High resolution medical image processing for autonomous diagnostics using vision transformer and machine learning
AssigneeDigital Diagnostics Inc.
InventorEvan Cary Harvey
Life of this patent
Estimating the term…
Abstract
A disease diagnosis tool uses a vision transformer to determine a diagnosis of a disease from a high-resolution image of a body part. A disease diagnosis tool receives a high-resolution image of a body part. The disease diagnosis tool divides the high-resolution image into a plurality of tiles. For each tile, the disease diagnosis tool generates an embedding having a position encoding corresponding to the tile's position in the high-resolution image. The disease diagnosis tool inputs the embeddings for the tiles into a linear projection model whose output feeds a transformer and receives, as output from a model comprising the transformer, a diagnosis of a disease of the body part.
Claims20
A method for autonomously diagnosing a disease of a patient, the method comprising:
- receiving a high-resolution image of a body part;
- dividing the high-resolution image into a plurality of tiles;
- generating a plurality of embeddings comprising of an embedding for each tile of the plurality of tiles, each embedding having a position encoding corresponding to its tile's position in the image;
- inputting the plurality of embeddings into a linear projection model whose output feeds a transformer;
- and
- receiving, as output from a model comprising the transformer, a diagnosis of a disease of the body part.
A non-transitory computer-readable medium comprising memory with instructions encoded thereon for autonomously diagnosing a disease of a patient, the instructions, when executed by one or more processors, causing the one or more processors to perform operations, the instructions comprising instructions to:
- receive a high-resolution image of a body part;
- divide the high-resolution image into a plurality of tiles;
- generate a plurality of embeddings comprising an embedding for each tile of the plurality of tiles, each embedding having a position encoding corresponding to its tile's position in the image;
- input the plurality of embeddings into a linear projection model whose output feeds a transformer;
- and
- receive, as output from a model comprising the transformer, a diagnosis of a disease of the body part.
A method comprising:
- receiving a high-resolution image;
- dividing the high-resolution image into a plurality of tiles;
- generating a plurality of embeddings comprising an embedding for each tile of the plurality of tiles corresponding to its tile's position in the image;
- inputting the plurality of embeddings into a linear projection model whose output feeds a model;
- and
- performing a task based on a classification output by the model.
A non-transitory computer-readable medium comprising memory with instructions encoded thereon that, when executed, cause one or more processors to perform operations, the instructions comprising instructions to:
- receive a high-resolution image;
- divide the high-resolution image into a plurality of tiles;
- generate a plurality of embeddings comprising an embedding for each tile of the plurality of tiles corresponding to its tile's position in the image;
- input the plurality of embeddings into a linear projection model whose output feeds a model;
- and
- perform a task based on a classification output by the model.
Dependencies are inferred from the claim text.