demanding time-consuming, manual classifications of yeast replicative 12 lifespans [5].