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: Mar 1, 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

3.6K

Marginal Shape Deep Learning: Applications to Pediatric Lung Field Segmentation.

Awais Mansoor1, Juan J Cerrolaza1, Geovanny Perez2

  • 1Sheikh Zayed Institute for Pediatric Surgical Innovation, Childrens National Medical Center, Washington DC.

Proceedings of Spie--The International Society for Optical Engineering
|June 9, 2017
PubMed
Summary
This summary is machine-generated.

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

Unlocking 3D baby face photogrammetry: Multi-view BabyMorph reconstruction from uncalibrated photographs.

Expert systems with applications·2026
Same author

LUMEN: LONGITUDINAL MULTI-MODAL RADIOLOGY MODEL FOR PROGNOSIS AND DIAGNOSIS.

Proceedings. IEEE International Symposium on Biomedical Imaging·2026
Same author

Human airway epithelial nasal and bronchial progenitor cell programs are defined in early infancy.

American journal of respiratory cell and molecular biology·2026
Same author

Data-driven personalized surgical treatment selection for metopic and sagittal craniosynostosis.

Journal of neurosurgery. Pediatrics·2026
Same author

Engineering MXenes: Tunable Mechanical Properties and Applications in Structural Systems.

Materials (Basel, Switzerland)·2026
Same author

Generative AI for spatial tumor growth on MRI: a proof-of-principle study in pediatric diffuse midline glioma.

BMC medicine·2026

This study introduces Marginal Shape Deep Learning (MaShDL) for segmenting deformable objects in medical images. MaShDL improves upon traditional methods by robustly estimating shape parameters, achieving higher accuracy in lung field segmentation.

Area of Science:

  • Medical Imaging
  • Computer Vision
  • Machine Learning

Background:

  • Deep learning (DL) excels at image analysis but struggles with deformable object segmentation.
  • Existing DL methods are limited to pixel classification, not parameter estimation for shape.
  • Classical shape models are prone to local minima and illumination variations.

Purpose of the Study:

  • To develop a novel framework, Marginal Shape Deep Learning (MaShDL), for deformable shape segmentation.
  • To integrate DL's feature learning with statistical shape models for robust parameter estimation.
  • To address the limitations of iterative methods and improve runtime performance.

Main Methods:

  • Proposed MaShDL framework combines DL classifiers with statistical shape models.
Keywords:
chest radiographdeep learninglung fieldshape learningstatistical shape models

More Related Videos

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging
10:44

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

1.3K

Related Experiment Videos

Last Updated: Mar 1, 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

3.6K
Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging
10:44

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

1.3K
  • Shape parameters are estimated by learning classifiers in marginal eigenspaces.
  • Evaluated on segmenting lung fields in 314 pediatric chest radiographs.

    • Main Results:

    • MaShDL achieved a mean Dice similarity coefficient of 0.927 for lung field segmentation.
    • This significantly outperformed classical Active Shape Models (ASM) which achieved 0.888 (p=0.01).
    • The method demonstrated robustness to local minima and illumination changes.

    Conclusions:

    • MaShDL is the first DL framework for parametric shape learning in deformable object delineation.
    • The framework offers improved accuracy and robustness compared to traditional methods.
    • MaShDL provides an efficient solution for complex multi-parameter estimation problems in medical imaging.