User Ratings

Easy to Follow Achieved Expected Results Overall Rating Add your comments

Detection and Location of Hybridization Domains on Chromosomes by Image Cytometry

Laura Mascio¹

¹Lawrence Livermore National Laboratory, Livermore, California

Publication Name:

Current Protocols in Cytometry

Unit Number:

Unit 10.9

DOI:

10.1002/0471142956.cy1009s03

Online Posting Date:

May, 2001

GO TO THE FULL TEXT:

PDF or HTML at Wiley Online Library

Are you the author of this protocol? Login or register and return to this page.

Abstract

Because of time considerations in performing image analysis for detection of very small amounts of fluorescence probe, there are advantages in using automated detection systems. This unit provides considerable detail on several algorithms used for location of hybridization domains that reduce the tedium and improve objectivity and repeatability. Not only are these algorithms described and discussed in detail, but sources for the programs are also provided. Here is a carefully crafted textual and graphical description of how to go about learning the art of automated image analysis.

GO TO THE FULL PROTOCOL:

PDF or HTML at Wiley Online Library

Unit Introduction
Basic Protocol: Using Custom Algorithms to Detect and Localize Hybridization Domains in Metaphase Chromosomes
Commentary
Literature Cited
Figures

GO TO THE FULL PROTOCOL:

PDF or HTML at Wiley Online Library

Materials

Basic Protocol: Using Custom Algorithms to Detect and Localize Hybridization Domains in Metaphase Chromosomes

Materials

DNA sample counterstained with 4¢,6-diamidino-2-phenylindole (DAPI) or propidium iodide (PI) and hybridized with probes of interest (units 8.1-8.3)
Fluorescence microscope with filter wheel containing single-bandpass filters at the excitation source and corresponding triple- (or double-) bandpass emission filters
High-resolution CCD or other digital camera
Computer for controlling image acquisition and performing image analysis
Software for image processing and analysis (e.g., Khoros from http://www.khoral.com, NIH image from http://rsb.info.nih.gov/NIH-image/download.html, or SCIL-Image from http://www.tno.nl/instit/tpd/product/scil/; or see references in protocol steps for instructions on generating the appropriate algorithms)

Looking for materials? Suppliers Guide

GO TO THE FULL PROTOCOL:

PDF or HTML at Wiley Online Library

Figures

Figure 10.9.1

Progression of steps toward FLpter determination. (A) The DNA counterstain image. (B) The FITC emission image. (C) DRT segmentation of FITC hybridization domains (spots inside white patches). (D) Texas red (TR) emission image. (E) DRT segmentation of TR hybridization domains. (F) Superposition of hybridization domain segmentations (from C and E) onto the chromosome, along with the results from medial axis determination. Figure (from Mascio et al., 1995) used by permission of the University of California, Lawrence Livermore National Laboratory, and the U.S. Department of Energy.

View Image
Figure 10.9.2

Histogram of a sample image. With the isodata threshold method, a marker starts at the mid-intensity level and then is adjusted until it converges to a position that is midway between the centers of mass of the two parts into which it divides the histogram. Figure reprinted with permission from Mascio et al. (1995).

View Image
Figure 10.9.3

Segmenting the chromosome. (A) An edge-preserving, smoothing filter is applied to the original image and then two thresholds are applied independently. (B, C) The isodata method yields a rough estimate for the location of the border. (D) The second-derivative method yields threads that follow intensity gradients. (E) A union of these two produces only the threads that lie within the rough border. The result is used as a mask to extract corresponding intensity values from the original image. The average of these intensities is then used as a single-value threshold on the original DNA counterstain image. Figure reprinted with permission from Mascio et al. (1995).

View Image
Figure 10.9.4

Extending the chromosomal skeleton (A) to form the medial axis (J). (B) Form a line orthogonal to the end of the skeleton. (C) Determine the geometric center of the area with no backbone. (D) Extend skeleton to the geometric center. (E–I) Repeat steps (B) through (D) twice. (I) Complete a line across the tip of the chromosome. Extend the skeleton to this line. Figure reprinted with permission from Mascio et al. (1995).

View Image
Figure 10.9.5

Pictorial summary of the four major steps in the DRT algorithm. (A) The original image is divided into three regions of interest. (B) The algorithm computes a conservative threshold value (one that is likely to include all object pixels or blobs). This is followed by a step to remove spurious signals left by the conservative threshold. (C) The algorithm computes another threshold for each region, this time using a specific and strict threshold value that is likely to include only object pixels. This step leaves objects with a diameter greater than two pixels (pinnacles) and eliminates any remaining spurious noise pixels. (D) The algorithm combines the results of steps (B) and (C) with a logical “or” operator, which adds the images together to yield the hybridization domains of interest. Figure (from Mascio et al., 1995) used by permission of the University of California, Lawrence Livermore National Laboratory, and the U.S. Department of Energy.

View Image
Figure 10.9.6

Detecting regions with local contrast. The graphs shown are of intensity vs. location along a line through an object. (A) The effect of dilating and eroding the regions containing some signal. (B) The eroded result is subtracted from the dilated result so as to remove the slowly changing background signal and leave the dilated regions level with one another. (C) Choose a “noise threshold” n_t (indicated by the horizontal dotted line) such that pixels with intensity over the threshold define region masks; the masks are used to “cookie cut” regions from the original signal. This threshold defines the minimum strength of an acceptable signal after background subtraction, but the exact value of this threshold is not critical. The goal is to isolate large regions that contain within them a signal of interest. The selection of the noise threshold will only nominally affect the size of the region, and the region will be analyzed more carefully during the blob and pinnacle detection steps of the DRT algorithm (Fig. 10.9.5B–C). To complete the region definition, the region masks that result from the noise threshold are used to isolate the regions of interest in the original image. Figure reprinted with permission from Mascio et al. (1995).

View Image

Literature Cited

Literature Cited
	Dorst, L. and Smeulders, A.W.M. 1987. Length estimators for digitized contours. Comput. Graphics Image Process. 40:311-333.
	Freeman, H. 1970. Boundary encoding and processing. In Picture Processing and Psychopictorics (B.S. Lipkin and A. Rosenfeld, eds.) pp. 241-266. Academic Press, New York.
	Giardina, C.R. and Dougherty, E.R. 1998. Morphological Methods in Image and Signal Processing. PrenticeHall, Englewood Cliffs, N.J.
	Kallioniemi, A., Kallioniemi, O.-P., Mascio, L., Sudar, D., Pinkel, D., Deaven, L., and Gray, J.W. 1994. Physical mapping of chromosome 17 cosmids. Genomics 20:125-128.
	Kuwahara, M., Hachimura, K., Eiho, S., and Kinoshita, M. 1976. Processing of RI-angiocardiographic images. In Digital Processing of Biomedical Images (K. Preston and M. Onoe, eds.) pp.187-203. Plenum, New York.
	Lawrence, J.B. 1990. A fluorescence in situ hybridization approach for gene mapping and the study of nuclear organization genome analysis. Genet. Phys. Mapping 1:1-39.
	Lichter, P., Tang, C., Call, K., Hermanson, G., Evans, G.A., Housman, D., and Ward, D.C. 1990. High-resolution mapping of human chromosome 11 by in situ hybridization with cosmid clones. Science 247:64-69.
	Mascio, L.N., Verbeek, P.W., Sudar, D., Kuo, W.-L., and Gray, J.W. 1995. Semiautomated DNA probe mapping using digital imaging microscopy: I. System development. Cytometry 19:51-59.
	Pinkel, D., Landegent, J., Collins, C., Fuscoe, J., Segraves, R., Lucas, J., and Gray, J.W. 1988. Fluorescence in situ hybridization with human chromosome-specific libraries: Detection of trisomy 21 and translocations of chromosome 4. Proc. Natl. Acad. Sci. U.S.A. 85:9138-9142.
	Ridler, T.W. and Calvard, S. 1978. Picture thresholding using an iterative selection method. IEEE Trans. Systems Man Cybernet. 8:630-632.
	Sakamoto, M., Pinkel, D., Mascio, L., Sudar, D., Peters, D., Kuo, W.-L., Yamakawa, K., Nakamura, Y., Drabkin, H., Jericevic, Z., Smith, L., and Gray, J.W. 1995. Semi-automated DNA probe mapping using digital imaging microscopy. II. System performance. Cytometry 19:60-69.
	Serra, J. 1982. Image Analysis and Mathematical Morphology. Academic Press, London.
	Verbeek, P.W., Vrooman, H.A., and Van Vliet, L.J. 1988. Low-level image processing by max-min filters. Signal Process. 17:249-258.
	Verwer, B.J.H. 1988. Improved metrics in image processing applied to the Hildritch skeleton. Proceedings of the 9th International Conference on Pattern Recognition. Rome, Italy, Nov. 14-17, 1988, pp. 137-142. Computer Society Press, Washington, D.C.
	Vossepoel, A.M. and Smeulders, A.W.M. 1982. Vector code probabilities and metrication error in the representation of straight lines of finite length. Comput. Vision Graphics Image Process. 20:347-364.
Key Reference
	Mascio, L.N., Verbeek, P.W., Sudar, D., Kuo, W.-L., and Gray, J.W. 1995. See above 1995
This paper presents and discusses the algorithm described in this unit.