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 Video

Updated: Dec 10, 2025

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

GENERALIZABLE MULTI-SITE TRAINING AND TESTING OF DEEP NEURAL NETWORKS USING IMAGE NORMALIZATION.

John A Onofrey¹, Dana I Casetti-Dinescu¹, Andreas D Lauritzen¹

¹Department of Radiology & Biomedical Imaging, Yale University, New Haven, CT, USA.

Proceedings. IEEE International Symposium on Biomedical Imaging

|September 3, 2020

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

End to end AI system for surgical gesture sequence recognition and clinical outcome prediction.

NPJ digital medicine·2026

Same author

Enhancing foundation model transfer for prostate cancer detection with patch-level contrastive learning.

NPJ digital medicine·2026

Same author

STNAGNN: Data-driven Spatio-temporal Brain Connectivity beyond FC.

Proceedings of machine learning research·2026

Same author

AI model predicts patient outcomes from surgical gestures and provides insights into explainability.

npj digital surgery·2026

Same author

Editorial Comment.

The Journal of urology·2026

Same author

Prostate Cancer Detection on Micro-Ultrasound Raw Data Using a Deep Learning Neural Network.

Ultrasound in medicine & biology·2026

Deep learning models for medical image analysis struggle to generalize across different hospitals. Evaluating models on data from multiple sites is crucial for reliable prostate segmentation using MRI scans.

Area of Science:

Medical image analysis
Deep learning
Radiology

Background:

Generalizability of deep learning models in medical imaging is crucial for clinical use.
MRI data intensity varies significantly across sites, scanners, and individuals.
Training deep learning models with data from all possible sources is impractical.

Purpose of the Study:

To evaluate the effectiveness of five image normalization methods for training deep neural networks.
To assess the generalization capabilities of models trained for prostate gland segmentation in MRI.
To determine the importance of intra-site and inter-site evaluation for model robustness.

Main Methods:

Trained a deep neural network for prostate gland segmentation using MRI data.

Keywords:

deep learning image segmentation magnetic resonance imaging multi-site evaluation prostate

More Related Videos

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Related Experiment Videos

Last Updated: Dec 10, 2025

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Utilized 600 MRI prostate gland segmentations from two distinct clinical sites.

Compared the performance of models trained with different image normalization techniques.

Main Results:

Training with single-site data resulted in models that failed to generalize to unseen sites.
Both intra-site and inter-site evaluations are critical for assessing model robustness.
Image normalization methods showed varying impacts on model generalization.

Conclusions:

Models trained on single-site MRI data exhibit poor generalization to multi-site data.
Robust evaluation requires assessing model performance within and across different clinical sites.
Intensity normalization is a key consideration for developing generalizable deep learning models for multi-site medical imaging.