1. Improved accuracy in early identification of ischaemic stroke using convolutional neural network with K-Nearest Neighbors.
- Author
-
Manikandan, S., Tamilselvi, M., and Sajiv, G.
- Subjects
CONVOLUTIONAL neural networks ,K-nearest neighbor classification ,ISCHEMIC stroke ,CONFIDENCE intervals ,SAMPLE size (Statistics) - Abstract
The primary goal of this research is to recognize signs of brain stroke in MRI pictures. We will contrast a novel convolutional neural network that seeks to improve specificity and accuracy with K-Nearest Neighbors. Ten participants from the K-Nearest Neighbors group and ten participants from the other group were compared in this study. To determine the total sample size, power software was used with the following parameters: 0.1 enrollment ratio, 98% pre-test power, and a 95% confidence interval. A 0.05 alpha threshold was used. With 99% accuracy and 89% specificity, the suggested method performs noticeably better than the most advanced Convolutional Neural Network and the K-Nearest Neighbors algorithm. The statistical study indicated that the specificity level was p=0.045 and the accuracy level was p=0.005. When it comes to brain stroke classification, the new Convolutional Neural Network classifiers outperform K-Nearest Neighbors. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF