Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2002-07-10
2004-07-27
Jones, Mary Beth (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
Reexamination Certificate
active
06768918
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to diagnosis of disease. More particularly, the invention relates to in vivo diagnosis by optical methods.
BACKGROUND OF THE INVENTION
Colonic polyps appear as two major types, neoplastic and non-neoplastic. Non-neoplastic polyps are benign with no direct malignant potential and do not necessarily need to be resected. Hyperplastic polyps, juvenile polyps, mucosal prolapse and normal mucosal polyps are examples of non-neoplastic polyps. Conversely, neoplastic polyps are pre-malignant, a condition requiring resection and further surveillance. Examples of premalignant neoplastic polyps are tubular adenoma, villous adenoma and tubulovillous adenoma.
Conventional laser-induced fluorescence emission and reflectance spectroscopy can distinguish between neoplastic and non-neoplastic tissue with accuracies approaching about 85%. However, typically these methods require that the full spectrum be measured with algorithms dependent on many emission wavelengths.
SUMMARY OF THE INVENTION
The invention provides in vivo diagnostic methods based upon the normalized intensity of light emitted from tissue. In particular, it is an observation of the invention that relevant diagnostic information is provided by comparing the intensities of light emitted from a tissue at two different wavelengths, both normalized over the intensity of light emitted from the same tissue at about 431 nm.
Thus, according to the invention, a comparison of the intensities of two different wavelengths normalized using the intensity at about 431 nm provides diagnostic insight. Preferred methods of the invention comprise obtaining a fluorescent emission having a first intensity at a first wavelength and a second intensity at a wavelength; normalizing the first and second intensities with respect to an intensity at a wavelength of about 431 nm to produce first and second normalized intensities; and determining a state of health of the tissue based upon a comparison of the first and second normalized intensities.
In one embodiment, methods of the invention comprise determining the state of health of the tissue using a classifier function in which the first and second normalized intensities are inputs. In one embodiment, the classifier function is a discrimination function, preferably a linear discrimination function. In other embodiments, the discrimination function is a non-linear discrimination function.
The invention can be applied to analyze a broad range of tissues. Preferably, the tissue to be analyzed is a tissue comprising epithelial cells. In one embodiment, the tissue is selected from the group consisting of cervical tissue, colonic tissue, esophogeal tissue, bladder tissue, and bronchial tissue.
Classifying or comparing normalized intensities into one or more groups may be performed by any acceptable means. There are numerous acceptable approaches to such classifications. For example, one general method of grouping the two normalized intensities is a Bayesian-based classifier using Mahalanobis distances. The Mahalanobis distance is well-known in statistical analysis, and is used to measure a distance between data in a multidimensional space based on characteristics that represent a degree of relationship among the data. Bayesian probabilities have been known in statistical analysis for many years. Specific Bayesian Mahalanobis-based classifier can be selected from linear discriminant analysis, quadratic discriminant analysis, and regularized discriminant analysis. As those familiar with statistical analysis will recognize, linear discrimination analysis and quadratic discriminant analysis are methods that are computationally efficient. Regularized discriminant analysis uses a biasing method based on two parameters to estimate class covariance matrices.
Other ways of comparing the normalized intensities include a binary tree classifier, and an unsupervised learning cluster classifier. Unsupervised learning is characterized by the absence of explicit examples showing what an input/output relation should be. Examples of an unsupervised learning cluster classifier include hierarchical clustering analysis, principal component analysis, fuzzy c-means analysis, and fuzzy k-means analysis. Each of the forgoing analytical techniques is well known in the statistical analysis literature. For example, the fuzzy c-means algorithm divides a data set having an integer number n data points into an integer number c fuzzy clusters, where n>c, while determining a location for each cluster in a multi-dimensional space.
In another aspect, the invention features systems for determining the state of health of a tissue. Systems of the invention comprise an illumination source for illuminating a tissue; a detector for receiving from the tissue light comprising a first intensity at a first wavelength and a second intensity at a second wavelength; a computational module for normalizing the first and second intensities with respect to received light having an intensity at a wavelength of about 431 nm to produce first and second normalized intensities; and an analysis module for determining a state of health of the tissue based upon a comparison of the first and second normalized intensities.
In a preferred embodiment, a system of the invention comprises an optical fiber as the illumination source. The detector may receive light from the tissue by way of a plurality of optical fibers. In a preferred embodiment, at least one of the optical fibers of the system is placed directly in contact with tissue. Preferably, the light received from the tissue is fluorescent light. The analysis module of a system of the invention may comprise a Bayesian Mahalanobis-based classifier function. The Bayesian Mahalanobis-based classifier may be selected from the group consisting of linear discriminant analysis, quadratic discriminant analysis, and regularized discriminant analysis. The analysis module may also comprise a binary tree classifier function or an unsupervised learning cluster classifier. In some embodiments, the unsupervised learning cluster classifier is selected from the group consisting of hierarchical clustering analysis, principal component analysis, fuzzy c-means analysis, and fuzzy k-means analysis.
Systems and methods of the invention are useful in examining a tissue comprising epithelial cells. A method of the invention comprises laser-induced fluorescence using light around 337 nm and a threshold classification model that depends on two fluorescence intensity ratios normalized by the intensity of fluorescence at about 431 nm.
The invention enables determining whether a polyp is neoplastic. Systems and methods of the invention enable such determination at the time of endoscopy particularly for diminutive polyps. In a preferred embodiment, the invention provides for identification of polyps (or other features) under about 10 mm in size. In a further preferred embodiment, the invention provides for identification of polyps (or other features) under about 10 mm in size in real time.
The combination of a new design of a fiberoptic probe for making measurements, an analytic method based on a small number of data points, and a simple method of obtaining a normalization factor for the data used provides enhanced diagnostic accuracy in distinguishing between neoplastic and non-neoplastic polyps.
The invention provides methods that reliably distinguish between neoplastic and non-neoplastic tissue at the time of endoscopy, colonoscopy, colposcopy, or other similar examinations. As a result, patients with non-neoplastic lesions are not subjected to the risk, discomfort and expense of biopsies or excisions. Patients with neoplastic lesions can be identified immediately and treated.
The foregoing and other objects, aspects, features, and advantages of the invention will become more apparent from the following description and from the claims.
REFERENCES:
patent: 3013467 (1961-12-01), Minsky
patent: 3632865 (1972-01-01), Haskell et al.
patent: 3809072 (1974-05-01), Ersek et al.
patent: 3890462
Jain Ruby
Jones Mary Beth
MediSpectra, Inc.
Testa Hurwitz & Thibeault LLP
LandOfFree
Fluorescent fiberoptic probe for tissue health... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Fluorescent fiberoptic probe for tissue health..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Fluorescent fiberoptic probe for tissue health... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3217534