Computer aided detection of microcalcification clusters
View Patent ↗A method for computer-aided detection of microcalcification clusters using a digital computer obtains digital mammography data for a single view image and normalizes and filters the image data to reduce noise. A first mask is generated and applied to the image data for defining the breast structure, forming a first cropped image. A second mask is generated and applied to the image data for defining muscle structure, forming a second cropped image. An artifact mask corresponding to vascular calcifications and known imaging artifacts is generated and applied to the first and second cropped images, defining first and second artifact-masked cropped images. In a repeated sequence, portions of each artifact-masked cropped image are processed using an enhancement algorithm and reducing edge effects to obtain a set of microcalcification cluster candidates and suspected microcalcification clusters. Image processing algorithms remove false positives from the listing of microcalcification clusters and classify candidate microcalcification clusters to identify true positives.
1. A method for removing false positive elements from mammography image data, comprising the steps of:
a) processing the image data to obtain a set of candidate microcalcifications;
b) classifying each microcalcification in the set of candidate microcalcifications using a neural network;
c) classifying each microcalcification in the set of candidate microcalcifications using a kernel based classifier;
d) combining results of steps b) and c) to remove false positives and obtain a set of classified microcalcifications;
e) grouping members of the set of classified microcalcifications to form a candidate set of microcalcification clusters;
f) classifying each microcalcification cluster in the set of candidate microcalcification clusters using a neural network;
g) classifying each microcalcification cluster in the set of candidate microcalcification clusters using a kernel based classifier; and
h) combining results of steps f) and g) to remove false positives and obtain a set of classified microcalcification clusters.
2. The method of step 1, wherein the step of classifying each microcalcification in the set of candidate microcalcifications comprises ranking each microcalcification according to a set of rules.
3. The method of step 1, wherein the step of classifying each microcalcification cluster in the set of candidate microcalcification clusters comprises ranking each microcalcification cluster according to a set of rules.
4. The method of step 1, wherein the step of classifying each microcalcification cluster in the set of candidate microcalcification clusters using a neural network further comprises training the neural network with the step of adding white noise to at least one image in the set of images used for training.
5. The method of step 1, wherein the step of classifying each microcalcification cluster in the set of candidate microcalcification clusters using a neural network further comprises training the neural network with the step of randomizing the order of the input data during successive training iterations.
6. The method of step 1, wherein the step of classifying each microcalcification cluster in the set of candidate microcalcification clusters using a neural network further comprises training the neural network with the step of normalizing input values for pixel intensities to between −1 and +1.
7. The method of step 1, wherein the step of classifying each microcalcification cluster in the set of candidate microcalcification clusters using a neural network further comprises training the neural network with the step of normalizing input values to obtain a standard deviation of 1.0.