RNN-CNN Based Cancer Prediction Model for Gene Expression

RNN-CNN Based Cancer Prediction Model for Gene  Expression

Mrs. Uma H R Asst. Professor,

Dept of computer science and engineering,  BGS Institute of Technology, Adhichunchanagiri university

BG Nagara, Karnataka

Sonashree N,4th year ,8th semester

Computer Science and Engineering, BGS Institute of Technology, Adhichunchanagiri university

BG Nagara, Karnataka

ABSTRACT

One of those sicknesses that is generally destructive to individuals is disease. The best way to forestall any damage to mankind is by its initial disclosure and treatment. Different kinds of tests are directed in the clinical labs for the identification of malignant growth. Disease can likewise be distinguished at the hereditary level. For this unmistakable AI and profound learning techniques as of now exist. This paper proposes a crossover strategy in view of Repetitive Brain Organization (RNN) and Convolution Brain Organization (CNN) to foresee various kinds of disease like Bosom, Lung, Uterine, Kidney, Prostate and colon malignant growth from quality articulation information. The bottleneck highlights are removed utilizing the sandwich stacked strategy in light of VGG16 and VGG19 pre-prepared models. A while later, the proposed half breed classifier in light of RNN-CNN has been utilized to order the information into different classes. The proposed model performs better compared to the next existing strategies, for example, VGG16, VGG19, ResNet50, Commencement V3 and MobileNet classifier as far as different execution measurements like exactness, Mean Square Mistake (MSE), accuracy, review, and F1 score. RNN-CNN classifier gives the most elevated precision of 0.978 among the wide range of various existing techniques for Dataset 1 and the most noteworthy exactness of 0.994 for Dataset 2 at 80% preparation information. Then again, RNN-CNN classifier gives the least MSE of 0.101 among the wide range of various existing strategies for Dataset 1 and the most reduced MSE of 0.006 for Dataset 2 at 80% preparation information.

INDEX TERMS Cancer, CNN, deep learning, gene expression, machine learning, RNN.