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

Stabilized binary hierarchic classifier in cytopathologic diagnosis.

H G Bartels, P H Bartels, M Bibbo

    Analytical and Quantitative Cytology
    |December 1, 1984
    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

    A PC-based system for the objective analysis of histologic specimens through quantitative contextual karyometry.

    Applied optics·2010
    Same author

    Knowledge engineering in quantitative histopathology.

    Applied optics·2010
    Same author

    Karyometry in the early detection and chemoprevention of intraepithelial lesions.

    European journal of cancer (Oxford, England : 1990)·2005
    Same author

    Study of viral integration of HPV-16 in young patients with LSIL.

    Journal of clinical pathology·2003
    Same author

    Transcontinental communication and quantitative digital histopathology via the Internet; with special reference to prostate neoplasia.

    Journal of clinical pathology·2002
    Same author

    Quantitative study of ductal breast cancer progression: nuclear signatures for evaluation of progression grade.

    Advances in clinical pathology : the official journal of Adriatic Society of Pathology·2001
    Same journal

    Chromatin pattern and DNA content in acquired immune deficiency syndrome (AIDS): indicators of viral infection revealed by DNA image cytometry.

    Analytical and quantitative cytology·1984
    Same journal

    Quantitative analysis of heterogeneity of estrogen binding in human breast tumor specimens.

    Analytical and quantitative cytology·1984
    Same journal

    Analysis of number, size and distribution patterns of lipid vacuoles in benign and malignant mesothelial cells.

    Analytical and quantitative cytology·1984
    Same journal

    Automated measurement of acid phosphatase reaction product in human kidneys treated with aminoglycoside antibiotics.

    Analytical and quantitative cytology·1984
    Same journal

    The use of cationic polyelectrolytes in the preparation of cell monolayers for automated cell scanning and diagnostic cytopathology.

    Analytical and quantitative cytology·1984
    Same journal

    Simulation model for gynecologic specimen classification in a high-resolution prescreening system.

    Analytical and quantitative cytology·1984
    See all related articles

    A new binary tree classifier (BTC) algorithm improves cell image analysis by preventing overtraining. This computer-assisted method ensures accurate classification, even with limited data, outperforming traditional classifiers.

    Area of Science:

    • Computer-assisted medical image analysis
    • Machine learning algorithms
    • Pattern recognition in biology

    Background:

    • Overtraining is a common issue in hierarchical classification, especially with limited sample sizes relative to data dimensionality.
    • Truncation effects from preceding decision nodes can negatively impact classification accuracy in complex datasets.
    • Single-stage classifiers often struggle when category mean vectors are not well-separated or covariance matrices are unequal.

    Purpose of the Study:

    • To develop a robust binary tree classifier (BTC) algorithm for computer-assisted cell image analysis.
    • To address and overcome the problem of overtraining in hierarchical decision structures.
    • To enhance classification accuracy in challenging scenarios with non-ideal data distributions.

    Main Methods:

    Related Experiment Videos

    • Implementation of a binary tree classifier (BTC) with specific provisions for representative sampling at each decision node.
    • Ensuring decision rules are based on subpopulations routed to each node, mitigating truncation effects.
    • Comparative analysis against single-stage classifiers under conditions of poor class separation and unequal covariance matrices.

    Main Results:

    • The developed BTC algorithm effectively overcomes overtraining issues common in hierarchical classifiers.
    • Classification accuracy remained statistically consistent between training and test datasets, indicating robustness.
    • The BTC demonstrated superior performance compared to single-stage classifiers in complex, non-ideal data scenarios.

    Conclusions:

    • The binary tree classifier (BTC) offers a significant advancement in computer-assisted cell image analysis.
    • The algorithm's design effectively handles data limitations and complex distributions, leading to reliable classification.
    • BTC provides a more accurate and stable classification solution, particularly valuable in research where test-set performance deterioration is a concern.