On the detection of pneumonia using deep learning and explainability
Loading...
Date
2024-06
Journal Title
Journal ISSN
Volume Title
Publisher
University Mohamed Boudiaf of M’sila
Abstract
Pneumonia detection from chest X-ray images is achieved using the pre-trained VGG16 deep
learning model, known for its ability to extract relevant features from images. The process
involves feeding the X-ray images into the VGG16 model, which then analyzes and classifies
them to detect signs of pneumonia. To build trust in the model’s decisions, explainability
techniques such as Causal Explanation CNN (CexCNN) and CAM are employed. These
techniques provide visual explanations by highlighting the regions of the X-ray images that the
model focuses on when making its predictions. CexCNN offers casual robust explanations,
indicating which features are most influential in the decision-making process, thus leading to
gain confidence in the CNN model. Which is very crucial especially in the medical field.
Description
Keywords
Deep learning, CNN, CexCNN, VGG16, ImageNet, Computer vision, heatmaps, Boundingbox, X-Ray