Interpretable chest X-ray localization using principal component-based feature selection in deep learning

doi:10.1016/j.engappai.2025.112358

Interpretable chest X-ray localization using principal component-based feature selection in deep learning

2025-09-27

0

PRE

AI

DOI：10.1016/j.engappai.2025.112358

Original

Original request for help

Save

Abstract

En 中文

• Proposed a novel PCA-based feature selection approach for CNN heatmap generation. • Enhances interpretability while preserving accuracy in chest X-ray localization. • Achieves 98.2% sensitivity, 99.4% specificity, and 97.5% DSC. • Notably improves inference speed, achieving 0.10 ms compared to VGG16 (4.1 ms). • Offers a model-agnostic and efficient alternative to CAM-based methods.

Interpretable chest X-ray localization using principal component-based feature selection in deep learning

Abstract

Journal

Organization