Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{195241,
        author = {Brahmam Odugu and Rohith Odugu and Herlin Kaur and Yuvrat Mittal},
        title = {Optimizing Early Disease Detection via Multi-Scale Ensemble CNNs: A Reproducible Framework for Medical Imaging Analytics},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {10},
        pages = {7313-7325},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=195241},
        abstract = {Early disease detection is a critical factor in improving patient prognosis and reducing healthcare burden, particularly in conditions such as cancer, cardiovascular disorders, and neurological diseases. Advances in deep learning, especially convolutional neural networks (CNNs), have significantly enhanced the capabilities of automated medical image analysis. However, conventional CNN-based approaches are often limited by single-scale feature extraction, reduced sensitivity to subtle abnormalities, and lack of robustness across heterogeneous datasets. Additionally, reproducibility challenges in medical artificial intelligence (AI) hinder reliable clinical translation.
This study proposes a Multi-Scale Ensemble Convolutional Neural Network (MSE-CNN) framework designed to address these limitations. The framework integrates multi-scale feature extraction with ensemble learning to capture both fine-grained local features and global contextual information from medical images. Multiple CNN architectures operating at different spatial resolutions are combined using a weighted ensemble strategy, enabling improved generalization and reduced prediction variance. Furthermore, a reproducibility and methodological alignment (RMA) framework is incorporated to standardize data preprocessing, model training, and evaluation protocols.
Experimental evaluation on benchmark datasets, including ChestX-ray14 and BraTS, indicates that the proposed framework achieves consistent improvements in sensitivity, F1-score, and area under the receiver operating characteristic curve (AUROC) compared to baseline models such as ResNet, DenseNet, and EfficientNet. The results suggest that the integration of multi-scale learning and ensemble strategies enhances the detection of small and early-stage abnormalities while maintaining robustness across datasets.
In conclusion, the MSE-CNN framework provides a unified and reproducible approach for improving early disease detection in medical imaging. The proposed methodology has the potential to support clinical decision-making and advance the deployment of AI-driven diagnostic systems in real-world healthcare settings.},
        keywords = {Medical Imaging, Deep Learning, Multi-Scale CNN, Ensemble Learning, Early Disease Detection, Reproducibility, Healthcare AI},
        month = {March},
        }