Related Experiment Videos
CCEO-DCABNet: Chronological Chaotic Evolution Optimization-Enabled Hybrid Deep Learning for Multiclass Disease
Leena Patil1,2, Bindu Garg1, Massimo Donelli3
1Bharati Vidyapeeth (Deemed to be) University, College of Engineering, Pune 411043, India.
Diagnostics (Basel, Switzerland)
|July 15, 2026
Summary
This study introduces a novel deep learning model, CCEO-DCABNet, for accurate multiclass lung disease classification from chest X-rays. The federated learning approach ensures data privacy while achieving high diagnostic performance.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer-Aided Diagnosis
Background:
- Chest X-rays are crucial for diagnosing lung diseases, but classification is challenging due to data heterogeneity, overlapping features, and privacy concerns.
- Distinguishing between various lung diseases is difficult owing to similar clinical and imaging characteristics.
Purpose of the Study:
- To propose a novel deep channel-attention broad convolutional neural network (CCEO-DCABNet) for accurate multiclass lung disease classification.
- To address data privacy challenges using a federated learning (FL) framework.
- To enhance classification performance through chaotic evolution optimization.
Main Methods:
- Implemented a federated learning framework for collaborative model training without raw data sharing.
- Applied Gaussian filter denoising and multiscale unsharp masking for image preprocessing.
- Utilized a deep channel-attention broad convolutional neural network (DCABNet) optimized by chaotic evolution optimization (CCEO).
Main Results:
- The CCEO-DCABNet model achieved high performance metrics: 96.98% accuracy, 96.41% true positive rate (TPR), and 97.45% true negative rate (TNR).
- Demonstrated effective multiclass classification of lung diseases from chest X-ray images.
Conclusions:
- The proposed CCEO-DCABNet framework effectively classifies multiple lung diseases while preserving data privacy via federated learning.
- The model shows superior classification performance, supporting reliable computer-aided diagnosis in clinical settings.