Image analysis – Applications – Biomedical applications
Reexamination Certificate
1999-11-19
2001-07-10
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06259807
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates generally to image processing, and more particularly to identifying objects of interest in a sample.
It is known to perform image processing on images of biological samples where different dyes are caused to reside on different portions of the sample, each dye residing on a particular feature that characterizes an object of interest. The sample is then illuminated and imaged in a manner that the different features of interest can be distinguished from each other and from the background. This typically entails acquiring separate images using the appropriate light sources, filters, and optical setup so that each image's particular type of feature appears recognizably.
It is also known that certain brightfield dyes, while nominally assumed to stain different portions of the sample, are not perfectly selective, and stain both features of interest and features not of interest. For example, a dye that is considered to stain nuclei will often also, to a lesser extent, stain cytoplasmic structures. Further, many of the dyes in common use are characterized by a broad absorption spectrum, and therefore objects stained with the dye may show up in images acquired with a different illumination scheme.
SUMMARY OF THE INVENTION
The present invention provides robust and efficient techniques for analyzing samples to find objects of interest that are interspersed with other objects.
In short, a sample is prepared to impart optical properties to objects so that objects of interest, when imaged on an imaging medium under a plurality of different illumination schemes, exhibit a combination of features that is different from combinations of features exhibited by other objects. This combination is therefore referred to as the unique combination. The objects of interest arc found by analyzing images arising from the respective illumination schemes to determine instances where the unique combination of features from the different images meets a predetermined proximity constraint (e.g., overlap or near-overlap).
The illumination schemes and the optical properties are preferably such that the sample is imaged on the imaging medium under the plurality of different illumination schemes without having to move any optical elements into or out of the path between the sample and the imaging medium. Thus, images corresponding to the illumination schemes are largely immune to registration problems, thereby making the overlap of the unique combination of features a reliable indication of an object of interest.
The images can be separately acquired images, each taken with a respective one of the illumination schemes, or can be derived from a lesser number of images, each taken with a combination of the illumination schemes in effect simultaneously. Moreover, multiple images, corresponding to different illumination schemes, can be combined to form a pre-combined image. This pre-combined image can then be treated as one of the images whose features are input to the analysis to determine the proximity constraint.
In a specific example: the objects of interest are fetal nucleated red blood cells (NRBCs) and the other objects include non-nucleated red blood cells (RBCs) and nucleated white blood cells (WBCs); the objects in the sample are stained with a fluorescent dye that selectively stains nuclei and a dye that selectively stains fetal hemoglobin in the cytoplasm of fetal RBCs; there are two different illumination schemes, namely UV excitation to provide fluorescent emissions from the stained cell nuclei and brightfield transmission of light that is absorbed by the stained cytoplasm; and the unique combination of features is the fluorescent emissions by cell nuclei in response to the UV excitation and the absorption by fetal hemoglobin of the brightfield illumination.
In a specific example of a pre-combined image: the objects of interest are micrometastatic cells in bone marrow and the other objects include normal cells; the objects in the sample are stained with a dye that selectively stains nuclei and a dye that selectively stains cytoplasms of cells expressing cytokeratin; the illumination schemes include two brightfield transmissions using two different color filters; and the pre-combined image is a linear combination of the two images. In a more specific instance of this example: the dyes are hematoxylin, which stains nuclei, and new fuschin, which stains cytoplasms of cells positive for cytokeratin; the two illumination schemes are broadband illumination through a red filter and a green filter.
In specific embodiments, the image that is expected to contain the fewest features is analyzed first to determine candidate regions of interest for subsequent processing. This is accomplished by finding regions in the image that contain the feature that objects of interest (as well as other objects) exhibit in the image, and then examining the corresponding regions in the images to determine whether or not the images all contain instances of the respective features in sufficient proximity to denote the presence of an object of interest.
In specific embodiments, the images (or preferably candidate regions thereof) are processed to generate respective contrast masks, which are combined and further processed to provide the locations of objects of interest in the images. The combination can include a logical AND operation between the masks, possibly themselves morphologically dilated to form a seed image. The further processing can include reconstructing the seed within the masks to provide the desired regions representing the features of interest where they denote the presence of an object of interest.
A further understanding of the nature and advantages of the present invention may be realized by reference to the remaining portions of the specification and the drawings.
REFERENCES:
patent: 3999047 (1976-12-01), Green
patent: 4125828 (1978-11-01), Resnick et al.
patent: 5093866 (1992-03-01), Douglas-Hamilton et al.
patent: 5162990 (1992-11-01), Odeyale et al.
patent: 5294799 (1994-03-01), Aslund et al.
patent: 5432054 (1995-07-01), Saunders et al.
patent: 5629149 (1997-05-01), Santamaria et al.
patent: 5662813 (1997-09-01), Sammons et al.
patent: 5764792 (1998-06-01), Kennealy
patent: 5989835 (1999-11-01), Dunlay et al.
patent: 0732 582 A2 (1996-01-01), None
patent: WO 94/02830 (1994-02-01), None
patent: WO 97/05562 (1997-02-01), None
patent: WO 97/26617 (1997-07-01), None
Bianchi, Diana W. et al., “PCR quantitation of fetal cells in maternal blood in normal and aneuploid pregnancies,”Am.J.Hum.Genet.61:822-829 (1997).
Ravkin, Ilya et al., “Automated microscopy system for detection and genetic characterization of fetal nucleated red blood cells on slides,”SPIE, Reprinted from: Proceedings of Optical Investigation of Cells In Vitro and In Vivo, vol. 3260, pp. 180-191 (Jan. 25-28, 1998, San Jose, CA).
Ravkin, Ilya et al., “Automatic counting of FISH spots in interphase cells for prenatal characterization of aneuploidies,”SPIE, Reprinted from: Proceedings of Optical Diagnostics of Living Cells II, vol. 3604, pp. 208-217 (Jan. 25-26, 1999, San Jose, CA).
Applied Imaging Corp.
Johns Andrew W.
Nakhjavan Shervin
Townsend and Townsend / and Crew LLP
LandOfFree
Identification of objects of interest using multiple... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Identification of objects of interest using multiple..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Identification of objects of interest using multiple... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2481713