Understanding the Impact of Deep Learning Model Parameters on Breast Cancer Histopathological Classification Using...

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Deep learning for diagnostics

Background:

• Artificial intelligence (AI) shows promise for enhancing early and accurate disease diagnosis, such as breast cancer.
• Effective AI in clinical settings requires not only accuracy but also interpretability and reliability.
• Weakly supervised deep learning models are being developed for medical image analysis.

Purpose of the Study:

• To analyze the impact of varying model parameters on the performance of a weakly supervised deep learning model for breast cancer detection.
• To enhance the interpretability and reliability of AI models in medical applications.

Main Methods:

• Utilized Analysis of Variance (ANOVA) to investigate parameter influence on deep learning model performance.
• Employed attention mechanisms within the deep learning model for classification and region identification, enhancing interpretability.
• ANOVA was used to determine the statistical significance of each parameter's effect on model outcomes.

Main Results:

• Identified specific model parameters that significantly influence performance, affecting sensitivity and specificity.
• ANOVA revealed key factors driving the model's image classification accuracy.
• The analysis provided a deeper understanding of model operations and areas for improvement.

Conclusions:

• Applying ANOVA to deep learning models in medical applications yields crucial insights into performance-influencing parameters.
• This analytical approach enhances AI model interpretability and trustworthiness for clinical adoption.
• Understanding parameter effects facilitates the development of more transparent and efficient AI tools for breast cancer detection.