Diagnosis of a disease condition using an automated diagnostic model
A method of identifying an object of interest can comprise obtaining first samples of an intensity distribution of one or more object of interest, obtaining second samples of an intensity distribution of confounder objects, transforming the first and second samples into an appropriate first space, performing dimension reduction on the transformed first and second samples, whereby the dimension reduction of the transformed first and second samples generates an object detector, transforming one or more of the digital images into the first space, performing dimension reduction on the transformed digital images, whereby the dimension reduction of the transformed digital images generates one or more reduced images, classifying one or more pixels of the one or more reduced images based on a comparison with the object detector, and identifying one or more objects of interest from the classified pixels.
1. A system for diagnosing a disease condition in a patient, the system comprising:
one or more computer processors on one or more computers;
a computer readable medium on the one or more computers;
an object detection model component stored on the computer readable medium, the object detection model component comprising a set of parameters of a supervised procedure that is trained to:
receive an input image, and
output a probability that the input image contains an object of interest at each of one or more locations within the input image;
an automated diagnostic model component stored on the computer readable medium, the automated diagnostic model component comprising a set of parameters of a diagnostic model that is trained to:
receive the probability that the input image contains the object of interest at each of the one or more locations within the input image, and
output a probabilistic diagnosis of a disease condition depicted in a input image; and
a diagnostic system component stored on the computer readable medium, the diagnostic system component containing instructions configured to cause the one or more computer processors to perform steps comprising:
receiving the input image of the portion of the body of the patient,
passing the input image to the object detection model component,
receiving, from the object detection model component, the probability that the input image contains the object of interest at each of the one or more locations within the input image,
passing, to the automated diagnostic model component, the probability that the input image contains the object of interest at each of the one or more locations within the input image,
receiving, from the automated diagnostic model component, the probabilistic diagnosis of the disease condition for the input image, and
outputting the probabilistic diagnosis of the disease condition.
2. The system of claim 1 , wherein passing the input image to the object detection model component comprises:
obtaining one or more samples for each of a set of different regions of the input image; and
applying each of the obtained samples of the input image to the supervised procedure to output a probability that the sample contains one or more objects of interest, where the one or more objects of interest are indicative of the disease condition.
3. The system of claim 1 , wherein the supervised procedure comprises a plurality of trained filters, each trained filter configured to identify the probability that the input image contains the object of interest.
4. The system of claim 1 , wherein the input image is of a portion of an eye of the patient, and wherein the disease condition comprises a disorder manifesting in the eye.
5. The system of claim 1 , wherein the input image is of a portion of an eye of the patient, and wherein the disease condition comprises a disorder manifesting in a retina of the patient.
6. The system of claim 1 , wherein the one or more objects of interest comprise an object that has a similar appearance to another object that indicates a disease condition in a patient but that does not itself indicate a disease condition in a patient.
7. The system of claim 1 , wherein the object of interest comprises an object selected from a group consisting of: a microaneurysm, a dot hemorrhage, a flame-shaped hemorrhage, a sub-intimal hemorrhage, a sub-retinal hemorrhage, a pre-retinal hemorrhage, a micro-infarction, a cotton-wool spot, a yellow exudate, and a drusen.
8. The system of claim 1 , wherein the supervised procedure is trained by a method comprising:
receiving a plurality of training examples, each training example comprising a training image and a label indicating whether the training image contains an object of interest at each of one or more locations within the training image; and
fitting the supervised procedure to the plurality of training examples.
9. The system of claim 8 , wherein each of the training images is annotated to indicate one or more reference points that correspond to a center of a feature of interest in the training image.
10. The system of claim 1 , wherein the diagnostic model is trained by a method comprising:
receiving a plurality of training examples, each training example comprising a probability that a training input image contains an object of interest at each of one or more locations within the training input image and a label indicating whether the training input image depicts a disease condition; and
fitting the diagnostic model to the plurality of training examples.
11. A method for diagnosing a disease condition in a patient, the method comprising:
receiving an input image of a portion of a body of a patient;
passing the input image to an object detection model that is trained to generate an output that comprises, for each of a set of reference locations in the input image, a probability that the input image contains an object of interest at the reference location, where one or more of the objects of interest are indicative of the disease condition;
passing the output from the object detection model to an automated diagnostic model that is trained to provide a probabilistic diagnosis of the disease condition for the input image based on the output from the object detection model; and
outputting the probabilistic diagnosis of the disease condition from the automated diagnostic model.
12. The method of claim 11 , wherein passing the input image to an object detection model comprises:
obtaining one or more samples for each of a set of different regions of the input image; and
applying each of the obtained samples of the input image to the object detection model to output a probability that the sample contains one or more objects of interest, where the object of interest is indicative of the disease condition.
13. The method of claim 11 , wherein the object detection model comprises a plurality of trained filters, each trained filter configured to identify a probability that the input image contains an object of interest.
14. The method of claim 11 , wherein the input image is of a portion of an eye of the patient, and wherein the disease condition comprises a disorder manifesting in the eye.
15. The method of claim 11 , wherein the input image is of a portion of an eye of the patient, and wherein the disease condition comprises a disorder manifesting in a retina of the patient.
16. The method of claim 11 , wherein the one or more objects of interest comprise an object that has a similar appearance to another object that indicates a disease condition in a patient but that does not itself indicate a disease condition in a patient.
17. The method of claim 11 , wherein the one or more objects of interest comprise an object selected from a group consisting of: a microaneurysm, a dot hemorrhage, a flame-shaped hemorrhage, a sub-intimal hemorrhage, a sub-retinal hemorrhage, a pre-retinal hemorrhage, a micro-infarction, a cotton-wool spot, a yellow exudate, and a drusen.
18. The method of claim 11 , wherein the object detection model is trained by a method comprising:
receiving a plurality of training examples, each training example comprising a training image and a label indicating whether the training image contains an object of interest at each of one or more locations within the training image; and
fitting the object detection model to the plurality of training examples.
19. The method of claim 18 , wherein each of the training images is annotated to indicate one or more reference points that correspond to a center of a feature of interest in the training image.
20. The method of claim 11 , wherein the automated diagnostic model is trained by a method comprising:
receiving a plurality of training examples, each training example comprising a probability that a training input image contains an object of interest at each of one or more locations within the training input image and a label indicating whether the training input image depicts a disease condition; and
fitting the automated diagnostic model to the plurality of training examples.