A novel pruning-enhanced hybrid approach for efficient and accurate brain tumor diagnosis

Accurate and efficient brain tumor diagnosis is a persistent challenge in medical imaging due to the complexity and variability of tumor structures. Deep learning has demonstrated strong potential; however, balancing diagnostic performance with computational efficiency remains a significant barrier...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Biomedical signal processing and control Ročník 112; s. 108466
Hlavní autor: Dogan, Yahya
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Ltd 01.02.2026
Témata:
Brain tumor diagnosis
Genetic algorithm
Lion Optimization Algorithm
Model pruning
Particle Swarm Optimization
Transfer learning
Brain tumor diagnosis
Model pruning
Transfer learning
Genetic algorithm
Lion Optimization Algorithm
Particle Swarm Optimization
ISSN:1746-8094
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Accurate and efficient brain tumor diagnosis is a persistent challenge in medical imaging due to the complexity and variability of tumor structures. Deep learning has demonstrated strong potential; however, balancing diagnostic performance with computational efficiency remains a significant barrier to real-world deployment, particularly in resource-limited clinical settings. This study introduces a novel hybrid framework combining a pruning-enhanced EfficientNetV2B3 model with metaheuristic-based feature selection and traditional machine learning classifiers. Multiple state-of-the-art CNN architectures were initially evaluated, and EfficientNetV2B3 was selected as the most effective model. A sparsity-driven threshold-based pruning strategy was then applied, reducing parameters by approximately 78% (from 12.9M to 2.86M) while retaining competitive classification accuracy. Feature vectors extracted from the pruned model were refined using Genetic Algorithm, Particle Swarm Optimization, and Lion Optimization Algorithm. The pruned EfficientNetV2B3 model maintained high performance, achieving 97.70% ± 0.11 accuracy. Among all evaluated classifiers, Random Forest combined with genetic Algorithm-selected features yielded the best results: 98.64% ± 0.25 accuracy, 98.67% ± 0.24 precision, and 98.66% ± 0.26 F1-score. Additionally, visualization techniques such as t-SNE and Grad-CAM were used to confirm feature separability and enhance model interpretability. The proposed hybrid framework demonstrates that significant model compression can be achieved without compromising diagnostic accuracy. It offers a computationally efficient and interpretable solution suitable for deployment in real-time and low-resource clinical environments.
ISSN:1746-8094
DOI:10.1016/j.bspc.2025.108466