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

Related Concept Videos

Absorption of Radiation

Absorption of Radiation

The rate of heat transfer by emitted radiation is described by the Stefan-Boltzmann law of radiation:

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Kolmogorov-Arnold Guided Local-Global Attention for Medical Image Classification.

Journal of imaging informatics in medicine·2026

See all related articles

Search research articles

Related Experiment Video

Updated: Jan 13, 2026

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

TET Loss: A Temperature-Entropy Calibrated Transfer Loss for Reliable Medical Image Classification.

¹School of Computer and Artificial Intelligence, Shandong Jianzhu University, No.1000 Fengming Road, Ganggou Subdistrict, Jinan, Shandong, 250101, People's Republic of China. panweichao01@outlook.com.

Journal of Imaging Informatics in Medicine

|January 6, 2026

Summary

TET Loss improves deep learning for medical imaging by calibrating confidence and reducing overconfidence. This novel approach enhances model reliability and interpretability without adding computational cost.

Keywords:

Entropy regularization ImageNet Medical image classification Temperature scaling Transfer learning

More Related Videos

Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging

Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging

Published on: May 5, 2011

Related Experiment Videos

Last Updated: Jan 13, 2026

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging

Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging

Published on: May 5, 2011

Area of Science:

Artificial Intelligence
Medical Imaging Analysis
Computer Vision

Background:

Deep learning models for medical image classification often suffer from overconfident predictions and domain mismatch from natural image pretraining.
This impacts their generalization capabilities and clinical reliability in real-world applications.

Purpose of the Study:

To introduce TET Loss (Temperature-Entropy calibrated Transfer Loss Function), a novel objective function designed to enhance the reliability and robustness of deep learning models in medical image classification.
To address overconfidence and domain mismatch issues inherent in transfer learning for medical imaging.

Main Methods:

Proposes TET Loss, a plug-and-play objective function combining temperature scaling for logit moderation and entropy regularization for uncertainty-aware learning.
The method is model-agnostic and adds no inference-time overhead.
Evaluated across four public benchmarks (BreastMNIST, DermaMNIST, PneumoniaMNIST, RetinaMNIST) using CNNs, transformers, and hybrid backbones with short 10-epoch fine-tuning.

Main Results:

TET Loss consistently improved performance across various architectures and datasets.
EfficientViT-M2 on BreastMNIST saw an F1 score increase from 53.9% to 66.7%.
BiFormer-Tiny achieved an F1 of 86.1% with an AUC of 94.1% on BreastMNIST.
RMT-T3 on PneumoniaMNIST reached an F1 of 96.4% and an AUC of 99.1%, outperforming models trained longer.
Grad-CAM visualizations indicated improved lesion localization and reduced spurious activations, suggesting enhanced interpretability.

Conclusions:

TET Loss offers a lightweight and effective solution for improving the reliability and robustness of medical imaging systems by calibrating prediction confidence.
The method enhances model generalization and interpretability without increasing computational complexity.
TET Loss represents a significant advancement towards more trustworthy AI in clinical settings.