Image analysis – Applications – Biomedical applications
Reexamination Certificate
2001-10-04
2004-02-24
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06697509
ABSTRACT:
BACKGROUND
In many diagnostic tests of tissue samples, the amount of a marker stain present in one or more cells may be measured semi-quantitatively. For example, several methods enable the development of custom antibodies to selected proteins. These antibodies may be linked to colored stains or stain converting enzymes to develop a color test such as an immunohistochemistry (IHC) test. In this method, technicians take a thin slice of a tumor and apply an antibody, a chemical that specifically binds to estrogen receptors. If the antibody sticks to receptors on some tumor cells, adding various other chemicals will stain those cells so that they stand out when viewed under a microscope. A technician or pathologist then visually inspects the specimen, counting the proportion of cells that stain.
While the human vision system is well adapted for detecting locations and patterns of color, humans tend to have relatively poor discrimination for the absolute intensity of a color, i.e., the brightness or dullness of a color. This is a consequence of the wide dynamic range of human vision. However, in IHC tests, for example, it is generally necessary to determine the color intensity. This is because many of the tested proteins are normally present in low quantities. Moreover, the diagnostically significant event may be an elevation of the level of the protein that translates to a darker shade of the stain color.
However, when pathologists score IHC tests, the results are generally given in a semi-quantitative scale. This scale may have a whole number score that ranges from 0 to 3. It is estimated that the average accuracy of manual reads is about ±0.8 on this scale. In many IHC tests, this may be enough uncertainty to give rise to diagnostic uncertainties. For instance, in the test for the Her2NU protein in breast cancer, a result of 0 or 1 may be an indication not to treat with Herceptin, and a score of 2 or 3 may be an indication to treat. Therefore, borderline cases may be suspect if the reading error is large enough to cause a “true” 1 to be mistaken for a 2, or visa versa.
SUMMARY
In an embodiment, an imager is used to generate an image of a sample, e.g., a tissue sample including diaminobenzidine tetrahydrochloride (DAB) stained cells. A user, e.g., a pathologist, selects a region to be scored in the image. A mean intensity value of a selected color is determined from the pixels in the selected region. The selected color may be the complement of a color to be scored. In the case of a DAB test, the complementary color is blue. A score is generated in response to the mean intensity value using a stored calibration curve.
To generate the calibration curve, mean intensity values for selected regions in a number of imaged samples may be correlated to user input scores for the same regions. To generate a user input score, the user may select the region in the image to be scored. The pixels outside of the selected region may be masked. Any pixel having a color value outside of a selected color threshold corresponding to the color to be scored may also be masked. The user may then input a score for the masked region. The mean intensity values for the same regions may then be correlated to the user input scores.
More accurate and precise scores may be generated using the mean intensity value and the calibration curve than the user input scores. For example, the user input scores may be whole number, e.g., on a scale of 0 to 3, which is indicative of the accuracy and precision available with human vision and perception. The machine generated scores may be on the same scale, but may include fractional values, e.g., a score of 2.8, due to its higher accuracy and precision.
REFERENCES:
patent: 5202931 (1993-04-01), Bacus
patent: 5420042 (1995-05-01), Schafer et al.
patent: 5846749 (1998-12-01), Slamon et al.
patent: 5981279 (1999-11-01), Weiss
patent: 6007996 (1999-12-01), McNamara et al.
patent: 6061425 (2000-05-01), Sato
patent: 6165734 (2000-12-01), Garini et al.
patent: 6183981 (2001-02-01), Gonzalez-Lima
patent: 6251586 (2001-06-01), Mulshine et al.
Chromavision Medical Systems, Inc.
Johns Andrew W.
Nakhjavan Shervin
LandOfFree
Method and apparatus for scoring the uptake of markers in cells does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for scoring the uptake of markers in cells, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for scoring the uptake of markers in cells will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3280752