Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Dec 16, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

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

Published on: November 30, 2022

3.2K

Deep Learning Pre-training Strategy for Mammogram Image Classification: an Evaluation Study.

Kadie Clancy1, Sarah Aboutalib2, Aly Mohamed3

  • 1Department of Computer Science, University of Pittsburgh, 3240 Craft Place, Pittsburgh, PA, 15213, USA.

Journal of Digital Imaging
|July 2, 2020
PubMed
Summary

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

Survivor Perspectives on Artificial Intelligence Integration in Breast Cancer Treatment: A Qualitative Study of Trust, Equity, and Application.

European journal of breast health·2026
Same author

<math><mi>Δ</mi> <mi>t</mi></math> -Mamba3D: A Time-Aware Spatio-Temporal State-Space Model for Breast Cancer Risk Prediction.

Proceedings of the ... AAAI Conference on Artificial Intelligence. AAAI Conference on Artificial Intelligence·2026
Same author

Safe Breast Cancer Diagnosis Resilient to Mammographic Adversarial Samples.

Artificial Intelligence and Imaging for Diagnostic and Treatment Challenges in Breast Care : first Deep Breast Workshop, Deep-Breath 2024, held in conjunction with MICCAI 2024, Marrakesh, Morocco, October 10, 2024, Proceedings. Deep-Bre...·2026
Same author

Mixture of Experts Made Personalized: Federated Prompt Learning for Vision-Language Models.

... International Conference on Learning Representations·2026
Same author

Interim Analysis of Dose-Escalated Preoperative Breast Irradiation: A Phase Ib study.

medRxiv : the preprint server for health sciences·2025
Same author

National Emergency Bariatric Surgical Audit (NEBSA): a protocol for a multi-center prospective study of unplanned interventions following emergency bariatric surgery.

International journal of surgery protocols·2025

Pre-training strategy significantly impacts deep learning performance in digital mammography for identifying breast cancer. The best model, AlexNet, achieved AUCs from 0.68 to 0.77, highlighting the importance of effective training approaches.

Area of Science:

  • Radiology and Medical Imaging
  • Artificial Intelligence in Healthcare
  • Deep Learning for Medical Diagnosis

Background:

  • Accurate interpretation of digital mammograms is crucial for breast cancer screening.
  • Deep learning models show promise in analyzing medical images but require optimal training strategies.
  • Distinguishing false recalls (benign findings) from malignancy is a key challenge in mammography interpretation.

Purpose of the Study:

  • To evaluate the impact of various pre-training strategies on the performance of deep learning models for digital mammography analysis.
  • To identify the most effective pre-training approaches for classifying mammogram findings, including malignancy, benign, and false recalls.
  • To compare the performance of different convolutional neural network (CNN) architectures and pre-training datasets.
Keywords:
Breast cancerDeep learningDigital mammographyTraining strategyTransfer learning

Related Experiment Videos

Last Updated: Dec 16, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

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

Published on: November 30, 2022

3.2K

Main Methods:

  • Retrospective analysis of 1303 patients and 4935 digital mammogram images.
  • Assessment of six CNN model structures using four diverse pre-training datasets (>1.4 million images).
  • Evaluation of transfer learning, layer freezing, varied network structures, and multi-view input strategies for binary and triple-class classification.
  • Performance measured using the area under the receiver operating characteristic curve (AUC).

Main Results:

  • An AlexNet model, incrementally pre-trained on ImageNet and a large Breast Density dataset, demonstrated the best performance.
  • The best model achieved AUCs ranging from 0.68 to 0.77 across six classification tasks.
  • Four out of five tested pre-training strategies significantly improved performance in distinguishing recalled-benign mammograms compared to the baseline.

Conclusions:

  • Pre-training strategy is a critical factor influencing the performance of deep learning models in digital mammography.
  • Effective pre-training can lead to significant improvements, particularly in differentiating benign findings from malignant ones.
  • The findings underscore the need for careful selection and optimization of pre-training methods for medical image analysis tasks.