Computer-Aided Cytogenetic Method Of Cancer Diagnosis
AssigneeNew Jersey Institute of Technology
InventorsRoman Andrushkiw, Natalya V. Boroday, Dmitriy A. Klyushin, Yuriy I. Petunin
Life of this patent
Estimating the term…
Abstract
The present invention relates to noninvasive diagnostic systems for cancer detection comprising RGB-imaging of cancer cells, buccal epithelium cells and uses of the system for drug discovery. The present invention provides novel algorithms for the detection of malignancy associated changes of buccal epithelial cells based on RGB analysis.
Claims40
A method for computer-aided diagnosis of breast cancer based on analysis of malignancy associated changes in buccal epithelium, the method comprising a first step (a) and a second step (b) wherein the first step (a) comprises:
- i) obtaining at least one training scanogram from a sample of buccal epithelium obtained from a patient with confirmed breast cancer or confirmed fibroadenomatosis;
- ii) for each training scanogram computing the ratio of model class volumes;
- iii) constructing a confidence region;
- iv) determining if a ratio of an investigated sample belongs to the confidence region, wherein if the ratio does belong, then I=1;
- and wherein the second step (b) comprises:
- i) computing a relief index;
- ii) constructing a confidence region;
- iii) such that if relief index of an investigated sample belongs to the confidence region, then J=1;
- wherein I and J are indicators;
- wherein if I=1 and J=1, then a breast cancer, else not breast cancer, and thereby determining a diagnosis of breast cancer based on the analysis.
A computer-controlled system comprising a digital imager that provides a scanogram of a cell, and an operably linked controller comprising computer-implemented programming implementing a method for computer-aided diagnosis of breast cancer based on analysis of malignancy associated changes in buccal epithelium, the method comprising a first step (a) and a second step (b) wherein the first step (b) comprises:
- a) obtaining at least one training scanogram from a sample of buccal epithelium obtained from a patient with confirmed breast cancer or confirmed fibroadenomatosis;
- b) for each training scanogram computing the ratio of model class volumes;
- c) constructing a confidence region;
- d) determining if a ratio of an investigated sample belongs to the confidence region, wherein if the ratio does belong, then I=1;
- and wherein the second step comprises:
- a) computing a relief index;
- b) constructing a confidence region;
- c) if relief index of an investigated sample belongs to the confidence region, then J=1;
- wherein I and J are indicators;
- wherein if I=1 and J=1, then a breast cancer, else not breast cancer, and
- whereby a diagnosis of breast cancer based on analysis of malignancy associated changes in buccal epithelium is determined.
A method for computer-aided breast cancer diagnosis, the method comprising the steps:
- a) obtaining a RGB-image of a scanogram from a sample of buccal epithelium obtained from a patient with confirmed breast cancer patient or confirmed fibroadenomatosis;
- b) computing 112 indexes, wherein the indexes comprise vector indexes and scalar indexes;
- c) constructing confidence ellipsoids for breast cancer and fibroadenomatosis on vector indexes;
- d) constructing confidence intervals of breast cancer and fibroadenomatosis on scalar indexes,
- wherein i) the number N of falling out of ellipsoids is computed, ii) if the number exceeds 1 then breast cancer, and iii) if (N+M for fibroadenomatosis<if N+M for breast cancer), then fibroadenomatosis;
- and
- wherein i) the number M of falling out of intervals is computed, ii) if the number exceeds 3, then breast cancer, iii) if (N+M for fibroadenomatosis≧if N+M for breast cancer), then breast cancer,
- thereby determining a diagnosis of breast cancer or fibroadenomatosis.
A computer-controlled system comprising a digital imager that provides a scanogram of a cell, and an operably linked controller comprising computer-implemented programming implementing a method for computer-aided breast cancer diagnosis, the method comprising the steps:
- a) obtaining a RGB-image of a scanogram from a sample of buccal epithelium obtained from a patient with confirmed breast cancer patient or confirmed fibroadenomatosis
- b) computing 112 indexes, wherein the indexes comprise vector indexes and scalar indexes;
- c) constructing confidence ellipsoids for breast cancer and fibroadenomatosis on vector indexes;
- d) constructing confidence intervals of breast cancer and fibroadenomatosis on scalar indexes,
- wherein i) the number N of falling out of ellipsoids is computed, ii) if the number exceeds 1 then breast cancer, and iii) if (N+M for fibroadenomatosis<if N+M for breast cancer), then fibroadenomatosis;
- and
- wherein i) the number M of falling out of intervals is computed, ii) if the number exceeds 3, then breast cancer, iii) if (N+M for fibroadenomatosis≧if N+M for breast cancer), then breast cancer,
- thereby determining a diagnosis of breast cancer or fibroadenomatosis.
A method for the differential diagnosis of breast cancer and fibroadenomatosis, the method comprising the steps:
- a) measuring scanograms of interphase nuclei of samples of buccal epithelium obtained from a patient with confirmed breast cancer patient or confirmed fibroadenomatosis;
- b) measuring scanogram indices;
- c) constructing a correlation matrix;
- d) finding numbers NBC and NFAM of falling out beyond the confidence intervals constructed for breast cancer and fibroadenomatosis, wherein BC=breast cancer and FAM=fibroadenomatosis;
- and
- e) making a diagnosis regarding the presence or absence of breast cancer or fibroadenomatosis.
A computer-controlled system comprising a digital imager that provides a scanogram of a cell, and an operably linked controller comprising computer-implemented programming implementing a method for the differential diagnosis of breast cancer and fibroadenomatosis, the method comprising the steps:
- a) measuring scanograms of interphase nuclei of samples of buccal epithelium obtained from a patient with confirmed breast cancer patient or confirmed fibroadenomatosis;
- b) measuring scanogram indices;
- c) constructing a correlation matrix;
- d) finding numbers NBC and NFAM of falling out beyond the confidence intervals constructed for breast cancer and fibroadenomatosis, wherein BC=breast cancer and FAM=fibroadenomatosis;
- and
- e) making a diagnosis regarding the presence or absence of breast cancer or fibroadenomatosis.
A method for diagnosis of breast cancer and fibroadenomatosis, the method comprising the steps:
- a) obtaining scanograms from a sample of buccal epithelium from a confirmed breast cancer patient and/or a confirmed fibroadenomatosis patient;
- b) assigning a green component and a red component for each scanogram;
- c) finding the center;
- d) constructing concentric squares;
- e) computing the average p-statistics between the squares in breast cancer training samples and fibroadenomatosis training samples;
- f) finding minimal p-statistics and maximal p-statistics,
- wherein for an investigated scanogram, compute N(P), wherein if N(P)>0, then breast cancer;
- wherein if N(P)=0, then do not make any decision;
- wherein if N(P)<0, then fibroadenomatosis;
- thereby determining a diagnosis for breast cancer or fibroadenomatosis.
A computer-controlled system comprising a digital imager that provides a scanogram of a cell, and an operably linked controller comprising computer-implemented programming implementing a method for diagnosis of breast cancer and fibroadenomatosis, the method comprising the steps:
- a) obtaining scanograms from a sample of buccal epithelium from a confirmed breast cancer patient and/or a confirmed fibroadenomatosis patient;
- b) assigning a green component and a red component for each scanogram;
- c) finding the center;
- d) constructing concentric squares;
- e) computing the average p-statistics between the squares in breast cancer training samples and fibroadenomatosis training samples;
- f) finding minimal p-statistics and maximal p-statistics,
- wherein for an investigated scanogram, compute N(P), wherein if N(P)>0, then breast cancer;
- wherein if N(P)=0, then do not make any decision;
- wherein if N(P)<0, then fibroadenomatosis;
- thereby determining a diagnosis for breast cancer or fibroadenomatosis.
Dependencies are inferred from the claim text.