Systems and methods for image pattern recognition
View Patent ↗Systems and methods for image pattern recognition comprise digital image capture and encoding using vector quantization (“VQ”) of the image. A vocabulary of vectors is built by segmenting images into kernels and creating vectors corresponding to each kernel. Images are encoded by creating a vector index file having indices that point to the vectors stored in the vocabulary. The vector index file can be used to reconstruct an image by looking up vectors stored in the vocabulary. Pattern recognition of candidate regions of images can be accomplished by correlating image vectors to a pre-trained vocabulary of vector sets comprising vectors that correlate with particular image characteristics. In virtual microscopy, the systems and methods are suitable for rare-event finding, such as detection of micrometastasis clusters, tissue identification, such as locating regions of analysis for immunohistochemical assays, and rapid screening of tissue samples, such as histology sections arranged as tissue microarrays (TMAs).
1. A computer implemented method for recognition of patterns in a digital image, comprising:
creating a vector set comprising a plurality of image characteristic vectors, the vector set corresponding to an image characteristic;
identifying a candidate region of an image, the candidate region comprising a plurality of candidate region vectors;
determining a correlation for each candidate region vector in the plurality of candidate region vectors with each image characteristic vector in the vector set; and
averaging the correlations to determine the probability that the candidate region exhibits the image characteristic.
2. The method of claim 1 , wherein the determining step further comprises dividing the number of times an image characteristic vector is observed in the candidate region by the number of times the image characteristic vector is observed overall.
3. The method of claim 1 , wherein the candidate region of an image comprises the entire image.
4. The method of claim 1 , wherein the candidate region of an image comprises the a portion of entire image.
5. The method of claim 4 , wherein the candidate region of an image comprises a portion of a microscope slide sample.
6. The method of claim 1 , wherein a high average of the correlations indicates a high probability of a match between an image pattern in the candidate region of the image and the image characteristic.
7. The method of claim 1 , wherein a low average of the correlations indicates a low probability of a match between an image pattern in the candidate region of the image and the image characteristic.
8. The method of claim 1 , further comprising creating a plurality of vector sets, wherein each vector set comprises a plurality of image characteristic vectors, the plurality of vector sets corresponding to a plurality of image characteristics.
9. The method of claim 8 , wherein the determining step further comprises determining a correlation for each candidate region vector in the plurality of candidate region vectors with each image characteristic vector in a plurality of vector sets.
10. A computer implemented method for recognition of patterns in a digital image, comprising:
receiving an identification of a region of a digital image containing a desired image characteristic;
creating a vector set for the identified region, the vector set comprising a plurality of image characteristic vectors, the vector set corresponding to the desired image characteristic;
receiving an identification of a candidate region of a digital image, the candidate region comprising a plurality of candidate region vectors;
correlating each candidate region vector in the plurality of candidate region vectors with each image characteristic vector in the vector set to provide a plurality of correlations; and
averaging the plurality of correlations to determine a probability that the candidate region includes the desired image characteristic.
11. A system for recognition of patterns in a digital image, comprising:
a vocabulary comprising a plurality of image vectors, each image vector including a plurality of values representing pixel intensities for a plurality of pixels in a subsection of a digital image;
a plurality of vector sets, each vector set including a plurality of image vectors and corresponding to a desired image characteristic;
a correlation module configured to correlate each image vector in a candidate region of a digital image with each image vector in a plurality of vector sets to determine a probability that the desired image characteristic is present in the candidate region of a digital image.