Classification method and classification device for classifying level of AMD
A classification method and a classification device for classifying a level of an age-related macular degeneration are provided. The classification method includes the following. An object detection model and a first classification model are pre-stored. A fundus image is obtained. A bounding box is generated in the fundus image according to a macula in the fundus image detected by the object detection model. An intersection over union between a predetermined area and the bounding box in the fundus image is calculated. A classification of the fundus image is generated according to the first classification model in response to the intersection over union being greater than a threshold.
1. An eye examining device for classifying a level of age-related macular degeneration, the eye examining device comprising:
a transceiver;
a camera;
a storage medium, storing an object detection model, a first classification model, and a second classification model different from the first classification model; and
a processor, coupled to the camera, the storage medium and the transceiver, wherein the processor is configured to:
capture a fundus image through the camera;
generate a bounding box in the fundus image according to a macula in the fundus image detected by the object detection model;
calculate an intersection over union between a predetermined area and the bounding box in the fundus image, wherein the fundus image and the predetermined area are rectangles, a center point of the predetermined area is located at a geometric center of the fundus image, a first edge of the predetermined area is distanced from a first boundary of the fundus image by a first distance, and a second edge of the predetermined area is distanced from a second boundary of the fundus image by the first distance, wherein the second edge is an opposite edge of the first edge, and the second boundary is an opposite boundary of the first boundary;
input an image in the bounding box into the first classification model to generate a classification of the fundus image in response to the intersection over union being greater than a threshold, wherein the first classification model was trained by a portion of fundus image;
input the fundus image into the second classification model to generate the classification in response to the intersection over union being less than or equal to the threshold, wherein the second classification model was trained by an entire fundus image; and
output the classification through the transceiver, wherein the classification indicates one of a first stage, a second stage, a third stage, and a fourth stage of the age-related macular degeneration.
2. The eye examining device described in claim 1 , wherein the predetermined area is a first rectangle, and the processor is further configured to:
obtain an eyeball area in the fundus image; and
generate a length of a long edge and a length of a short edge of the first rectangle according to a diameter of the eyeball area.
3. The eye examining device described in claim 1 , wherein the first classification model and the second classification model have a same convolutional neural network architecture.
4. An eye examining method for classifying a level of age-related macular degeneration, the eye examining method comprising:
pre-storing an object detection model, a first classification model, and a second classification model different from the first classification model;
capturing a fundus image through a camera;
generating a bounding box in the fundus image according to a macula in the fundus image detected by the object detection model;
calculating an intersection over union between a predetermined area and the bounding box in the fundus image, wherein the fundus image and the predetermined area are rectangles, a center point of the predetermined area is located at a geometric center of the fundus image, a first edge of the predetermined area is distanced from a first boundary of the fundus image by a first distance, and a second edge of the predetermined area is distanced from a second boundary of the fundus image by the first distance, wherein the second edge is an opposite edge of the first edge, and the second boundary is an opposite boundary of the first boundary;
inputting an image in the bounding box into the first classification model to generate a classification of the fundus image in response to the intersection over union being greater than a threshold, wherein the first classification model was trained by a portion of fundus image;
inputting the fundus image into the second classification model to generate the classification in response to the intersection over union being less than or equal to the threshold, wherein the second classification model was trained by an entire fundus image; and
outputting the classification, wherein the classification indicates one of a first stage, a second stage, a third stage, and a fourth stage of the age-related macular degeneration.