Manuscript received November 27, 2024; revised January 21, 2025; accepted March 6, 2025; Published July 17, 2025.

Abstract—Deep learning has transformed medical image analysis; in particular, many different brain tumors are being produced. Accurate detection is crucial for effective treatment and contributes to prolonging the life of patients. While MRI is a standard diagnostic tool, manual interpretation can be slow and sometimes error prone. As such, automatic classification systems based on Convolutional Neural Networks (CNNs) are gaining importance. However, standard CNNs are generally good for capturing only the continuous features within one region of the data, and that means they need huge amounts of training samples to work well in practice. We also struggle to capture diverse shapes, sizes and positions of brain tumors. To meet this challenge, our paper proposes the SPP-MobileNet model, which integrates a Spatial Pyramid Pooling (SPP) block into the MobileNet architecture. With the SPP layer for multi-scale feature extraction, our classifier is much better at spotting tumor appearance changes without resizing images. By building this into MobileNet, SPP-MobileNet maintains all that model’s computational efficiency while boosting classification accuracy. On two Magnetic Resonance Imaging (MRI) datasets, the proposed model outperformed other state-of-the-art methods with an accuracy of 98.86% and perfect precision. Its recall rate was 97.68%, while the Matthews Correlation Coefficient value reached 97.75 %. These results suggest that SPP-MobileNet is a powerful tool for brain tumor classification, and it should go some way to improving diagnostic accuracy and speed. In the future, we will focus on tuning the model for more complex types of brain tumors and applying it across various other medical imaging tasks.

Keywords—brain tumor classification, MRI image analysis, neuroimaging, SPP-MobileNet

Cite: Karrar A. Kadhim, Ola N. Kadhim, and Fallah H. Najjar, "Integrating Spatial Pyramid Pooling for Multi-scale Brain Tumor Classification in Deep Learning," Journal of Image and Graphics, Vol. 13, No. 4, pp. 315-324, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.