Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

An optimal three-class linear observer derived from decision theory.

Xin He¹, Eric C Frey

¹Department of Radiology, Johns Hopkins School of Medicine, Baltimore, MD 21287, USA. xinhe@jhmi.edu

IEEE Transactions on Medical Imaging

|January 25, 2007

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

An international study of factors affecting variability of dosimetry calculations, part 5: impact of segmentation methods.

EJNMMI physics·2026

Same author

Intraprocedural C-arm dual-phase cone-beam enhancement patterns correlate with tumor absorbed dose after radioembolization.

Medical physics·2023

Same author

Girth-based administered activity for pediatric <sup>99m</sup> Tc-DMSA SPECT.

Medical physics·2023

Same author

Lack of repeatability of radiomic features derived from PET scans: Results from a <sup>18</sup> F-DCFPyL test-retest cohort.

The Prostate·2023

Same author

TransMorph: Transformer for unsupervised medical image registration.

Medical image analysis·2022

Same author

Automated segmentation of lung, liver, and liver tumors from Tc-99m MAA SPECT/CT images for Y-90 radioembolization using convolutional neural networks.

Medical physics·2021

Same journal

Informed-Exploration Reinforcement Learning for Automated Virtual Coronary Intervention Planning.

IEEE transactions on medical imaging·2026

Same journal

4D Reconstruction of Fetal Left Ventricle from Echocardiography via 2.5D Radial Segmentation and Graph-Fourier Reconstruction.

IEEE transactions on medical imaging·2026

Same journal

Generalised Medical Phrase Grounding.

IEEE transactions on medical imaging·2026

Same journal

EndoLRMGS: Combining Large Reconstruction Modelling and Gaussian Splatting for Complete Endoscopic Scene Reconstruction.

IEEE transactions on medical imaging·2026

Same journal

A Neural-Analytical Fusion Scatter Correction Method for Multi-Source CT Using Equivalent High-Order Scatter.

IEEE transactions on medical imaging·2026

Same journal

SynReEM: Synapse Reconstruction via Instance Structure Encoding in Anisotropic Electron Microscopic Volumes.

IEEE transactions on medical imaging·2026

See all related articles

We developed a three-class Hotelling observer (3-HO) for multiclass data classification. This mathematical observer offers clear optimality and avoids ambiguous decisions, improving diagnostic imaging tasks.

Area of Science:

Decision Theory
Statistical Pattern Recognition
Medical Imaging

Background:

Existing linear observers for multiclass data classification often lack defined optimality or have ambiguous decision regions.
The ideal observer provides a theoretical benchmark for classification performance but is often computationally intractable.

Purpose of the Study:

To derive an optimal linear observer for three-class classification problems.
To develop a mathematical observer with clearly defined optimality and unambiguous decision rules for diagnostic tasks.

Main Methods:

Derived a three-class Hotelling observer (3-HO) inspired by the decision-theoretic ideal observer.
Assumed multivariate Gaussian distributions with equal covariance matrices for initial derivation.
Extended applicability to non-Gaussian data with equal covariance matrices under specific linear relationships between two-class Hotelling templates.

Related Experiment Videos

Main Results:

Under Gaussian and equal covariance assumptions, the 3-HO constructed from two two-class Hotelling templates matches the performance of the three-class ideal observer (3-IO).
The 3-HO remains applicable and maximizes pairwise signal-to-noise ratios (SNR) even without Gaussian or equal covariance assumptions, provided a linear relationship holds.
The developed observer utilizes first- and second-order ensemble data statistics.

Conclusions:

A novel three-class linear mathematical observer (3-HO) has been developed.
The 3-HO demonstrates clear optimality under various conditions and eliminates ambiguous decision regions.
This observer is potentially valuable for optimizing and evaluating imaging techniques in three-class diagnostic applications.