Fibonacci-Net: A Lightweight CNN model for Automatic Brain Tumor Classification

TL;DR

This work tackles automatic brain tumor classification from imbalanced MRI datasets by introducing Fibonacci-Net, a lightweight CNN whose convolutional filters follow Fibonacci numbers to drastically reduce parameters. It incorporates two parallel concatenation blocks with an in-network Avg-2Max pooling technique to augment features and mitigate class imbalance, aided by depth-wise separable convolutions in the last blocks and global average pooling. The approach is trained from scratch and validated on three MRI datasets with 4, 15, and 44 classes, achieving state-of-the-art or near state-of-the-art accuracy (up to 96.2% on the 44-class dataset) while maintaining efficiency and good generalization, as shown by ablation studies and Grad-CAM explainability. Overall, the method offers a practical, parameter-efficient solution for robust brain tumor classification in imbalanced medical imaging settings, with interpretable AI supporting its focus on tumor regions.

Abstract

This research proposes a very lightweight model "Fibonacci-Net" along with a novel pooling technique, for automatic brain tumor classification from imbalanced Magnetic Resonance Imaging (MRI) datasets. Automatic brain tumor detection from MRI dataset has garnered significant attention in the research community, since the inception of Convolutional Neural Network (CNN) models. However, the performance of conventional CNN models is hindered due to class imbalance problems. The novelties of this work are as follows: (I) A lightweight CNN model is proposed in which the number of filters in different convolutional layers is chosen according to the numbers of Fibonacci series. (II) In the last two blocks of the proposed model, depth-wise separable convolution (DWSC) layers are employed to considerably reduce the computational complexity of the model. (III) Two parallel concatenations (or, skip connections) are deployed from 2nd to 4th, and 3rd to 5th convolutional block in the proposed Fibonacci-Net. This skip connection encompasses a novel Average-2Max pooling layer that produces two stacks of convoluted output, having a bit different statistics. Therefore, this parallel concatenation block works as an efficient feature augmenter inside the model, thus, automatically alleviating the class imbalance problem to a certain extent. For validity purpose, we have implemented the proposed framework on three MRI datasets which are highly class-imbalanced. (a) The first dataset has four classes, i.e., glioma tumor, meningioma tumor, pituitary tumor, and no-tumor. (b) Second and third MRI datasets have 15 and 44 classes respectively. Experimental results reveal that, after employing the proposed Fibonacci-Net we have achieved 96.2% accuracy, 97.17% precision, 95.9% recall, 96.5% F1 score, and 99.9% specificity on the most challenging ``44-classes MRI dataset''.

Figures (5)

• Figure 1: Block Diagram of the proposed Fibonacci Net
• Figure 2: The first row indicates original MRI images; the second row are the down-sampled (/2) images after passing it through Avg-2Max pooling
• Figure 3: Basic units of the proposed model (a) parallel concatenation vs (b) sequential model
• Figure 4: $1^{st}$ and $2^{nd}$ graphs represent validation accuracy and validation loss (vs number of epochs) for 44-class dataset, $3^{rd}$ graph shows Macro-average of Precision, Recall and F1 score for several models on 44-class dataset
• Figure 5: Validity checking by Explainable AI: First column represents the original image, $2^{nd}$ column represents GradCam plot by Proposed "Fibonacci Net with pcb", $3^{rd}$ column represents GradCam plot by Proposed "Fibonacci Net without pcb"