Method and apparatus for motion correction and image enhancement for optical coherence tomography
Images of an object, such as OCT scans of a human eye, can include distortions and data gaps due to relative motion of the object and the image acquisition device. Methods and systems for correction of such distortions and data gaps are described herein. Motion-corrected data is arrived at by applying three-dimensional transforms to input three-dimensional data sets that represent at least partially overlapping regions of the imaged object. The three dimensional transforms are computed based on an objective function that accounts for similarity between the transformed three-dimensional data sets and the estimated motion of the object relative to an imaging instrument. Methods and systems described herein advantageously eliminate the need for postulated assumptions and reliance on landmarks and are capable of filling data gaps, thereby producing high quality, undistorted images of objects subject to movement during imaging. Multiple motion-corrected data sets can be merged or combined to produce a data set with improved image quality.
1. A computer-implemented method of correcting a data set representing an object for the object's motion relative to an optical coherence tomography (OCT) imaging instrument, comprising:
acquiring at least a first three-dimensional data set and a second three-dimensional data set representing overlapping regions of an object with the OCT imaging instrument, wherein the first three-dimensional data set and the second three-dimensional data set are obtained in an instrument coordinate system;
optionally, computing at least one of a first derivative three-dimensional data set corresponding to the first three-dimensional data set, a second derivative three-dimensional data set corresponding to the second three-dimensional data set, a first preprocessed three-dimensional data set corresponding to the first three-dimensional dataset, or a second preprocessed three-dimensional data set corresponding to the second three-dimensional dataset;
determining sample positions in an object coordinate system;
for at least the first and the second three-dimensional data sets, computing a first three-dimensional transform and a second three-dimensional transform respectively, wherein every transform associates a three-dimensional displacement vector with each sample position in the object coordinate system, and wherein computing the first and the second three-dimensional transforms includes:
a. for each of the first and the second three-dimensional transform, computing a first three-dimensional displacement vector for each sample position in the object coordinate system;
b. for each of the first and the second three-dimensional transform, computing, based on the first three-dimensional displacement vectors, positions in the instrument coordinate system;
c. for each of the first and the second three-dimensional transform, interpolating data of the respective three-dimensional data set or preprocessed three-dimensional data set at the positions in the instrument coordinate system, thereby producing a first set of interpolated values and a second set of interpolated values;
d. evaluating an objective function by calculating similarity between the first and the second sets of interpolated values and estimating motion of the object relative to the OCT imaging instrument; and
e. based on the value of the objective function, for each of the first and the second three-dimensional transform, computing a second three-dimensional displacement vector for each sample position in the object coordinate system; and
applying at least one three-dimensional transform to at least one of the first or the second three-dimensional data sets, or the first or the second derivative three-dimensional data sets to obtain at least one motion-corrected data set representing the object.
2. The method of claim 1 , wherein evaluating the objective function includes favoring similarity between the first and the second sets of interpolated values.
3. The method of claim 2 , wherein each three-dimensional data set is a set of values of a signal acquired from a region of the object, and wherein favoring similarity between the first and the second sets of interpolated values includes penalizing differences among signal values.
4. The method of claim 1 , wherein evaluating the objective function includes penalizing motion of the object relative to the imaging instrument.
5. The method of claim 1 , wherein applying at least one three-dimensional transform comprises applying the first three-dimensional transform to the first three-dimensional data set or the first derivative three-dimensional data set and applying the second three-dimensional transform to the second three-dimensional data set or derivative three-dimensional data set to obtain first and second motion-corrected data sets, respectively, and
further including combining data elements of the first and second motion-corrected data sets to obtain a merged motion-corrected data set having improved signal quality relative to the first or second motion-corrected data sets, individually.
6. The method of claim 5 , wherein said combining data elements further includes adjusting contributions of data elements of the first and second motion-corrected data sets to the merged motion-corrected data set based on at least one property of the first and the second three-dimensional transforms.
7. The method of claim 6 , wherein the at least one property of the first or the second three-dimensional transform is sampling density.
8. The method of claim 6 , wherein said adjusting further includes computing a weighted sum of the data elements of the first and second motion-corrected data sets.
9. The method of claim 1 , wherein the first three-dimensional data set is acquired using a first scan pattern and the second three-dimensional data set is acquired using a second scan pattern, wherein the first and the second scan patterns are complementary.
10. The method of claim 9 , wherein the complementary scan patterns are raster scans.
11. The method of claim 10 , wherein the raster scans are orthogonal.
12. The method of claim 1 , further comprising iteratively computing the first three-dimensional transform and the second three-dimensional transform.
13. The method of claim 1 , further comprising computing the first three-dimensional transform and the second three-dimensional transform using a numerical optimization method.
14. The method of claim 1 , wherein the first three-dimensional transform and the second three-dimensional transform are computed using a multi-resolution numerical optimization method.
15. The method of claim 1 , wherein the object is selected from the group consisting of the eye, retina, fovea, optic nerve head, or cornea.
16. The method of claim 1 , further including:
obtaining a first motion-corrected data set representing the object at a first time point,
obtaining a second motion-corrected data set representing the object at a second time point, and
comparing the first and the second motion-corrected data sets to quantitatively track changes in the object between the first and the second time points.
17. The method of claim 5 , further including:
obtaining a first merged motion-corrected data set representing the object at a first time point,
obtaining a second merged motion-corrected data set representing the object at a second time point, and
comparing the first and the second merged motion-corrected data sets to quantitatively track changes in the object between the first and the second time points.
18. The method of claim 16 or claim 17 , wherein the object is associated with a patient.
19. The method of claim 18 , wherein the first and the second time points correspond to the first and the second visits by the patient to a healthcare provider, respectively.
20. The method of claim 18 , wherein the first and the second time points correspond to the first and the second activities by the patient for scanning the object, respectively.
21. The method of claim 1 , wherein computing the first preprocessed three-dimensional data set or the second preprocessed data set or both is performed by at least one method selected from the group consisting of: image resampling, noise reduction, A-scan feature generation, tilt-compensation, and roll-off compensation.
22. The method of claim 1 , wherein at least one three-dimensional data set includes at least one of intensity data, Doppler shift data, or polarization data.
23. The method of claim 1 , wherein at least one derivative three-dimensional data set includes at least one of Doppler shift data or polarization data.
24. An optical coherence tomography (OCT) system correcting a data set representing an object for the object's motion relative to an OCT imaging instrument, comprising:
a memory configured to store at least one three-dimensional data set;
an acquisition module comprising an OCT imaging instrument, the acquisition module configured to acquire at least a first three-dimensional data set and a second three-dimensional data set representing overlapping regions of an object, wherein the first three-dimensional data set and the second three-dimensional data set are obtained in an instrument coordinate system;
a computing module configured to:
(i) optionally, compute at least one of a first derivative three-dimensional data set corresponding to the first three-dimensional data set, a second derivative three-dimensional data set corresponding to the second three-dimensional data set, a first preprocessed three-dimensional data set corresponding to the first three-dimensional dataset, or a second preprocessed three-dimensional data set corresponding to the second three-dimensional dataset;
(ii) determine sample positions in an object coordinate system;
(iii) for at least the first and the second three-dimensional data sets, compute a first three-dimensional transform and a second three-dimensional transform respectively, wherein every transform associates a three-dimensional displacement vector with each sample position in the object coordinate system, and wherein computing the first and the second three-dimensional transforms includes:
a. for each of the first and the second three-dimensional transform, computing a first three-dimensional displacement vector for each sample position in the object coordinate system;
b. for each of the first and the second three-dimensional transform, computing, based on the first three-dimensional displacement vectors, positions in the instrument coordinate system;
c. for each of the first and the second three-dimensional transform, interpolating data of the respective three-dimensional data set or preprocessed three-dimensional data set at the positions in the instrument coordinate system, thereby producing a first set of interpolated values and a second set of interpolated values;
d. evaluating an objective function by calculating similarity between the first and the second sets of interpolated values and estimating motion of the object relative to the OCT imaging instrument; and
e. based on the value of the objective function, for each of the first and the second three-dimensional transform, computing a second three-dimensional displacement vector for each sample position in the object coordinate system; and
a motion-correction module configured to apply at least one three-dimensional transform to at least one of the first or the second three-dimensional data sets, or the first or the second derivative three-dimensional data sets to obtain at least one motion-corrected data set representing the object.
25. The system of claim 24 , wherein evaluating the objective function includes favoring similarity between the first and the second sets of interpolated values.
26. The system of claim 24 , wherein each three-dimensional data set is a set of values of a signal acquired from a region of the object, and wherein favoring similarity between the first and the second sets of interpolated values includes penalizing differences among signal values.
27. The system of claim 24 , wherein evaluating the objective function includes penalizing motion of the object relative to the imaging instrument.
28. The system of claim 24 , wherein applying at least one three-dimensional transform comprises applying the first three-dimensional transform to the first three-dimensional data set or the first derivative three-dimensional data set and applying the second three-dimensional transform to the second three-dimensional data set or the second derivative three-dimensional data a set to obtain first and second motion-corrected data sets, respectively, and
further wherein the motion-correction module is configured to combine data elements of the first and second motion-corrected data sets to obtain a merged motion-corrected data set having improved signal quality relative to the first or second motion-corrected data sets, individually.
29. The system of claim 28 , wherein combining data elements further includes adjusting contributions of data elements of the first and second motion-corrected data sets to the merged motion-corrected data set based on at least one property of the first and the second three-dimensional transforms.
30. The system of claim 29 , wherein the at least one property of the first or the second three-dimensional transform is sampling density.
31. The system of claim 29 , wherein adjusting further includes computing a weighted sum of the data elements of the first and second motion-corrected data sets.
32. The system of claim 24 , wherein the data acquisition module is configured to acquire the first three-dimensional data set by using a first scan pattern and the second three-dimensional data set by using a second scan pattern, wherein the first and the second scan patterns are complementary.
33. The system of claim 32 , wherein the complementary scan patterns are raster scans.
34. The system of claim 33 , wherein the raster scans are orthogonal.
35. The system of claim 24 , wherein the computing module is configured to compute the first three-dimensional transform and the second three-dimensional transform iteratively.
36. The system of claim 24 , wherein the computing module is configured to compute the first three-dimensional transform and the second three-dimensional transform using a numerical optimization method.
37. The system of claim 24 , wherein the computing module is configured to compute the first three-dimensional transform and the second three-dimensional transform using a multi-resolution numerical optimization method.
38. The system of claim 24 , further including a time-comparison module configured to compare a first motion-corrected data set representing the object at a first time point and a second motion-corrected data set representing the object at a second time point,
wherein comparing the first and the second motion-corrected data sets quantitatively tracks changes in the object between the first and the second time points.
39. The system of claim 28 , further including a time-comparison module configured to compare a first merged motion-corrected data set representing the object at a first time point and a second merged motion-corrected data set representing the object at a second time point,
wherein comparing the first and the second merged motion-corrected data sets quantitatively tracks changes in the object between the first and the second time points.
40. The system of claim 38 or claim 39 , wherein the object is associated with a patient.
41. The system of claim 40 , wherein the object is selected from the group consisting of the eye, retina, fovea, optic nerve head, or cornea.
42. The system of claim 41 , wherein the first and the second time points correspond to the first and the second visits by the patient to a healthcare provider, respectively.
43. The system of claim 41 , wherein the first and the second time points correspond to the first and the second activities by the patient for scanning the object, respectively.
44. The system of claim 24 , wherein computing the first preprocessed three-dimensional data set or the second preprocessed data set or both is performed by at least one method selected from the group consisting of: image resampling, noise reduction, A-scan feature generation, tilt-compensation, and roll-off compensation.
45. The system of claim 24 , wherein at least one three-dimensional data set includes at least one of intensity data, Doppler shift data, or polarization data.
46. The system of claim 24 , wherein at least one derivative three-dimensional data set includes at least one of Doppler shift data or polarization data.
47. A non-transitory computer-readable medium having thereon a sequence of instructions, which, when executed by a processor:
cause an optical coherence tomography (OCT) imaging device to acquire at least a first three-dimensional data set and a second three-dimensional data set representing overlapping regions of an object with the OCT imaging instrument, wherein the first three-dimensional data set and the second three-dimensional data set are obtained in an instrument coordinate system;
optionally, cause the processor to compute at least one of a first derivative three-dimensional data set corresponding to the first three-dimensional data set, a second derivative three-dimensional data set corresponding to the second three-dimensional data set, a first preprocessed three-dimensional data set corresponding to the first three-dimensional dataset, or a second preprocessed three-dimensional data set corresponding to the second three-dimensional dataset;
cause the processor to determine sample positions in an object coordinate system;
for at least the first and the second three-dimensional data sets, cause the processor to compute a first three-dimensional transform and a second three-dimensional transform respectively, wherein every transform associates a three-dimensional displacement vector with each sample position in the object coordinate system, and wherein computing the first and the second three-dimensional transforms includes:
a. for each of the first and the second three-dimensional transform, computing a first three-dimensional displacement vector for each sample position in the object coordinate system;
b. for each of the first and the second three-dimensional transform, computing, based on the first three-dimensional displacement vectors, positions in the instrument coordinate system;
c. for each of the first and the second three-dimensional transform, interpolating data of the respective three-dimensional data set or preprocessed three-dimensional data set at the positions in the instrument coordinate system, thereby producing a first set of interpolated values and a second set of interpolated values;
d. evaluating an objective function by calculating similarity between the first and the second sets of interpolated values and estimating motion of the object relative to the OCT imaging instrument; and
e. based on the value of the objective function, for each of the first and the second three-dimensional transform, computing a second three-dimensional displacement vector for each sample position in the object coordinate system; and
cause the processor to apply at least one three-dimensional transform to at least one of the first or the second three-dimensional data sets, or the first or the second derivative three-dimensional data sets to obtain at least one motion-corrected data set representing the object.