Quality evaluation of fundus images apparatus using transfer learning and method thereof

• Main Author: NAM YUN YOUNG
• Format: Patent
• Language: English; Korean
• Published: 22.02.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS

The present invention relates to a device for evaluating quality of a fundus image using transfer learning and a method thereof. The present invention allows that only a good image is applied to learning and a bad image is discarded by evaluating quality of a fundus image after data pre-processing of the acquired fundus image using transfer learning utilizing a convolutional neural network (CNN) model, which is an image classification artificial intelligence model that has been trained on general images. The device comprises: an image acquisition unit; a transfer learning unit; an evaluation image acquisition unit; and an image quality evaluation unit. 본 발명은 안저 이미지 품질 평가 장치 및 방법에 관한 것으로, 더욱 상세하게는 일반 이미지에 대해 학습을 수행한 이미지 분류 인공지능 모델인 합성곱 신경망(Convolution Neural Network: CNN) 모델을 활용한 전이학습을 이용하고, 획득되는 안저 이미지를 데이터 전처리한 후, 안저 이미지의 품질을 평가하여 나쁜 이미지는 버리고 좋은 이미지만 학습에 적용할 수 있도록 하는 전이학습을 이용한 안저 이미지 품질 평가 장치 및 방법에 관한 것이다.