DME prognostic information prediction system based on integrated machine learning and application method thereof

Abstract

The invention provides a DME prognostic information prediction system based on integrated machine learning and an application method thereof. The DME prognostic information prediction system is characterized in that a preprocessing module for preprocessing data; a feature extraction module performs image feature extraction on the preprocessed image by using three deep learning models; a network construction module is used for constructing a deep learning network; a feature fusion module fuses the obtained image features; a data processing module processes the image fusion features and text data generated by the deep learning network to generate a probability distribution map; and a prediction module generates prediction values of CFT and BCVA according to the probability distribution map.The multi-mode human activity recognition system and the application method thereof can construct a deep learning network through the network construction module to process the OCT image, can obtain fused image features and text features of the clinical variables by adding the clinical variables, and finally, performs CFT and BCVA prediction by the prediction module, so that objective prediction values are accurately provided, and the prediction precision is effectively improved, and the defects of a traditional prediction method are overcome.

Description

technical field [0001] The present invention relates to the fields of clinical ophthalmology and computer engineering, and more specifically, relates to a DME prognosis information prediction system based on integrated machine learning, and also relates to an application method of the system. Background technique [0002] Patients with diabetic macular edema (DME) have significantly decreased visual acuity, morphologically manifested by retinal thickening from the fovea to the optic nerve. Intravitreal injection of anti-vascular endothelial growth factor (VEGF) drugs is currently the most important method for the treatment of DME, but different DME patients have different responses to anti-VEGF drugs, and about one-third of DME patients respond ineffectively (ineffective response is defined as three-month follow-up). In the first month after regular injection of anti-VEGF drugs, the decrease of central macular thickness (CFT) is less than 50 μm or the improvement of best cor...

AI Technical Summary

Problems solved by technology

The main reason is that the traditional prediction method is highly subjective, there are many and complex factors to be considered comprehensively, and a large part depends on the ophthalmologist's clinical experience and knowledge level

Therefore, for young doctors who lack clinical experience and doctors in community primary hospitals with relatively low medical level, it is very difficult to make accurate postoperative predictions

However, there is still no rapid, accurate, versatile, and widely used evaluation method for postoperative recovery of anti-VEGF in DME patients for clinical use.

Images