Effective tensor based PCA machine learning techniques for glaucoma detection and ASPP – EffUnet classification

Venkatachalam, K. and Bacanin, Nebojsa and Kabir, Enamul ORCID: https://orcid.org/0000-0002-6157-2753 and Parbu, P. (2021) Effective tensor based PCA machine learning techniques for glaucoma detection and ASPP – EffUnet classification. In: 10th International Conference on Health Information Science (HIS 2021), 25 Oct - 28 Oct 2021, Melbourne, Australia.


Main problem in current research area focused on generating automatic AI technique to detect bio medical images by slimming the dataset. Reducing the original dataset with actual unwanted noises can accelerate new data which helps to detect diseases with high accuracy. Highest level of accuracy can be achieved only by ensuring accuracy at each level of processing steps. Dataset slimming or reduction is NP hard problems due its resembling variants. In this research work we ensure high accuracy in two phases. In phase one feature selection using Normalized Tensor Tubal PCA (NTT-PCA) method is used. This method is based on tensor with single value decomposition (SVD) for accurate dimensionality reduction problems. The dimensionality reduced output from phase one is further processed for accurate classification in phase two. The classification of affected images is detected using ASPP – EffUnet. The atrous spatial pyramid pooling (ASPP) with efficient convolutional block in Unet is combined to provide ASPP – EffUnet CNN architecture for accurate classification. This two phase model is designed and implemented on benchmark datasets of glaucoma detection. It is processed efficiently by exploiting fundus image in the dataset. We propose novel AI techniques for segmenting the eye discs using EffUnet and perform classification using ASPP-EffUnet techniques. Highest accuracy is achieved by NTT-PCA dimensionality reduction process and ASPP-EffUnet based classification which detects the boundaries of eye cup and optical discs very curiously. Our resulting algorithm “NTT-PCA with ASPP-EffUnet “for dimensionality reduction and classification process which is optimized for reducing computational complexity with existing detection algorithms like PCA-LA-SVM,PCA-ResNet ASPP –Unet. We choose benchmark datasets ORIGA for our experimental analysis. The crucial areas in clinical setup are examined and implemented successfully. The prediction and classification accuracy of proposed technique is achieved nearly 100%.

Statistics for USQ ePrint 44936
Statistics for this ePrint Item
Item Type: Conference or Workshop Item (Commonwealth Reporting Category E) (Paper)
Refereed: Yes
Item Status: Live Archive
Additional Information: Permanent restricted access to Published version in accordance with the copyright policy of the publisher.
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Sciences (6 Sep 2019 - 31 Dec 2021)
Faculty/School / Institute/Centre: Current - Institute for Resilient Regions - Centre for Health Research (1 Apr 2020 -)
Date Deposited: 18 Nov 2021 04:31
Last Modified: 08 Mar 2022 03:01
Uncontrolled Keywords: glaucoma detection, tensor tubal PCA, ASPP, EffUnet, convolutional neural network
Fields of Research (2008): 08 Information and Computing Sciences > 0801 Artificial Intelligence and Image Processing > 080199 Artificial Intelligence and Image Processing not elsewhere classified
Fields of Research (2020): 40 ENGINEERING > 4003 Biomedical engineering > 400399 Biomedical engineering not elsewhere classified
Socio-Economic Objectives (2008): E Expanding Knowledge > 97 Expanding Knowledge > 970110 Expanding Knowledge in Technology
E Expanding Knowledge > 97 Expanding Knowledge > 970111 Expanding Knowledge in the Medical and Health Sciences
Socio-Economic Objectives (2020): 22 INFORMATION AND COMMUNICATION SERVICES > 2204 Information systems, technologies and services > 220403 Artificial intelligence
Identification Number or DOI: https://doi.org/10.1007/978-3-030-90885-0_17
URI: http://eprints.usq.edu.au/id/eprint/44936

Actions (login required)

View Item Archive Repository Staff Only