Detecting disease in retinal images

A method of generating output data providing an indication of the presence or absence of disease from a retinal image of a patient involves receiving data from the image associated with the presence and abundance of three different types of eye lesion, such as microaneurysms, blot haemorrhages and exudates. These data are then combined to generate output data indicating the presence or absence of a disease, such as diabetic retinopathy, and these may represent confidence values associated with the detection of each lesion type in the image. These data can also be processed to generate a weight for each data value which influences how they are combined, such that the correspondence of the disease indicating output data withpredetermined classification data is maximised. Each retinal image may comprise two images corresponding one each to the left and right eyes of the patient.

Attached files:


Patents:
GB 2,467,840

Inventor(s): FLEMING ALAN DUNCAN [GB]

Type of Offer: Sale



« More Data Processing Patents

---