Explainable Artificial Intelligence (XAI) for Deep Learning Based Medical Imaging Classification

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computer Vision

Background:

• Deep learning (AI) models offer high accuracy but suffer from a lack of interpretability (the "black-box" problem).
• Interpretable AI is crucial for reliable medical diagnosis and clinical decision-making.

Purpose of the Study:

• To develop an explainable AI model for medical image classification that enhances decision-making transparency.
• To improve the accuracy and efficiency of AI-driven medical diagnoses.

Main Methods:

• Image segmentation techniques were employed to provide insights into the AI model's classification process.
• The model was evaluated on five diverse medical imaging datasets, including COVID-19 and pneumonia chest X-rays.

Main Results:

• Achieved a testing and validation accuracy of 90.6% on a dataset comprising 6432 images.
• Demonstrated improved accuracy and reduced time complexity compared to traditional AI models.
• The segmentation-based approach enhanced the interpretability of the AI model's predictions.

Conclusions:

• The proposed explainable AI model offers a more transparent and interpretable solution for medical image classification.
• This approach has the potential to increase the accuracy and efficiency of AI in medical diagnosis.
• The model's reduced time complexity makes it a practical tool for clinical applications.