TY - JOUR
T1 - Comparative Analysis of Pre-trained ResNet and DenseNet Models for the Detection of Diabetic Macular Edema
AU - Pavithra, K. C.
AU - Kumar, Preetham
AU - Geetha, M.
AU - Bhandary, Sulatha V.
N1 - Publisher Copyright:
© Published under licence by IOP Publishing Ltd.
PY - 2023
Y1 - 2023
N2 - A major setback in Deep Learning (DL) is that a huge amount of data is essential to render the trained model more generalizable. Constructing a higher-order model based on insufficient data has a detrimental effect on testing performance. Transfer Learning (TL) involves less training data than conventional DL approaches and offers superior decision support. Healthcare datasets of reasonable sizes are generally inappropriate for training DL models. A promising solution to the issue would be to use TL methods for the classification of medical image datasets. This paper aims at the training and evaluation of six variants of pre-trained ResNet and three variants of pre-trained DenseNet models for Diabetic Macular Edema (DME) classification employing a public retinal Optical Coherence Tomography (OCT) image dataset. Among the ResNet implementations, ResNet101V2 has delivered the highest accuracy value of 95%. And among the DenseNet implementations, DenseNet201 has yielded an exceptional classification accuracy of 99%. As all three DenseNet versions, along with the ResNet101V2 version, have produced noticeably good results (accuracy values greater than 95%), they can be used to screen the retinal OCT images of DME patients.
AB - A major setback in Deep Learning (DL) is that a huge amount of data is essential to render the trained model more generalizable. Constructing a higher-order model based on insufficient data has a detrimental effect on testing performance. Transfer Learning (TL) involves less training data than conventional DL approaches and offers superior decision support. Healthcare datasets of reasonable sizes are generally inappropriate for training DL models. A promising solution to the issue would be to use TL methods for the classification of medical image datasets. This paper aims at the training and evaluation of six variants of pre-trained ResNet and three variants of pre-trained DenseNet models for Diabetic Macular Edema (DME) classification employing a public retinal Optical Coherence Tomography (OCT) image dataset. Among the ResNet implementations, ResNet101V2 has delivered the highest accuracy value of 95%. And among the DenseNet implementations, DenseNet201 has yielded an exceptional classification accuracy of 99%. As all three DenseNet versions, along with the ResNet101V2 version, have produced noticeably good results (accuracy values greater than 95%), they can be used to screen the retinal OCT images of DME patients.
UR - http://www.scopus.com/inward/record.url?scp=85176215350&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85176215350&partnerID=8YFLogxK
U2 - 10.1088/1742-6596/2571/1/012006
DO - 10.1088/1742-6596/2571/1/012006
M3 - Conference article
AN - SCOPUS:85176215350
SN - 1742-6588
VL - 2571
JO - Journal of Physics: Conference Series
JF - Journal of Physics: Conference Series
IS - 1
M1 - 012006
T2 - 2nd International Conference on Artificial Intelligence, Computational Electronics and Communication System, AICECS 2023
Y2 - 16 February 2023 through 17 February 2023
ER -