Summary of Deep Fusion Model For Brain Tumor Classification Using Fine-grained Gradient Preservation, by Niful Islam et al.
Deep Fusion Model for Brain Tumor Classification Using Fine-Grained Gradient Preservation
by Niful Islam, Mohaiminul Islam Bhuiyan, Jarin Tasnim Raya, Nur Shazwani Kamarudin, Khan Md Hasib, M. F. Mridha, Dewan Md. Farid
First submitted to arxiv on: 28 Jun 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel architecture for precise brain tumor classification is proposed, fusing pretrained ResNet152V2 and modified VGG16 models. The architecture undergoes fine-tuning to preserve fine gradients in deep neural networks, essential for effective brain tumor classification. The model achieves an accuracy of 98.36% and 98.04% on Figshare and Kaggle datasets, respectively. The architecture is streamlined, with only 2.8 million trainable parameters, making it suitable for edge devices. Quantization reduces the model size to 73.881 MB from its original size of 289.45 MB. Grad-CAM improves interpretability, offering insights into the decision-making process. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Brain tumors are a common cause of early death if not diagnosed early. Traditional diagnostic methods take too long and can be inaccurate. Computer vision-based approaches have shown promise in brain tumor classification, but existing models are often too big to deploy in areas with limited technology. This research aims to create an accurate and fast way to classify brain tumors that works even in underdeveloped regions. The proposed architecture combines two well-known models (ResNet152V2 and VGG16) and fine-tunes them for better results. The model is very good, achieving 98% accuracy on two different datasets. It’s also small enough to work on devices with limited resources. |
Keywords
» Artificial intelligence » Classification » Fine tuning » Quantization
