Deep Learning Approach Outperforms Conventional Methods in Brain Cancer Detection

Strategic Insights -

Researchers at Gandhigram Rural Institute have made a significant breakthrough in brain cancer detection using deep convolutional neural networks (CNNs). The study aimed to develop an optimal brain tumor detection system for high-grade and low-grade glioma lesions from magnetic resonance imaging (MRI) human brain scans. The research concluded that the deep learning approach yielded better results than conventional state-of-the-art methods.

Key Takeaways:

  • The researchers developed six CNN models, each with different hyperparameter settings and architectures, to find the optimal brain tumor detection system.
  • The models were trained and tested on BraTS2013 and whole brain atlas data sets, with the FLSCBN model yielding the best classification results for brain tumor detection.
  • The experimental results revealed that the deep learning approach outperformed conventional state-of-the-art methods.
  • The study used conventional CNN architecture with different combinations and settings of hyperparameters, including conventional CNN architecture, dropout, stopping criteria, batch normalization, and dropout.
  • The research was conducted at Gandhigram Rural Institute's Department of Computer Sciences and Applications, and involved authors Thiruvenkadam Kalaiselvi, Thiyagarajan Padmapriya, Padmanaban Sriramakrishnan, and Venugopal Priyadharshini.
  • The study's findings have implications for the development of more accurate and efficient brain cancer detection systems.

Statistics:

  • 6 CNN models were developed to find the optimal brain tumor detection system.
  • The models were trained and tested on 2 data sets: BraTS2013 and whole brain atlas.
  • The FLSCBN model yielded the best classification results for brain tumor detection.
  • The study used conventional CNN architecture with different combinations and settings of hyperparameters.
  • The deep learning approach outperformed conventional state-of-the-art methods in brain cancer detection.

Sources:

  • Advanced study results on Oncology - Brain Cancer have been published. (OCT 19, 2020)
  • Development of Automatic Glioma Brain Tumor Detection System Using Deep Convolutional Neural Networks (International Journal of Imaging Systems and Technology, 2020).
  • NewsRx. New Brain Cancer Study Findings Have Been Reported by Investigators at Gandhigram Rural Institute (Development of Automatic Glioma Brain Tumor Detection System Using Deep Convolutional Neural Networks). Journal of Engineering. October 19, 2020; p 1456