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 Videos

Segmentation of anatomical structures in chest radiographs using supervised methods: a comparative study on a public

Bram van Ginneken1, Mikkel B Stegmann, Marco Loog

  • 1Image Sciences Institute, University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. bram@isi.uu.nl

Medical Image Analysis
|May 28, 2005
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

Community-based tuberculosis screening with computer-aided detection technology alone and in combination with point-of-care C-reactive protein testing: a paired screen-positive trial.

The Lancet. Infectious diseases·2026
Same author

Assessment of modifications to a blind-sweep ultrasound protocol for improved lower-uterus imaging by novice operators.

Scientific reports·2026
Same author

Large Language Model Automated Extraction of Clinical Signs and Symptoms From Emergency Department Reports for Machine Learning Prediction Models: Development and Validation Study.

JMIR medical informatics·2026
Same author

Applications and implicit assumptions in dementia risk scores: A scoping review of the LIBRA score.

Journal of Alzheimer's disease reports·2026
Same author

Leveraging open-source large language models for clinical information extraction in resource-constrained settings.

JAMIA open·2025
Same author

Artificial intelligence-based identification of thin-cap fibroatheromas and clinical outcomes: the PECTUS-AI study.

European heart journal·2025

This study compares three automated methods for segmenting lung fields, heart, and clavicles in chest X-rays. Pixel classification achieved human-level performance for lung segmentation, while all methods showed potential for heart and clavicle segmentation.

Area of Science:

  • Medical Imaging
  • Computer Vision
  • Radiology

Background:

  • Accurate segmentation of anatomical structures in chest radiographs is crucial for quantitative analysis.
  • Existing automated methods require rigorous comparison against human performance and each other.

Purpose of the Study:

  • To evaluate and compare three supervised segmentation methods for lung fields, heart, and clavicles in posterior-anterior chest radiographs.
  • To assess the performance of active shape models, active appearance models, and pixel classification against manual segmentations by human observers.

Main Methods:

  • Comparison of active shape models (ASM), active appearance models (AAM), and a multi-resolution pixel classification (PC) method using a filter bank of Gaussian derivatives and k-nearest-neighbors.
  • Parameter optimization for ASM and modifications to AAM by including external areas were investigated.

Related Experiment Videos

  • Evaluation on a public database of 247 manually segmented chest radiographs.

    • Main Results:

    • Pixel classification achieved comparable performance to human observers for lung field segmentation.
    • All methods performed comparably for heart segmentation, but were significantly outperformed by human observers.
    • Clavicle segmentation proved challenging for all automated methods, with ASM performing best but still substantially below human performance.
    • Hybrid systems, particularly majority voting for heart segmentation, showed improved results.
    • Automated cardio-thoracic ratio computation demonstrated good agreement with the gold standard across methods.

    Conclusions:

    • Pixel classification is a robust method for lung field segmentation in chest radiographs, achieving human-level accuracy.
    • While automated heart and clavicle segmentation requires further improvement, hybrid approaches show promise.
    • The study provides publicly available data and results to foster further research in medical image segmentation.