Abstract: A device and method for glaucoma auxiliary diagnosis, and a non-transitory storage medium are provided. The device includes an obtaining unit and a processing unit. The obtaining unit is configured to obtain a color fundus image of a patient. The processing unit is configured to perform feature extraction on the color fundus image to obtain a first feature map. The processing unit is further configured to perform image segmentation on the color fundus image according to the first feature map to obtain an optic disc image in the color fundus image, where the optic disc image corresponds to an optic disc area in the color fundus image. The processing unit is further configured to perform feature extraction on the optic disc image and the color fundus image according to the first feature map to obtain a probability that the patient has glaucoma.

Abstract: The present disclosure provides a method, device, computer apparatus, and storage medium for training recognition model and recognizing fundus features. The method includes: obtaining a color fundus image sample associated with a label value, inputting the color fundus image sample into a preset recognition model containing initial parameters; extracting a red channel image; inputting the red channel image into the first convolutional neural network to obtain a first recognition result and a feature image of the red channel image; combining the color fundus image sample with the feature image to generate a combined image, and inputting the combined image into the second convolutional neural network to obtain a second recognition result; obtaining a total loss value through a loss function, and when the total loss value is less than or equal to a preset loss threshold, ending the training of the preset recognition model.

Abstract: A method for detecting and locating a lesion in a medical image is provided. A target medical image of a lesion is obtained and input into a deep learning model to obtain a target sequence. A first feature map output from the last convolution layer in the deep learning model is extracted. A weight value of each network unit corresponding to each preset lesion type in a fully connected layer is extracted. For each preset lesion type, a fusion feature map is calculated according to the first feature map and the corresponding weight value and resampled to the size of the target medical image to generate a generic activation map. The maximum connected area in each generic activation map is determined, and a mark border surrounding the maximum connected area is created. A mark border corresponding to each preset lesion type is added to the target medical image.

Abstract: The present disclosure provides a method, device, computer apparatus, and storage medium for training recognition model and recognizing fundus features. The method includes: obtaining a color fundus image sample associated with a label value, inputting the color fundus image sample into a preset recognition model containing initial parameters; extracting a red channel image; inputting the red channel image into the first convolutional neural network to obtain a first recognition result and a feature image of the red channel image; combining the color fundus image sample with the feature image to generate a combined image, and inputting the combined image into the second convolutional neural network to obtain a second recognition result; obtaining a total loss value through a loss function, and when the total loss value is less than or equal to a preset loss threshold, ending the training of the preset recognition model.

Abstract: A method for determining a centerline of a blood vessel in an image associated with a subject is provided. The method includes obtaining a centerline model used for identifying a centerline of a blood vessel and identifying the centerline of the blood vessel based on the centerline model.

Abstract: Methods and systems for determining a region of interest (ROI) in an image associated with a subject are provided. Slice information of multiple slices of an image are identified and multiple slice ranges are determined based on the slice information. Based on the slice ranges, multiple sub-images of the image are then determined. For each sub-image, a template and a corresponding probability graph are acquired, a registration result is determined by registering the sub-image with the template, and a ROI in the sub-image is identified based on the registration result. A ROI of the image can be determined by combining the ROIs of the sub-images.

Abstract: A device and method for glaucoma auxiliary diagnosis, and a non-transitory storage medium are provided. The device includes an obtaining unit and a processing unit. The obtaining unit is configured to obtain a color fundus image of a patient. The processing unit is configured to perform feature extraction on the color fundus image to obtain a first feature map. The processing unit is further configured to perform image segmentation on the color fundus image according to the first feature map to obtain an optic disc image in the color fundus image, where the optic disc image corresponds to an optic disc area in the color fundus image. The processing unit is further configured to perform feature extraction on the optic disc image and the color fundus image according to the first feature map to obtain a probability that the patient has glaucoma.

Abstract: A method and device for neural network-based optical coherence tomography (OCT) image lesion detection, and a medium are provided. The method includes the following. An OCT image is obtained. The OCT image is inputted into a lesion-detection network model. A position, a category score, and a positive score of each lesion box in the OCT image are outputted through the lesion-detection network model. A lesion detection result of the OCT image is obtained according to the position, the category score, and the positive score of each lesion box. The lesion-detection network model includes a category detection branch configured to obtain, for each of the anchor boxes, a position and a category score of the anchor box, and a lesion positive score regression branch configured to obtain, for each of the anchor boxes, a positive score of whether the anchor box belongs to a lesion, to reflect severity of lesion positive.

Abstract: A neural network model training method and apparatus, a computer device, and a storage medium are provided. The method includes: obtaining a model prediction value of each of all reference samples based on a trained deep neural network model, calculating a difference measurement index between the model prediction value of each reference sample and a real annotation corresponding to the reference sample, and using a target reference sample whose difference measurement index is less than or equal to a preset threshold as a comparison sample; using a training sample whose similarity with the comparison sample meets a preset augmentation condition as a to-be-augmented sample; and performing data augmentation on the to-be-augmented sample, and using the obtained target training sample as a training sample to train the trained deep neural network model until model prediction values of all verification samples in a verification set meet a preset training ending condition.

Abstract: A method for detecting and locating a lesion in a medical image is provided. A target medical image of a lesion is obtained and input into a deep learning model to obtain a target sequence. A first feature map output from the last convolution layer in the deep learning model is extracted. A weight value of each network unit corresponding to each preset lesion type in a fully connected layer is extracted. For each preset lesion type, a fusion feature map is calculated according to the first feature map and the corresponding weight value and resampled to the size of the target medical image to generate a generic activation map. The maximum connected area in each generic activation map is determined, and a mark border surrounding the maximum connected area is created. A mark border corresponding to each preset lesion type is added to the target medical image.

Abstract: A method for extracting a blood vessel may include acquiring an image relating to a blood vessel, the image including multiple slices; determining a region of interest in the image; establishing a blood vessel model; and extracting the blood vessel from the region of interest based on the blood vessel model.

Abstract: A method for extracting a blood vessel is provided. An image relating to a blood vessel may be acquired. The image may include multiple slices. A region of interest in the image may be determined. A blood vessel model may be obtained. The blood vessel may be extracted from the region of interest based on the blood vessel model.

Abstract: A method for extracting a blood vessel may include acquiring an image relating to a blood vessel, the image including multiple slices; determining a region of interest in the image; establishing a blood vessel model; and extracting the blood vessel from the region of interest based on the blood vessel model.