Medical analytics system
Systems and methods of a medical analytics system are described herein. The medical analytics system can include a machine learning model for processing patient tissue images for either training the machine learning model or for clinical use, such as providing information for assisting a clinician with at least diagnosing a disease or condition of a patient. Implementations of the medical analytics system can further include a user interface that is configured to allow a user to interact with a patient image for assisting with diagnosing at least a part of the tissue captured in the patient image.
1. A system, comprising:
at least one processor; and
at least one memory including program code which when executed by the at least one processor provides operations comprising:
projecting a three-dimensional image of a patient tissue into a plurality of two-dimensional grayscale images;
applying at least one transformation algorithm to a first set of two-dimensional grayscale images to generate a first set of transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images of the first set of transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images, a plurality of feature values comprising a feature vector;
projecting the three-dimensional image into a two-dimensional color image;
applying at least one color transformation algorithm to a first set of two-dimensional color images to generate a first set of color-transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional color image and to each color-transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to the at least one two-dimensional color image and to each of the transformed two-dimensional color images, a plurality of color feature values comprising the feature vector;
collecting patient information;
generating, based on the collected patient information, one or more patient values comprising the feature vector;
training a machine learning model based on the feature vector and an associated diagnosis of the patient tissue, the machine learning model comprising a classifier having a weighted value assigned to each of the plurality of feature values, the plurality of color feature values, and the patient value.
2. The system of claim 1 , further comprising:
defining, based on the training, a bio-marker that identifies one or more of the color value, the patient value, and at least one of the plurality of feature values, the bio-marker being part of the trained classifier for determining a diagnosis of an undiagnosed tissue captured in an undiagnosed three-dimensional image.
3. The system of claim 2 , wherein the diagnosis includes whether the undiagnosed tissue is malignant or benign.
4. The system of claim 2 , wherein the undiagnosed tissue is lung tissue or breast tissue.
5. The system of claim 1 , wherein the three-dimensional image includes a part of a captured three-dimensional image generated by a three-dimensional imaging system.
6. The system of claim 1 , wherein the three-dimensional image is projected onto at least two different dimensional planes thereby generating at least two different two-dimensional grayscale images.
7. A computer-implemented method, comprising:
projecting a three-dimensional image of a patient tissue into a plurality of two-dimensional grayscale images;
applying at least one transformation algorithm to a first set of two-dimensional grayscale images to generate a first set of transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images of the first set of transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images, a plurality of feature values comprising a feature vector;
projecting the three-dimensional image into a two-dimensional color image;
applying at least one color transformation algorithm to a first set of two-dimensional color images to generate a first set of color-transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional color image and to each color-transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to the at least one two-dimensional color image and to each of the transformed two-dimensional color images, a plurality of color feature values comprising the feature vector;
collecting patient information;
generating, based on the collected patient information, one or more patient values comprising the feature vector;
training a machine learning model based on the feature vector and an associated diagnosis of the patient tissue, the machine learning model comprising a classifier having a weighted value assigned to each of the plurality of feature values, the plurality of color feature values, and the patient value.
8. The computer-implemented method of claim 7 , further comprising:
defining, based on the training, a bio-marker that identifies one or more of the color value, the patient value, and at least one of the plurality of feature values, the bio-marker being part of the trained classifier for determining a diagnosis of an undiagnosed tissue captured in an undiagnosed three-dimensional image.
9. The computer-implemented method of claim 8 , wherein the diagnosis includes whether the undiagnosed tissue is malignant or benign.
10. The computer-implemented method of claim 8 , wherein the undiagnosed tissue is lung tissue or breast tissue.
11. The computer-implemented method of claim 7 , wherein the three-dimensional image includes a part of a captured three-dimensional image generated by a three-dimensional imaging system.
12. The computer-implemented method of claim 7 , wherein the three-dimensional image is projected onto at least two different dimensional planes thereby generating at least two different two-dimensional grayscale images.
13. A non-transitory computer-readable storage medium including program code which when executed by at least one processor causes operations comprising:
projecting a three-dimensional image of a patient tissue into a plurality of two-dimensional grayscale images;
applying at least one transformation algorithm to a first set of two-dimensional grayscale images to generate a first set of transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images of the first set of transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to at least one two-dimensional grayscale image and to each transformed two-dimensional grayscale images, a plurality of feature values comprising a feature vector;
projecting the three-dimensional image into a two-dimensional color image;
applying at least one color transformation algorithm to a first set of two-dimensional color images to generate a first set of color-transformed two-dimensional grayscale images;
applying at least one feature algorithm to at least one two-dimensional color image and to each color-transformed two-dimensional grayscale images;
generating, based on the applying of the at least one feature algorithm to the at least one two-dimensional color image and to each of the transformed two-dimensional color images, a plurality of color feature values comprising the feature vector;
collecting patient information;
generating, based on the collected patient information, one or more patient values comprising the feature vector;
training a machine learning model based on the feature vector and an associated diagnosis of the patient tissue, the machine learning model comprising a classifier having a weighted value assigned to each of the plurality of feature values, the plurality of color feature values, and the patient value.
14. The computer-readable storage medium of claim 13 , further comprising:
defining, based on the training, a bio-marker that identifies one or more of the color value, the patient value, and at least one of the plurality of feature values, the bio-marker being part of the trained classifier for determining a diagnosis of an undiagnosed tissue captured in an undiagnosed three-dimensional image.
15. The computer-readable storage medium of claim 14 , wherein the diagnosis includes whether the undiagnosed tissue is malignant or benign.
16. The computer-readable storage medium of claim 14 , wherein the undiagnosed tissue is lung tissue or breast tissue.
17. The computer-readable storage medium of claim 13 , wherein the three-dimensional image includes a part of a captured three-dimensional image generated by a three-dimensional imaging system.
18. The computer-readable storage medium of claim 13 , wherein the three-dimensional image is projected onto at least two different dimensional planes thereby generating at least two different two-dimensional grayscale images.