A Semantic Adversarial Network for Detection and Classification of Myopic Maculopathy  

作　　者：Qaisar Abbas Abdul Rauf Baig Ayyaz Hussain 

机构地区：[1]College of Computer and Information Sciences,Imam Mohammad Ibn Saud Islamic University(IMSIU),Riyadh,11432,Saudi Arabia [2]Department of Computer Science,Quaid-i-Azam University,Islamabad,44000,Pakistan

出　　处：《Computers, Materials & Continua》2023年第4期1483-1499,共17页计算机、材料和连续体（英文）

基　　金：the Deanship of Scientific Researchat Imam Mohammad Ibn Saud Islamic University (IMSIU) for funding and supporting this workthrough Research Partnership Program no. RP-21-07-04.

摘　　要：The diagnosis of eye disease through deep learning (DL) technologyis the latest trend in the field of artificial intelligence (AI). Especially indiagnosing pathologic myopia (PM) lesions, the implementation of DL is adifficult task because of the classification complexity and definition system ofPM. However, it is possible to design an AI-based technique that can identifyPM automatically and help doctors make relevant decisions. To achieve thisobjective, it is important to have adequate resources such as a high-qualityPM image dataset and an expert team. The primary aim of this research isto design and train the DLs to automatically identify and classify PM intodifferent classes. In this article, we have developed a new class of DL models(SAN-FSL) for the segmentation and detection of PM through semanticadversarial networks (SAN) and few-short learning (FSL) methods, respectively.Compared to DL methods, the conventional segmentation methodsuse supervised learning models, so they (a) require a lot of data for trainingand (b) fixed weights are used after the completion of the training process.To solve such problems, the FSL technique was employed for model trainingwith few samples. The ability of FSL learning in UNet architectures is beingexplored, and to fine-tune the weights, a few new samples are being providedto the UNet. The outcomes show improvement in the detection area andclassification of PM stages. Betterment in the result is observed by sensitivity(SE) of 95%, specificity (SP) of 96%, and area under the receiver operatingcurve (AUC) of 98%, and the higher F1-score is achieved using 10-fold crossvalidation.Furthermore, the obtained results confirmed the superiority of theSAN-FSL method.

关 键 词：Artificial intelligence CARDIOVASCULAR vision loss deep learning few-shot learning semantic segmentation myopic maculopathy 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] R77[自动化与计算机技术—计算机科学与技术]