Automated method and system for nuclear analysis of biopsy images

A method for automatically analyzing a subject biopsy to determine whether the subject biopsy exhibits a disease, the method comprising the steps of:

• finding candidate nuclei outlines in an image of a training biopsy;
• generating an outline image of the found candidate nuclei outlines;
• overlaying the outline image of the candidate nuclei outlines over the same areas of the first image of the training biopsy where the nuclei outlines were found, on an interactive graphical user interface;
• labeling a first group of nuclei in the image of the training biopsy with a first label using ones of the candidate nuclei outlines in the outline image having a first visual characteristic, each of the nuclei in the first group being overlaid by a corresponding one of the candidate nuclei outlines having the first visual characteristic;
• labeling a second group of nuclei in the image of the training biopsy with a second label using ones of the candidate nuclei outlines in the outline image having a second visual characteristic, at least one of the nuclei in the second group being overlaid by a corresponding one of the candidate nuclei outlines having the second visual characteristic;
• training a classifier using features of the nuclei labeled with the first and second labels, the features capturing at least one of an edge map response along the outline of the labeled nuclei, a texture within the outline of the labeled nuclei, and a geometry of the outline of the labeled nuclei;
• automatically filtering nuclei outlines obtained from the image of the subject biopsy using the classifier;
• and
• determining from the filtered outlines whether the biopsy exhibits the disease.

A system for automatically analyzing a subject biopsy to determine whether the subject biopsy exhibits a disease, the system comprising:

• a nuclei finder for finding candidate nuclei outlines in an image of a training biopsy and generating an outline image of the found candidate nuclei outlines;
• a graphical user interface for:
• overlaying the outline image of the candidate nuclei outlines over the image of the training biopsy where the nuclei outlines were found;
• labeling a first group of nuclei in the image of the training biopsy with a first label using ones of the candidate nuclei outlines in the outline image having a first visual characteristic, each of the nuclei in the first group being overlaid by a corresponding one of the candidate nuclei outlines having the first visual characteristic;
• labeling a second group of nuclei in the image of the training biopsy with a second label using ones of the candidate nuclei outlines in the outline image having a second visual characteristic, at least one of the nuclei in the second group being overlaid by a corresponding one of the candidate nuclei outlines having the second visual characteristic;
• and
• a classifier trained using features of the nuclei labeled with the first and second labels, the features capturing at least one of an edge map response along the outline of the labeled nuclei, a texture within the outline of the labeled nuclei, and a geometry of the outline of the labeled nuclei, the classifier for:
• automatically filtering nuclei outlines obtained from the image of the subject biopsy;
• and
• determining from the filtered outlines whether the biopsy exhibits the disease.

A method for training a support vector machine classifier for use in automatic biopsy classification, the method comprising the steps of:

• finding candidate nuclei in a histological image of a training biopsy;
• generating an outline image of the found candidate nuclei;
• on an interactive graphical user interface, overlaying the outline image of the candidate nuclei over the same areas of the histological image of the training biopsy where the nuclei were found;
• on the interactive graphical user interface using the overlying outline image of the candidate nuclei outlines, labeling the candidate nuclei in the image of the training biopsy having well formed nuclei outlines as good candidate nuclei;
• on the interactive graphical user interface using the overlying outline image of the candidate nuclei outlines, labeling the candidate nuclei in the image of the training biopsy having poorly formed nuclei outlines as bad candidate nuclei with the interactive graphical user interface;
• encoding each of the labeled nuclei into nuclei feature vectors that capture at least one of an edge map response along the outline of the labeled nuclei, a texture within the outline of the labeled nuclei, and a geometry of the outline of the labeled nuclei;
• training the support vector machine classifier with the nuclei feature vectors of the labeled nuclei to generate a plurality of support vectors;
• and
• storing the support vectors for use by the support vector machine classifier in the automatic classification of an image of a subject biopsy.