Search research articles

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Multifocal and Multicentric Breast Cancer: Evaluation of Clinicopathological Features and Survival Outcomes.

Sisli Etfal Hastanesi tip bulteni·2026

Same author

Updates on intratumoral therapies in melanoma.

Cancer·2026

Same author

Prognostic value of the shock index in Fournier's gangrene: a retrospective cohort study comparing established mortality scoring systems.

Ulusal travma ve acil cerrahi dergisi = Turkish journal of trauma & emergency surgery : TJTES·2026

Same author

Comparative Analysis of Quantitative and Semiquantitative Ultrafast DCE-MRI, Morphological Findings, and ADC in Breast Lesion Characterization.

Academic radiology·2026

Same author

Virtual Non-Iodine Coronary Calcium Scoring on Photon-Counting CT: Patient- and Plaque-Level Analysis.

Diagnostics (Basel, Switzerland)·2026

Same author

Consistent explainable image quality assessment for medical imaging.

Health information science and systems·2026

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jul 2, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Quantitative evaluation of Saliency-Based Explainable artificial intelligence (XAI) methods in Deep Learning-Based

Esma Cerekci¹, Deniz Alis², Nurper Denizoglu³

¹Sisli Hamidiye Etfal Training and Research Hospital, Department of Radiology, Istanbul, Turkey.

European Journal of Radiology

|February 16, 2024

Summary

Quantitative evaluation of saliency-based Explainable Artificial Intelligence (XAI) methods in breast cancer detection shows limitations. While these methods offer some insight, they often fail to accurately pinpoint diagnostic decisions made by deep learning models.

Keywords:

Breast Cancer Deep Learning Mammogram XAI

More Related Videos

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

Related Experiment Videos

Last Updated: Jul 2, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

Area of Science:

Medical Imaging Analysis
Artificial Intelligence in Healthcare
Radiology

Background:

Explainable Artificial Intelligence (XAI) is crucial for understanding deep learning (DL) models in medical diagnostics.
Saliency methods are common XAI tools in medical imaging, but their performance lacks quantitative validation.
This study addresses the need for rigorous evaluation of these methods in breast cancer detection.

Purpose of the Study:

To quantitatively assess the performance of popular saliency XAI techniques for breast cancer detection using mammograms.
To provide evidence-based insights into the reliability of XAI in clinical diagnostic support.

Main Methods:

A dataset of 1496 mammograms (cancer-positive and negative) was curated, with ground-truth annotations by three radiologists.
A pre-trained deep learning model was utilized for breast cancer detection on MLO and CC views.
Saliency methods (Grad-CAM, Grad-CAM++, Eigen-CAM) were evaluated using the Pointing Game metric against ground-truth bounding boxes.

Main Results:

The deep learning model achieved a recall of 69%, precision of 88%, accuracy of 80%, and F1-Score of 0.77 on the test set.
Pointing Game Scores for Grad-CAM, Grad-CAM++, and Eigen-CAM were 0.41, 0.30, and 0.35, respectively, for all cancer patients.
Scores slightly improved to 0.41, 0.31, and 0.36 when considering only true-positive cases.

Conclusions:

Saliency-based XAI methods offer partial explainability for deep learning models in mammography.
These methods demonstrate limitations in accurately delineating the decision-making process of DL models for breast cancer detection.
Further research is needed to develop more reliable XAI techniques for clinical applications.