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: Jan 8, 2026

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

3.3K

The impact of pre-processing techniques on deep learning breast image segmentation.

Jéssica Catarino1,2, Nuno Cruz Garcia3, Sara Silva3

  • 1LASIGE, Faculdade de Ciências, Universidade de Lisboa, 1749-016, Lisbon, Portugal. jessica.catarino@research.fchampalimaud.org.

Scientific Reports
|December 16, 2025
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

Semi-automated detection of cleaning interactions using supervised machine learning.

Scientific reports·2026
Same author

Polymeric Nanogels for Skin Applications.

Gels (Basel, Switzerland)·2026
Same author

Digital twins for non-invasive tumor localization in breast-conserving surgery: a technical pilot study.

Breast (Edinburgh, Scotland)·2026
Same author

Editorial: Natural bioactives: a promising avenue for metabolic syndrome therapy.

Frontiers in nutrition·2026
Same author

Primer on Modelling Approaches for Omics Data.

Advances in experimental medicine and biology·2026
Same author

Western diet suppresses canonical intestinal stem cells and reprograms c-Kit⁺ reserve stem cells via proinflammatory dysbiosis.

bioRxiv : the preprint server for biology·2026
This summary is machine-generated.

Pre-processing techniques significantly impact Deep Learning model performance for breast cancer image segmentation. Tailored strategies, especially pixel intensity normalization, enhance segmentation accuracy in breast imaging analysis.

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Oncology

Background:

  • Breast cancer is a leading global health concern, necessitating advanced diagnostic tools.
  • Deep Learning (DL) methods show promise in medical image analysis, particularly for segmentation tasks.
  • The influence of pre-processing on DL model performance in breast imaging is not well-established.

Purpose of the Study:

  • To investigate the impact of various pre-processing techniques on DL model performance for breast image segmentation.
  • To compare a Domain Non-Specific pipeline with a Domain Specific pipeline for breast imaging.
  • To identify optimal pre-processing strategies for improving breast image segmentation accuracy.

Main Methods:

  • A U-Net segmentation model was applied to two public breast imaging datasets (CBIS-DDSM and Duke-Breast-Cancer-MRI).

More Related Videos

Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

10.5K

Related Experiment Videos

Last Updated: Jan 8, 2026

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

3.3K
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

10.5K
  • Systematic evaluation of pre-processing techniques including normalization, harmonization, resizing, and standardization.
  • Development and comparison of Domain Non-Specific and Domain Specific pre-processing pipelines.

    • Main Results:

    • Different pre-processing techniques led to significant variations in U-Net segmentation outcomes.
    • Pixel intensity normalization approaches showed a notable impact on model performance, confirmed by a 3-way ANOVA F-test.
    • The Domain Specific pipeline, preserving anatomical information, demonstrated potential for improved segmentation.

    Conclusions:

    • Pre-processing is a critical determinant of DL model performance in breast image segmentation.
    • Tailored pre-processing strategies, particularly for pixel intensity normalization, can enhance segmentation accuracy.
    • This study provides foundational insights for optimizing DL-based breast image analysis.