Systems and methods to process electronic images to determine salient information in digital pathology
Systems and methods are disclosed for identifying a diagnostic feature of a digitized pathology image, including receiving one or more digitized images of a pathology specimen, and medical metadata comprising at least one of image metadata, specimen metadata, clinical information, and/or patient information, applying a machine learning model to predict a plurality of relevant diagnostic features based on medical metadata, the machine learning model having been developed using an archive of processed images and prospective patient data, and determining at least one relevant diagnostic feature of the relevant diagnostic features for output to a display.
1. A computer-implemented method for identifying a diagnostic feature of a digitized pathology image, the method comprising:
receiving one or more digitized images of a pathology specimen, and medical metadata comprising image metadata, specimen metadata, clinical information, and patient information;
applying a machine learning model to generate one or more predictions based on a presence of one or more pathological conditions in the one or more digitized images, the one or more predictions comprising a binary output to indicate a plurality of relevant diagnostic features based on medical metadata, the machine learning model having been developed using an archive of processed images and prospective patient data comprising at least one of a tissue type, a specimen type, and a stain type;
generating, by the machine learning model, at least one relevant diagnostic feature of the relevant diagnostic features for output to a display, the at least one relevant diagnostic feature being based on the presence of a region having the one or more pathological conditions beyond a statistical likelihood, the pathological conditions comprising biomarkers, cancer, grade of cancer, non-cancerous features, nuclear features, and/or cell count; and
providing, by the machine learning model, the at least one relevant diagnostic feature to the display as a region of interest indicated by at least one of an outline comprising a non-geometric shape.
2. The computer-implemented method of claim 1 , wherein the generating at least one relevant diagnostic feature further comprises:
determining a probability of diagnostic relevance for each of the plurality of relevant diagnostic features; and
determining a highest probability among the plurality of relevant diagnostic features for output to the display.
3. The computer-implemented method of claim 1 , wherein the generating at least one relevant diagnostic feature further comprises:
determining at least one probability of diagnostic relevance that exceeds a predetermined value among the plurality of relevant diagnostic features for output to the display.
4. The computer-implemented method of claim 1 , wherein the generating at least one relevant diagnostic feature further comprises:
determining a predetermined number of highest probabilities of diagnostic relevance among the plurality of relevant diagnostic features for output to the display.
5. The computer-implemented method of claim 1 , further comprising:
determining a ranking of probabilities of diagnostic relevance of each of the plurality of relevant diagnostic features;
automatically focusing the display on each of the relevant diagnostic features in order based upon the determining of the ranking of probabilities of diagnostic relevance; and
receiving a user input to modify a focus of the display from a first relevant diagnostic feature to a second relevant diagnostic feature.
6. The computer-implemented method of claim 1 , wherein at least one field of interest is indicated on a digitized pathology image.
7. The computer-implemented method of claim 1 , wherein the method further comprises storing a collection of data into a digital storage device.
8. The computer-implemented method of claim 1 , wherein the method further comprises generating a probability for biomarkers, cancer, and/or histological features on all points of a whole slide image.
9. The computer-implemented method of claim 1 , wherein the method further comprises identifying a set of relevant areas where the one or more pathological conditions is present in a whole slide image.
10. The computer-implemented method of claim 9 , wherein the method further comprises computing an overall score for each pathological condition.
11. A system for identifying a diagnostic feature of a digitized pathology image, the system comprising:
at least one memory storing instructions; and
at least one processor configured to execute the instructions to perform operations comprising:
receiving one or more digitized images of a pathology specimen, and medical metadata comprising at least one of image metadata, specimen metadata, clinical information and/or patient information;
applying a machine learning model to generate one or more predictions based on a presence of one or more pathological conditions in the one or more digitized images, the one or more predictions comprising a binary output to indicate a plurality of relevant diagnostic features based on medical metadata, the machine learning model having been developed using an archive of processed images and prospective patient data;
generating, by the machine learning model, at least one relevant diagnostic feature of the relevant diagnostic features for output to a display, the at least one relevant diagnostic feature being based on the presence of a region having the one or more pathological conditions beyond a statistical likelihood, the pathological conditions comprising biomarkers, cancer, grade of cancer, non-cancerous features, nuclear features, and/or cell count; and
providing, by the machine learning model, the at least one relevant diagnostic feature to the display as a region of interest indicated by at least one of an outline comprising a non-geometric shape.
12. The system of claim 11 , wherein the generating at least one relevant diagnostic feature further comprises:
determining a probability of diagnostic relevance for each of the plurality of relevant diagnostic features; and
determining a highest probability among the plurality of relevant diagnostic features for output to the display.
13. The system of claim 11 , wherein the generating of at least one relevant diagnostic feature further comprises:
determining at least one probability of diagnostic relevance that exceeds a predetermined value among the plurality of relevant diagnostic features for output to the display.
14. The system of claim 11 , wherein the generating at least one relevant feature further comprises:
determining a predetermined number of highest probabilities of diagnostic relevance among the plurality of relevant diagnostic features for output to the display.
15. The system of claim 11 , further comprising:
determining a ranking of probabilities of diagnostic relevance of each of the plurality of relevant diagnostic features;
automatically focusing the display on each of the relevant diagnostic features in order based upon the determining of the ranking of probabilities of diagnostic relevance; and
receiving a user input to modify a focus of the display from a first relevant diagnostic feature to a second relevant diagnostic feature.
16. The system of claim 11 , wherein at least one field of interest is indicated on a digitized pathology image.
17. A non-transitory computer-readable medium storing instructions that, when executed by a processor to perform a method for identifying a diagnostic feature of a digitized pathology image, the method comprising:
at least one memory storing instructions; and
at least one processor configured to execute the instructions to perform operations comprising:
receiving one or more digitized images of a pathology specimen, and medical metadata comprising at least one of image metadata, specimen metadata, clinical information and/or patient information;
applying a machine learning model to generate one or more predictions based on a presence of one or more pathological conditions in the one or more digitized images, the one or more predictions comprising a binary output to indicate a plurality of relevant diagnostic features based on medical metadata, the machine learning model having been developed using an archive of processed images and prospective patient data;
generating by the machine learning model, at least one relevant diagnostic feature of the relevant diagnostic features for output to a display, the at least one relevant diagnostic feature being based on the presence of a region having the one or more pathological conditions beyond a statistical likelihood, the pathological conditions comprising biomarkers, cancer, grade of cancer, non-cancerous features, nuclear features, and/or cell count; and
providing, by the machine learning model, the at least one relevant diagnostic feature to the display as a region of interest indicated by at least one of an outline comprising a non-geometric shape.