Deep Convolutional Neural Networks for Human Embryonic Cell Counting

We address the problem of counting cells in time-lapse microscopy images of developing human embryos. Cell counting is considered as an important step in analyzing biological phenomenon such as embryo viability. Traditional approaches to counting cells re

  • PDF / 843,668 Bytes
  • 10 Pages / 439.37 x 666.142 pts Page_size
  • 96 Downloads / 244 Views

DOWNLOAD

REPORT


College of Engineering and Computer Science, The Australian National University, Canberra, Australia [email protected] 2 CVLab, EPFL, Lausanne, Switzerland

Abstract. We address the problem of counting cells in time-lapse microscopy images of developing human embryos. Cell counting is considered as an important step in analyzing biological phenomenon such as embryo viability. Traditional approaches to counting cells rely on hand crafted features and cannot fully take advantage of the growth in data set sizes. In this paper, we propose a framework to automatically count the number of cells in developing human embryos. The framework employs a deep convolutional neural network model trained to count cells from raw microscopy images. We demonstrate the effectiveness of our approach on a data set of 265 human embryos. The results show that the proposed framework provides robust estimates of the number of cells in a developing embryo up to the 5-cell stage (i.e., 48 h post fertilization).

1

Introduction

Counting the number of objects in an image is an important and challenging computer vision problem that arises in many real-world applications ranging from crowd monitoring to biological research. In biological research counting cells is a fundamental first step for further analysis (e.g., cell mitosis detection and cell lineage analysis). In this paper we focus on the problem of determining the number of cells in time-lapse microscopy images of developing human embryos. We are primarily interested in images of embryos up to the 5-cell stage, which have been used in other works for computing biomarkers (e.g., cell timing parameters) to assess embryo viability in the context of in vitro fertilization (IVF) treatments [14,19]. Manual cell counting, is an extremely tedious process that is prone to error and subject to intra- and inter-individual variability. Automating the process has the benefit of reducing time and cost, minimizing errors, and improving consistency of results between individuals and clinics. To simplify the task and improve robustness, many researchers stain the cells prior to automatic counting [1,4,5,20]. However, cell staining is not feasible for many applications (such as IVF embryo assessment). Counting non-stained cells in dark-field microscopy images is difficult because of constraints in the imaging process. For example, the exposure time, the light c Springer International Publishing Switzerland 2016  G. Hua and H. J´ egou (Eds.): ECCV 2016 Workshops, Part I, LNCS 9913, pp. 339–348, 2016. DOI: 10.1007/978-3-319-46604-0 25

340

A. Khan et al.

(a)

(b)

(c)

(d)

(e)

Fig. 1. Examples of developing embryos: (a) one-cell stage, (b) two-cell stage, (c) threecell stage, (d) four-cell stage and (e) 5-or-more cell stage

intensity and the transparency of the specimen all cause variations in the image quality and result in faint cell boundaries. Analysis of human embryonic cells is further challenged by the fact that the cells exhibit variability in appearance and shape. Also, each embryo grows (ce