TY - GEN
T1 - Automated Classification of Diabetic Retinopathy Using Deep Learning Architecture
AU - D’Souza, Noel
AU - Siddalingaswamy, P. C.
AU - Reddy, N. V.Subba
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023.
PY - 2023
Y1 - 2023
N2 - Diabetic Retinopathy (DR) is an eye disease that can lead to blindness. It is one of the leading causes of vision loss in the world. Early detection of DR is critical to prevent permanent vision loss. DR is diagnosed by a trained ophthalmologist. In many developing countries, access to ophthalmologists for regular screening is often difficult. Using Machine Learning to overcome the need for skilled personnel to diagnose DR is a viable alternative. To automate the diagnosis of DR, we preprocessed our images to obtain 6 channels before training a modified Inception-v3 model on retinal fundus images obtained from Kaggle. We then trained a second model, combining data from both eyes, obtained as outputs from Inception-v3 to diagnose the stage of DR. Using this architecture, we were able to obtain a Kappa score of 0.821 on the dataset provided by Kaggle.
AB - Diabetic Retinopathy (DR) is an eye disease that can lead to blindness. It is one of the leading causes of vision loss in the world. Early detection of DR is critical to prevent permanent vision loss. DR is diagnosed by a trained ophthalmologist. In many developing countries, access to ophthalmologists for regular screening is often difficult. Using Machine Learning to overcome the need for skilled personnel to diagnose DR is a viable alternative. To automate the diagnosis of DR, we preprocessed our images to obtain 6 channels before training a modified Inception-v3 model on retinal fundus images obtained from Kaggle. We then trained a second model, combining data from both eyes, obtained as outputs from Inception-v3 to diagnose the stage of DR. Using this architecture, we were able to obtain a Kappa score of 0.821 on the dataset provided by Kaggle.
UR - http://www.scopus.com/inward/record.url?scp=85174523778&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85174523778&partnerID=8YFLogxK
U2 - 10.1007/978-981-99-4284-8_26
DO - 10.1007/978-981-99-4284-8_26
M3 - Conference contribution
AN - SCOPUS:85174523778
SN - 9789819942831
T3 - Lecture Notes in Networks and Systems
SP - 309
EP - 321
BT - Advanced Computational and Communication Paradigms - Proceedings of ICACCP 2023
A2 - Borah, Samarjeet
A2 - Gandhi, Tapan K.
A2 - Piuri, Vincenzo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 4th International Conference on Advanced Computational and Communication Paradigms, ICACCP 2023
Y2 - 16 February 2023 through 18 February 2023
ER -
