Huron Digital Pathology announced today that it has been granted a US patent for barcoding digital images. The invention forms the basis for Huron’s image search platform. Related to the patent, the company also announced the recent publication of a paper in Nature Digital Medicine describing a major validation of its image search technology.
United States patent #10,628,736, “Systems and Methods For Barcode Annotations For Digital Images” describes a content-based image retrieval system and method to generate barcodes from an input digital image. The invention allows whole slide images to be represented as a bunch of compact barcodes. Once barcoded, large archives of whole slide images can be quickly and easily compared to each other, opening up new ways for pathologists to connect to the expertise of their colleagues and the rich diagnostic data contained in the world’s pathology reports.
In March, Nature Digital Medicine published a paper, “Pan-Cancer Diagnostic Consensus Through Searching Archival Histopathology Images Using Artificial Intelligence.” The paper reports the results of a recent validation on 33,000 whole slides from 11,000 patients, 25 organs and 32 cancer subtypes from The Cancer Genome Atlas program (TCGA) by National Cancer Institute public dataset. The key finding of this validation study was that computational consensus appears to be possible for rendering diagnoses if a sufficiently large number of searchable cases are available for each cancer subtype. The paper was written by researchers, engineers and pathologists from Huron, Kimia Lab, Vector Institute and clinical partners.
Being able to search by the content of an image opens up amazing opportunities to advance the practice of pathology, and we expect image search to quickly become a ‘must-have’ component in digital pathology workflows. Our recent milestones are a great start for us in 2020. They give us momentum as we undertake planned clinical validations throughout the year.”
Patrick Myles, CEO, Huron Digital Pathology