Methods for diagnosing glaucoma utilizing combinations of FD-OCT measurements from three anatomical regions of the eye

1. A machine classifier based method of diagnosing glaucoma in a subject, comprising:

obtaining OCT images from three anatomic regions of the subject's eye, wherein said three anatomic regions consists of the mGCC, the ppNFL, and the ONH;

processing the images to obtain a predetermined collection of diagnostic parameters, wherein said collection of diagnostic parameters comprises the following characteristics:

a. each parameter is classified as belonging to one of the three anatomic regions,

b. in cases where the parameter is a pattern cross-correlation parameter, it is further classified as one of a GPCC type selected from superior, inferior, and even, and

c. at least the mGCC and the ppNFL regions are represented in the collection of diagnostic parameters; and

transforming the collection of diagnostic parameters by combining them across anatomic regions and GPCC types according to a predetermined combination formula to arrive at a set of reduced global parameters; and

applying a pre-trained machine classifier to the reduced global parameters to arrive at a single indicator for the subject's glaucoma status.

2. The method of claim 1 , wherein said predetermined collection of diagnostic parameters include parameters of all three types of GPCC for each anatomic region.

3. The method of claim 1 , wherein said predetermined collection of diagnostic parameters further include parameters for the ONH region.

4. The method of claim 1 , wherein said predetermined combination formula comprises transforming the diagnostic parameters into six reduced global parameters consisting of a global GPCC, a combined-region FLV, a combined-region GLV, a combined-region PCV, a cup/disk, and a rim parameter.

5. The method of claim 1 , wherein said machine classifier is one selected from the group consisting of LDF, logistic regression, SVM, and RVM.

6. The method of claim 1 , wherein said combination formula comprises the steps of:

combining diagnostic parameters of different anatomic regions; followed by

combining diagnostic parameters of different GPCC types.

7. The method of claim 1 , wherein said combination formula comprises the steps of:

combining diagnostic parameters of different GPCC types within each anatomic region; followed by

combining diagnostic parameters of different anatomic regions.

8. The method of claim 1 further comprising a step of adjusting the indicator to account for the patient's age.

9. The method of claim 1 , wherein said OCT images are obtained by an FD-OCT device.

10. A method of training a machine classifier for glaucoma diagnosis, comprising:

providing a training dataset consisting of an initial set of diagnostic parameters obtained from a sample population of subjects, wherein said subjects consists of a first population of healthy patient and a second population of patients suffering from perimetric glaucoma, and said initial set of diagnostic parameters has the following characteristics:

a. each diagnostic parameter is derived from OCT images of an anatomic region of a subject's eye selected from the group consisting of mGCC, ppNFL, and ONH, and

b. in cases where the diagnostic parameter is a pattern cross-correlation parameter, it is further classified as being one of three GPCC types selected from superior, inferior, and even;

selecting a trial set of diagnostic parameters such that at least the mGCC and the ppNFL regions are represented;

transforming the trial set of diagnostic parameters by combining them across regions and GPCC types to arrive at a set of reduced global parameters;

training the machine classifier to distinguish between healthy and glaucomatous using the reduced global parameters as inputs; and

optimizing the machine classifier using AROC as a guide.

11. The method of claim 10 , wherein said trail set of diagnostic parameters include parameters of all three GPCC types for each anatomic region.

12. The method of claim 10 , wherein said trial set of diagnostic parameters further include parameters for the ONH region.

13. The method of claim 10 , wherein said transforming step comprises the steps of:

combining the diagnostic parameters of different anatomic regions; followed by

combining diagnostic parameters of different GPCC types.

14. The method of claim 10 , wherein said transforming step comprises the steps of:

combining diagnostic parameters of different GPCC types within each anatomic region; followed by

combining diagnostic parameters of different anatomic regions.

15. The method of claim 10 , wherein said machine classifier is one selected from the group consisting of LDF, logistic regression, SVM, and RVM.

16. The method of claim 10 , wherein said OCT images are obtained by a FD-OCT device.

17. A system for diagnosing glaucoma in a patient, comprising:

a computing unit configured for receiving a predetermined collection of diagnostic parameters, transforming the collection of diagnostic parameters according to a predetermined formula to arrive at a reduced set of global parameters, and applying a pre-trained machine classifier to the reduced global parameters to arrive at a single indicator; and

an input/output unit operatively connected to the computing unit for receiving input from an end user and outputting the indicator to the end user,

wherein said collection of diagnostic parameters has the following characteristics:

a. each diagnostic parameter is derived from OCT images of an anatomic region of the patient's eye selected from the group consisting of the mGCC, the ppNFL, and the ONH,

b. in cases where the parameter is a pattern cross-correlation parameter, it is further classified as one of three GPCC types consisting of superior, inferior, and even; and

c. at least the mGCC and the ppNFL regions are represented in the collection of diagnostic parameters.

18. The system of claim 17 , wherein said collection of diagnostic parameters include all three types of GPCC for each anatomic region.

19. The system of claim 17 , wherein said collection of diagnostic parameters further include parameters for the ONH region.

20. The system of claim 17 , further comprising a OCT device for obtaining tomographic images of the patient's eye.

21. The system of claim 20 , wherein said OCT device is a FD-OCT.