BIOMEDICAL IMAGING GROUP (BIG)
Laboratoire d'imagerie biomédicale (LIB)

Whole Image-Based Classification Using CellProfiler

V. Uhlmann, S. Singh, C. Wählby, J. McKinney, A. Carpenter

Proceedings of BioImage Informatics 2012 (BII'12), Dresden, Federal Republic of Germany, September 16-19, 2012, pp. 59–60.

Much research in biology and medicine relies on information contained in images. Due to the arge number and size of the images acquired in these experiments, automated image classification methods are a critical tool for data analysis. Most of these methods rely on features extracted from specific regions of interest in the image such as the extent of individual cells. Deviating from this approach, Goldberg et. al. [1,2] proposed a new paradigm in which features are computed on the whole image, thereby circumventing the problem of image segmentation. Their whole image-based classification algorithm, WND-CHARM, has been demonstrated on a range of different datasets. The results from their experiments were very promising, though the algorithms were not adapted for non-expert use.

We therefore aimed to create and validate a WND-CHARM-inspired algorithm where feature extraction is carried out in the widespread and user-friendly open-source image analysis software CellProfiler [3]. We have aimed to keep the core idea of WND-CHARM while making it easier to adapt for a range of cell-biology applications. Our new implementation was benchmarked against WND-CHARM using a collection of reference datasets that were proposed previously, and was shown to exhibit competitive results. The algorithms were also successfully tested on typical examples of high-throughput cell-based assay data from the Broad Bioimage Benchmark Collection (http://broad.io/bbbc) and yielded promising results. Finally, we used our algorithm to automatically classify immunostaining patterns from selected tissue images of the Human Protein Atlas [4] and obtained classification accuracy competitive with other published results, indicating its eficacy in real-world biological datasets.

Similar to WND-CHARM, the proposed algorithm has two key advantages over traditional approaches of classifying biological images: (i) it captures several different morphological aspects of the image, making the method generalizable across a wide range of applications, and (ii) it obviates the need for image segmentation. In nearly all the examples considered, the proposed method was shown to perform as well as its predecessor, but with an improved user-friendliness as the feature-extraction process is designed as an editable CellProfiler pipeline. By incorporating the algorithm into CellProfiler, we aim to make whole-image based classification more accessible to the biological and medical research community.

References

  1. N. Orlov, L. Shamir, T. Macura, J. Johnston, D.M. Eckley, I.G. Goldberg, "WND-CHARM: Multi-Purpose Image Classification Using Compound Image Transforms," Pattern Recognition Letters, vol. 29, no. 11, pp. 1684-1693, August 2008.

  2. L. Shamir, N. Orlov, D.M. Eckley, T. Macura, J. Johnston, I. G Goldberg, "Wndchrm—An Open Source Utility for Biological Image Analysis, " Source Code for Biology and Medicine, vol. 3, no. 13, July 8, 2008.

  3. A.E: Carpenter, T.R. Jones, M.R. Lamprecht, C. Clarke, I.H. Kang, O. Friman, D.A. Guertin, J.H. Chang, R.A. Lindquist, J. Moffat, P. Golland, D.M. Sabatini, "CellProfiler: Image Analysis Software for Identifying and Quantifying Cell Phenotypes, " Genome Biology, vol. 7, no. 10, paper no. R100, October 31, 2006.

  4. M. Uhlen, P. Oksvold, L. Fagerberg, E. Lundberg, K. Jonasson, M. Forsberg, M. Zwahlen, C. Kampf, K. Wester, S. Hober, H. Wernerus, L. Björling, F. Ponten, "Towards a Knowledge-Based Human Protein Atlas, " nature biotechnology, vol. 28, no. 12, pp. 1248-1250, December 2010. 

@INPROCEEDINGS(http://bigwww.epfl.ch/publications/uhlmann1201.html,
AUTHOR="Uhlmann, V. and Singh, S. and W{\"{a}}hlby, C. and McKinney, J.
	and Carpenter, A.",
TITLE="Whole Image-Based Classification Using {CellProfiler}",
BOOKTITLE="Proceedings of BioImage Informatics 2012 ({BI'12})",
YEAR="2012",
editor="",
volume="",
series="",
pages="59--60",
address="Dresden, Federal Republic of Germany",
month="September 16-19,",
organization="",
publisher="",
note="")
© 2012 CBG. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from CBG. This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.