Ophthalmic information processing apparatus, ophthalmic apparatus, ophthalmic information processing method, and recording medium
An ophthalmic information processing apparatus includes an acquisition unit and a disease estimation unit. The acquisition unit is configured to acquire a plurality of images with different cross-sectional orientations from each other of a subject's eye. The disease estimation unit is configured to output estimation information for estimating whether or not the subject's eye is a glaucoma eye from the images, using a plurality of learned models obtained by performing machine learning for each type of the images.
1 . An ophthalmic information processing apparatus, comprising:
an acquisition circuit configured to acquire a plurality of images with different cross-sectional orientations from each other of a subject's eye, the images having image types including a B-scan image in a horizontal direction passing through a center of a optic disc or near the center, a B-scan image in a vertical direction passing through the center of the optic disc or near the center, a B-scan image in the vertical direction passing through a fovea or near the fovea, a projection image, and an en-face image; and
a disease estimation circuit configured to output estimation information for estimating whether or not the subject's eye is a glaucoma eye from the images, using a plurality of learned models obtained by performing machine learning for each type of the images,
wherein
the disease estimation circuit includes:
a plurality of estimator circuits configured to output confidence score information that represents confidence score that the subject's eye is the glaucoma eye, using each of the learned models for each type of the images; and
a classifier configured to output the estimation information from a plurality of confidence score information that are output from the estimator circuits, using a classification model obtained by performing machine learning.
2 . The ophthalmic information processing apparatus of claim 1 , further comprising
a learning circuit configured to generate the learned models by performing supervised machine learning for each type of the images.
3 . The ophthalmic information processing apparatus of claim 1 , further comprising
an image generator circuit configured to generate at least one of the images based on three-dimensional OCT data of the subject's eye.
4 . An ophthalmic apparatus, comprising:
an OCT device including a scanner and configured to perform optical coherence tomography on the subject's eye;
an image generator circuit configured to generate at least one of the images based on the three-dimensional OCT data acquired by the OCT unit; and
an ophthalmic information processing apparatus, wherein
the ophthalmic information processing apparatus, comprising:
an acquisition circuit configured to acquire a plurality of images with different cross-sectional orientations from each other of a subject's eye, the images having image types including a B-scan image in a horizontal direction passing through a center of a optic disc or near the center, a B-scan image in a vertical direction passing through the center of the optic disc or near the center, a B-scan image in the vertical direction passing through a fovea or near the fovea, a projection image, and an en-face image; and
a disease estimation circuit configured to output estimation information for estimating whether or not the subject's eye is a glaucoma eye from the images, using a plurality of learned models obtained by performing machine learning for each type of the images, and
the disease estimation circuit includes:
a plurality of estimator circuits configured to output confidence score information that represents confidence score that the subject's eye is the glaucoma eye, using each of the learned models for each type of the images; and
a classifier configured to output the estimation information from a plurality of confidence score information that are output from the estimator circuits, using a classification model obtained by performing machine learning.
5 . An ophthalmic information processing method, comprising:
an acquisition step of acquiring a plurality of images with different cross-sectional orientations from each other of a subject's eye, the images having image types including a B-scan image in a horizontal direction passing through a center of a optic disc or near the center, a B-scan image in a vertical direction passing through the center of the optic disc or near the center, a B-scan image in the vertical direction passing through a fovea or near the fovea, a projection image, and an en-face image; and
a disease estimation step of outputting estimation information for estimating whether or not the subject's eye is a glaucoma eye from the images, using a plurality of learned models obtained by performing machine learning for each type of the images,
wherein
the disease estimation step includes:
a plurality of estimation steps of outputting confidence score information that represents confidence score that the subject's eye is the glaucoma eye, using each of the learned models for each type of the images; and
a classification step of outputting the estimation information from a plurality of confidence score information that are output from the plurality of estimation steps, using a classification model obtained by performing machine learning.
6 . The ophthalmic information processing method of claim 5 , further comprising
a learning step of generating the learned models by performing supervised machine learning for each type of the images.
7 . The ophthalmic information processing method of claim 5 , further comprising
an image generating step of generating at least one of the images based on three-dimensional OCT data of the subject's eye.
8 . A computer readable non-transitory recording medium in which a program for causing a computer to execute each step of the ophthalmic information processing method is recorded, wherein
the ophthalmic information processing method comprises:
an acquisition step of acquiring a plurality of images with different cross-sectional orientations from each other of a subject's eye, the images having image types including a B-scan image in a horizontal direction passing through a center of a optic disc or near the center, a B-scan image in a vertical direction passing through the center of the optic disc or near the center, a B-scan image in the vertical direction passing through a fovea or near the fovea, a projection image, and an en-face image; and
a disease estimation step of outputting estimation information for estimating whether or not the subject's eye is a glaucoma eye from the images, using a plurality of learned models obtained by performing machine learning for each type of the images, wherein
the disease estimation step includes:
a plurality of estimation steps of outputting confidence score information that represents confidence score that the subject's eye is the glaucoma eye, using each of the learned models for each type of the images; and
a classification step of outputting the estimation information from a plurality of confidence score information that are output from the plurality of estimation steps, using a classification model obtained by performing machine learning.