A Class-Wise Deep Ensemble Framework Using ResNet101 and DenseNet201 for Brain Tumor Classification.

Motea Alsamawi¹, Waled Hussein Al-Arashi², Mohammed M Alkhawlani²

¹Department of Biomedical Engineering, University of Science and Technology, Sana'a, Yemen. moteaibir@gmail.com.

Journal of Imaging Informatics in Medicine

|December 1, 2025

Summary

Related Concept Videos

01:26

Classification of Systems-I

Linearity is a system property characterized by a direct input-output relationship, combining homogeneity and additivity.
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:

You might also read

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

An audit of root canal filling quality performed by undergraduate pre-clinical dental students, Yemen.

BMC medical education·2019

See all related articles

This study introduces an ensemble learning framework for brain tumor classification, achieving 96.50% accuracy. The class-wise ensemble method significantly improves diagnostic precision for brain lesions.

Area of Science:

Medical Image Analysis
Artificial Intelligence in Medicine
Machine Learning for Diagnostics

Background:

Accurate brain tumor classification is crucial for effective medical diagnosis and treatment planning.
Existing methods may have limitations in achieving high diagnostic accuracy for diverse brain lesions.
Deep learning models like ResNet101 and DenseNet201 show promise but can be further optimized.

Purpose of the Study:

To develop and evaluate an ensemble learning framework for enhanced brain tumor classification.
To integrate ResNet101 and DenseNet201 architectures using a class-wise model selection strategy.
To improve the accuracy and reliability of automated brain tumor diagnosis from MRI data.

Main Methods:

An ensemble learning framework was designed, integrating ResNet101 and DenseNet201.

Keywords:

Brain tumor diagnosis Class-wise selection Convolutional neural network Deep learning Ensemble deep learning Tumor classification Tumor detection

Models were trained on a 7041-image brain MRI dataset with an 80/20 split.

Image preprocessing included resizing, CLAHE, and sharpening for enhanced feature extraction.

Main Results:

The proposed class-wise ensemble achieved an overall accuracy of 96.50%.
Performance metrics included precision (96.43%), specificity (98.84%), recall (96.20%), and F1 score (96.25%).
The ensemble significantly outperformed individual ResNet101 and DenseNet201 models.

Conclusions:

The class-wise ensemble learning framework demonstrates robustness and efficiency in brain tumor classification.
This approach offers a reliable and highly precise method for automated brain tumor diagnosis.
Ensemble methods, particularly class-wise selection, can significantly enhance diagnostic performance in medical imaging.

Related Experiment Video