Automated method and system for advanced non-parametric classification of medical images and lesions
A computer-aided diagnosis (CAD) scheme to aid in the detection, characterization, diagnosis, and/or assessment of normal and diseased states (including lesions and/or images). The scheme employs lesion features for characterizing the lesion and includes non-parametric classification, to aid in the development of CAD methods in a limited database scenario to distinguish between malignant and benign lesions. The non-parametric classification is robust to kernel size.
1. A method of analyzing a medical image to determine information concerning a disease that may be evidenced by a lesion in the medical image, the method comprising:
extracting data corresponding to at least one feature of the lesion from the medical image; and
determining the information concerning the disease, based on non-parametric smoothing of the extracted data over a database of previously stored feature data with one of a fixed or adaptive kernel, K, the adaptive kernel being wider in a region where the extracted data are more sparse, narrower in a region where the extracted data are more dense.
2. The method of claim 1 , wherein the information comprises at least one from a group including:
a decision on whether a lesion is present in the medical image;
a characterization of a likelihood that the lesion is malignant;
a characterization of a stage of cancer of the lesion;
a characterization of the lesion as being malignant or benign; and
a characterization of a likelihood that a malignancy will develop in the future.
3. The method of claim 1 , wherein the extracting data step comprises:
analyzing a surrounding environment of the lesion.
4. The method of claim 3 , wherein the analyzing step comprises:
assessing a parenchymal pattern surrounding the lesion in human breast tissue in a mammogram constituting the medical image.
5. The method of claim 1 , wherein the extracting data step comprises:
determining at least one feature from a group of features comprising:
skewness of gray-values,
spiculation,
margin definition,
shape,
density,
homogeneity,
texture,
asymmetry, and
temporal stability.
6. The method of claim 1 , where K is a paraboloid, Gaussian, or Lorentzian kernel.
7. The method of claim 1 , wherein the information comprises an estimate of a probability density function (PDF) of a distribution of the at least one lesion feature over the database, and the PDF is calculated by the mathematical equation
PDF( {right arrow over (x)} )=Σ i K ( {right arrow over (x)}−{right arrow over (x)} i )
where {right arrow over (x)} represents the extracted data, and {right arrow over (x)} i represents previously stored feature data.
8. A system, comprising:
a data extraction device configured to extract data corresponding to at least one feature of the lesion from a medical image; and
a processor configured to determine the information concerning the disease, based on non-parametric smoothing of the extracted data over a database of previously stored feature data with one of a fixed or adaptive kernel, K, the adaptive kernel being wider in a region where the extracted data are more sparse, narrower in a region where the extracted data are more dense.
9. A computer readable storage medium containing instructions configured to cause a computing device to execute a method comprising:
extracting data corresponding to at least one feature of the lesion from the medical image; and
determining the information concerning the disease, based on non-parametric smoothing of the extracted data over a database of previously stored feature data with one of a fixed or adaptive kernel, K, the adaptive kernel being wider in a region where the extracted data are more sparse, narrower in a region where the extracted data are more dense.