Deep learning based genome analysis and NGS-RNA LL identification with a novel hybrid model. (November 2020)

• Record Type: Journal Article
• Authors: Ramamurthy, Madhumitha; Krishnamurthi, Ilango; Vimal, S.; Robinson, Y. Harold
• Is Part Of: Bio systems. Volume 197(2020)
• Publication Date: 2020-11
• Subjects: SegNet -- MultiResUNet -- Krill herd optimization -- Genome sequencing -- NGS disease Prediction -- Convolutional neural network; Biological systems -- Periodicals; Biology -- Periodicals; Biology -- Periodicals; Evolution -- Periodicals; Biologie -- Périodiques; Évolution -- Périodiques; 570
• DOI: 10.1016/j.biosystems.2020.104211
• Languages: English
• ISSNs: 0303-2647

Abstract

Abstract: The conventional image segmentation techniques have a lot of issues with highest computational cost and low level accuracy for medical image diagnosis and genome analysis. The deep learning based optimization models utilize to predict the liver cancer with RNA genome using CT images and the prediction of genome classification with NGS is a higher probable in recent medical disease classification. This paper proposes a hybrid deep learning technique constructs with SegNet, MultiResUNet, and Krill Herd optimization (KHO) algorithm to perform the extraction of the liver lesions and RNA sequencing that the optimization techniques used into the deep learning method. The proposed technique implements the SegNet for segregating the liver with genome from the CT scan; the MultiResUNet is constructed to perform the extractions of liver lesions. The KHO algorithm is combined with the deep learning approaches for tuning the hyper parameters to every Convolutional neural network model and enhances the segmentation process which may elaborately identifies the sequence that causes the liver classification disease. The proposed technique is compared with the related techniques on liver lesion classification (LL) for NGS in genome. The performance results show that the proposed technique is better to other algorithms on various performance metrics.