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

On evaluating brain tissue classifiers without a ground truth.

Sylvain Bouix1, Marcos Martin-Fernandez, Lida Ungar

  • 1Psychiatry Neuroimaging Laboratory, Department of Psychiatry, Brigham and Women's Hospital, Boston, MA, USA. sylvain@bwh.harvard.edu

Neuroimage
|May 29, 2007
PubMed
Summary

Evaluating brain tissue classifiers using common agreement methods offers valuable insights without ground truth. These techniques effectively identify outliers and assess classifier consistency in MR image analysis.

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

Plasma Phosphorylated Tau 217 in Participants at Risk for Chronic Traumatic Encephalopathy.

JAMA network open·2026
Same author

Shared and specific associations of amygdala nuclei volumes with PTSD symptom domains and childhood trauma: An ENIGMA-PGC PTSD mega-analysis.

Molecular psychiatry·2026
Same author

Youth Soccer Participation and Brain Health Outcomes in Adolescent Athletes.

JAMA network open·2026
Same author

Comparative profiles of anomalous self-experiences and social cognition in clinical high risk for psychosis and autism spectrum disorder.

Schizophrenia research. Cognition·2026
Same author

Brain regional susceptibility to tauopathy in individuals at risk for chronic traumatic encephalopathy.

Alzheimer's & dementia : the journal of the Alzheimer's Association·2026
Same author

Dose-dependent white matter changes associated with repetitive head impacts in former American football players.

Brain communications·2026

Area of Science:

  • Medical Imaging
  • Computer Vision
  • Neuroscience

Background:

  • Accurate brain tissue classification from MR images is crucial for neurological studies.
  • Establishing a definitive ground truth for MR image segmentation is challenging.

Purpose of the Study:

  • To present and evaluate techniques for assessing brain tissue classifiers without relying on ground truth data.
  • To compare the effectiveness of common agreement principles versus ground truth-based evaluation.

Main Methods:

  • Utilized Williams' index for common agreement measurement.
  • Applied STAPLE (Simultaneous Truth and Performance Level Estimation) algorithm.
  • Employed Multidimensional Scaling for similarity data visualization.
  • Evaluated eleven segmentation algorithms on forty MR images.

Related Experiment Videos

Main Results:

  • Common agreement methods successfully detected outliers and discriminated between consistent and variable classifiers.
  • These techniques provided insights into the overall similarity between different segmentation algorithms.
  • Comparison with ground truth revealed that common agreement alone may not capture all performance nuances.

Conclusions:

  • Evaluation of brain tissue classifiers using common agreement principles is informative, especially when ground truth is unavailable.
  • These methods aid in identifying classifier performance characteristics like consistency and variability.
  • While valuable, common agreement metrics may not fully replace expert-based ground truth for precise performance evaluation.