A Deep Learning-Based EffConvNeXt Model for Automatic Classification of Cystic Bronchiectasis: An Explainable AI
Veysi Tekin1, Muhammed Tekinhatun2, Salih Taha Alperen Özçelik3
1Department of Chest Diseases, Faculty of Medicine, Dicle University, Diyarbakır, Turkey.
Journal of Imaging Informatics in Medicine
|September 25, 2025
Summary
A new deep learning model, EffConvNeXt, accurately distinguishes cystic bronchiectasis and pneumonia on chest X-rays. This hybrid model combines EfficientNetB1 and ConvNeXtTiny, improving diagnostic accuracy for critical respiratory conditions.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Respiratory Medicine
Background:
- Cystic bronchiectasis and pneumonia are significant global health concerns impacting morbidity and mortality.
- Accurate and timely diagnosis of these respiratory conditions is essential for improving patient outcomes.
- Overlapping features on chest X-rays (CXRs) present diagnostic challenges, necessitating advanced analytical tools.
Purpose of the Study:
- To develop and evaluate a novel deep learning model, EffConvNeXt, for enhanced classification of cystic bronchiectasis, pneumonia, and normal cases from CXRs.
- To leverage a hybrid approach combining EfficientNetB1 and ConvNeXtTiny to improve diagnostic accuracy and efficiency in medical image analysis.
- To address the limitations of individual deep learning models by integrating their strengths for superior performance in CXR interpretation.
Main Methods:
- The study proposed the EffConvNeXt model, a hybrid architecture integrating EfficientNetB1 and ConvNeXtTiny.
- The model was trained and validated using a dataset of 5899 CXR images from Dicle University Medical Faculty.
- Performance was evaluated by comparing EffConvNeXt against individual models (ConvNeXtTiny and EfficientNetB1) and other deep learning models.
Main Results:
- The individual ConvNeXtTiny model achieved 97.12% accuracy, and EfficientNetB1 achieved 97.79% accuracy.
- The proposed EffConvNeXt model demonstrated a superior accuracy rate of 98.25%, representing a 0.46% improvement over the best individual model.
- EffConvNeXt outperformed other tested deep learning models in classifying CXR images for cystic bronchiectasis and pneumonia.
Conclusions:
- The EffConvNeXt model offers a reliable and automated solution for differentiating cystic bronchiectasis and pneumonia on CXRs.
- The hybrid deep learning approach enhances diagnostic accuracy, supporting clinical decision-making in respiratory disease diagnosis.
- This advanced model shows significant potential for rapid and precise analysis of medical imaging in clinical settings.
