Elsevier

Pattern Recognition

Volume 42, Issue 4, April 2009, Pages 498-508
Pattern Recognition

Recognition and analysis of cell nuclear phases for high-content screening based on morphological features

https://doi.org/10.1016/j.patcog.2008.08.003Get rights and content

Abstract

Automated analysis of molecular images has increasingly become an important research in computational life science. In this paper some new and efficient algorithms for recognizing and analyzing cell phases of high-content screening are presented. The conceptual frameworks are based on the morphological features of cell nuclei. The useful preprocessing includes: smooth following and linearization; extraction of morphological structural points; shape recognition based morphological structure; issue of touching cells for cell separation and reconstruction. Furthermore, the novel detecting and analyzing strategies of feed-forward and feed-back of cell phases are proposed based on gray feature, cell shape, geometrical features and difference information of corresponding neighbor frames. Experiment results tested the efficiency of the new method.

Introduction

The tracing and recognition of cell phases using fluorescence microscopy images play an important role for any automated high-content screening that helps scientists to better understand the complex process of cell division or mitosis [1], [2], [3], [4], [5], [6], [7], [8]. High-content screening concerns with the tracking of cell-cycle progression (interphase, prophase, metaphase, anaphase and telophase), which can be identified by measuring nuclear changes. The most difficult task of such analysis is finding different stages during cell mitosis. For example, six image frames (Frames 7–12) taken from a database of high-content screening are enhanced by contrast stretching [9], and shown in Figs. 1(a–f), respectively. One problem is that there are some touching nuclei in a frame, which can be found from the frames in Figs. 1(a–f). In order to determine which phase a cell is, the shape and size of each cell need to be found, and touching nuclei make it difficult [5]. Therefore, it is necessary that the detection, segmentation and reconstruction of touching nuclei are investigated in this paper.

A typical nuclear migration during cell division is shown in Fig. 2. For the cell of being at bottom-left in different frames, it is in prophase in the frames of Figs. 2(a–d), it is in metaphase in the frames of Figs. 2(e and f), and it is in anaphase in the frames of Figs. 2(g and h) (two spindles being formed). It is difficult to distinguish interphase of nuclei from other phases [5], [6], [8].

Related work: Some methods are used to recognize these nuclear phases, and these are LBG: Linde-Buzo–Gray's method, VQ: vector quantization, FE: fuzzy entropy, FCM: fuzzy c-means, GMM: Gaussian mixture models [7]. There are three main questions in these methods. One is that the prior information of nuclear mitosis is not used. The second one is that the cases of touching nuclei are not considered. The last one is that the extracted features are not enough. In fact, if only geometrical and gray features are used, it is very difficult to distinguish the differences between the interphase and prophase.

In this paper, we develop new and efficient algorithms for detecting and analyzing cell phases based on all morphological features such as gray feature, shape recognition, geometrical features and prior information of normal cellular cycle. The contribution of this paper is that several new methods are used to recognize and analyze the phases of cell nuclei. These novel methods are shape recognition based morphological structure, new segmentation method for touching nuclei, new method for detecting the cellular metaphase feed-forward detection and feed-back detection, which are used and defined firstly. The general scheme of the proposed method is graphically shown in Fig. 3. The rest of this paper is organized as follows. Section 2 describes the preprocessing of cell images. Section 3 describe the method of detecting and analyzing cell nuclei phases. Finally, Section 4 concludes the findings of our work.

Section snippets

Preprocessing

Otsu's method [10] is used to segment images in Fig. 1, Fig. 2, and the binary results are shown in Fig. 4, Fig. 5, respectively.

Accurate description of image contours plays an important role for the shape analysis and recognition of images. Line segment, critical points and their convexity and concavity are useful features to analyze the shape of the image contour. There exist several methods and algorithms for the description of contours [11], [12], [13], [14], [15]. However, these methods

Tracing recognition and analysis of cell nuclei phases

For tracing detection and analysis of cell nuclei phases, some geometrical features of each frame need to be extracted as follows: binarizing the frame image; labeling the frame image; extract the area of each object; extract the centroid of each object; extract the lengths of the major and minor axes of each object.

A cell's average time in the normal cellular cycle is 2175 min (about series of 145 frames) for the cell nuclei database used by us based on statistical results of some samples.

Experiments and conclusion

An efficient and new method has developed to recognize, trace, and analyze the phases of nuclei in frames of a high-contents cell-cycle based on shape recognition, morphological and geometrical features, gray feature and prior information of normal cellular cycle. This method simulates artificial intelligence. All related thresholds such as Lt, (Lmaa/Lmia)>1.5, Gnt, Gat are found from 496 sample nuclei where 153 nuclei are in metaphase, 77 nuclei are in anaphase, 80 nuclei are telophase, and

Acknowledgment

This work was supported by the Australia Research Council ARC-DP Grant (DP0665598) to T.D. Pham. The cell images were provided by Dr. Randy King of the Department of Cell Biology, Harvard Medical School.

About the Author—DONGGANG YU received his Ph.D. in Faculty of Information and Communication Technologies from Swinburne University of Technology, Melbourne, Australia. His research interests are image processing, pattern recognition and bioinformatics. He is doing research as a research fellow (lecturer) in School of Design, Communication and Information Technology, The University of Newcastle, Australia.

References (18)

  • A.M.N. Fu et al.

    A curvature angle bend function based method to characterize contour shapes

    Pattern Recognition

    (1997)
  • D. Yu et al.

    An efficient algorithm for smoothing, linearization and detection of structure feature points of binary image contours

    Pattern Recognition

    (1997)
  • S. Fox

    Accommodating cells in HTS

    Drug Discovery World

    (2003)
  • Y. Feng

    Practicing cell morphology based screen

    Eur. Pharm. Rev.

    (2002)
  • R. Dunkle

    Role of image informatics in accelerating drug discovery and development

    Drug Discovery World

    (2003)
  • J.C. Yarrow

    Phenotypic screening of small molecule libraries by high throughput cell imaging

    Comb. Chem. High Throughput Screen.

    (2003)
  • X. Chen et al.

    Automated segmentation, classification, and tracking cancer cell nuclei in time-lapse microscopy

    IEEE Trans. Biomed. Eng.

    (2006)
  • T.D. Pham, D. Tran, X. Zhou, S.T.C. Wong, An automated procedure for cellphase imaging identification, in: Proceedings...
  • T.D. Pham et al.

    Integrated algorithms for image analysis and classification of nuclear division for high-content cell-cycle screening

    Int. J. Comput. Intell. Appl.

    (2006)
There are more references available in the full text version of this article.

About the Author—DONGGANG YU received his Ph.D. in Faculty of Information and Communication Technologies from Swinburne University of Technology, Melbourne, Australia. His research interests are image processing, pattern recognition and bioinformatics. He is doing research as a research fellow (lecturer) in School of Design, Communication and Information Technology, The University of Newcastle, Australia.

About the Author—TUAN D. PHAM received his Ph.D. degree in 1995 from the University of New South Wales. His current research interests include image processing, molecular and medical image analysis, pattern recognition, bioinformatics, biomedical informatics, fuzzy-set algorithms, genetic algorithms, neural networks, geostatistics, signal processing, fractals and chaos. His research has been funded by the Australian Research Council, academic institutions, and industry. Dr. Pham is an editorial board member of several journals and book series including Pattern Recognition (Elsevier), Current Bioinformatics (Bentham), Recent Patents on Computer Science (Bentham), Proteomics Insights (open access journal, Libertas Academica Press), Book Series on Bioinformatics and Computational BioImaging (Artech House), International Journal of Computer Aided Engineering and Technology (Inderscience Publishers). He has been serving as chair and technical committee member of many international conferences in the fields of image processing, pattern recognition, computational intelligence, and computational life sciences.

About the AuthorXIAOBO ZHOU received his B.S. in Mathematics from Lanzhou University, China, in 1988, M.S. and Ph.D. in Mathematics from Peking University, Beijing, China, in 1995 and 1998. He was an Assistant Professor of Radiology, BWH and a research member of Harvard Center for Neurodegeneration and Repair—Center for Bioinformatics. He has worked in a number of areas including wavelet analysis, statistical signal processing, pattern recognition, remote sensing, wireless communications, life science data mining, bioinformatics, computational biology, systems biology, and molecular and cellular imaging analysis. His current research interests include bioinformatics and signal processing for high-content molecular and cellular imaging for drug development and for diagnosis and therapy, modeling and integration for multiscale and multimodality phenotype data, neuroinformatics, and bioinformatics for genomics and proteomics. Dr. Zhou is now an Associate Professor with the Methodist Hospital Research Institute, the Methodist Hospital-Weill Cornell Medical College.

About the Author—STEPHEN WONG is Vice Chairman of Radiology Department and chief of Medical Physics as well as director of the Bioinformatics Program and senior member at the Methodist Hospital Research Institute, the Methodist Hospital-Weill Cornell Medical College. His research focuses on creation and application of advanced imaging and bioinformatics techniques to solve complex biomedical problems. Dr. Wong published over 250 peer-reviewed papers and holds seven patents in biomedical technology. He serves on several NIH and NSF panels, non-profit organization boards, numerous journal editorial boards, and conference program committees.

View full text