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: May 19, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Deep ensemble optimized models for probabilistic CTV breast segmentation.

Cecilia Riani^1,2, Maria Giulia Ubeira-Gabellini¹, Gabriele Palazzo¹

¹Medical Physics, IRCCS San Raffaele Scientific Institute, Milan, Italy.

Frontiers in Artificial Intelligence

|May 18, 2026

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

Development and validation of a knowledge-based model for robotic radiosurgery planning for brain lesions.

Physics and imaging in radiation oncology·2026

Same author

The 'CIED Team' Approach to Device-Related Complications.

JACC. Case reports·2026

Same author

Radiosurgery for PRL-secreting pituitary adenomas refractory to medical and surgical treatment.

European journal of endocrinology·2026

Same author

Combined minibeam and conventional radiotherapy reduces severe skin toxicity and improves tumor control.

Scientific reports·2026

Same author

National survey to update the Italian diagnostic reference levels in interventional cardiology.

European journal of radiology·2026

Same author

Association of Cardiac and Pulmonary CT Imaging Features with Respiratory Side Effects After Whole-Breast Radiotherapy.

Cancers·2026

This summary is machine-generated.

Six deep learning models were evaluated for automatic clinical target volume (CTV) segmentation in whole-breast radiotherapy. Top models achieved accuracy comparable to inter-observer variability, enabling the creation of probability maps to reduce uncertainty in radiotherapy planning.

Area of Science:

Radiotherapy
Medical Imaging
Artificial Intelligence

Background:

Accurate segmentation of organs-at-risk (OARs) and clinical target volume (CTV) is crucial for optimizing radiotherapy and minimizing toxicity.
Deep learning (DL) models offer high segmentation accuracy but require careful tuning and reliable performance to address clinical contouring variability.

Purpose of the Study:

To systematically evaluate six advanced DL models for automatic CTV segmentation in whole-breast radiotherapy.
To leverage top-performing models for constructing probability maps to improve consistency and mitigate bias in clinical predictions.

Main Methods:

Six DL models (UNet, SegResNetDS, DynUNet, nnU-Net, MedSAM2-A, MedSAM2-B) were trained and tested on CT images for simultaneous right/left breast CTV segmentation.
Performance was assessed using Dice Similarity Coefficient, Average Surface Distance (ASD), and Hausdorff Distance (HD95), with statistical analysis via Friedman and Conover tests.

Keywords:

breast cancer breast radiotherapy deep learning model ensemble probabilistic CTV uncertainty

Related Experiment Videos

Last Updated: May 19, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Model-based probability maps were generated using top-performing models to quantify uncertainties.

Main Results:

All models showed satisfactory performance, comparable to inter-observer variability (Dice = 0.90).
UNet, DynUNet, nnU-Net, and MedSAM2-A achieved the highest accuracy (average ASD = 1.5 mm, HD95 = 3.8 mm).
Probability map analysis revealed significant uncertainty volumes (average difference = 123 cm³) at lateral and cranio-caudal CTV borders.

Conclusions:

Advanced DL models can effectively segment breast CTVs, achieving accuracy comparable to human experts.
The developed probability maps represent a novel approach to visualize and quantify segmentation uncertainties, aiding clinical decision-making.
This work supports the integration of DL for more consistent and reliable radiotherapy planning.