An interactive deep-learning workflow for head and neck gross tumour volume segmentation

Background and purpose: Deep learning (DL)-based auto-segmentation of head and neck cancer (HNC) gross tumour volumes remains challenging due to anatomical complexity and limited accuracy. We propose an interactive DL (iDL) workflow that integrates clinician input to enhance segmentation performance...

Full description

Saved in:
Bibliographic Details
Main Authors: Zixiang Wei, Jintao Ren, Jesper Grau Eriksen, Kenneth Jensen, Hanna Rahbek Mortensen, Stine Sofia Korreman, Jasper Nijkamp
Format: Article
Language:English
Published: Elsevier 2025-07-01
Series:Physics and Imaging in Radiation Oncology
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2405631625001253
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849393164388925440
author Zixiang Wei
Jintao Ren
Jesper Grau Eriksen
Kenneth Jensen
Hanna Rahbek Mortensen
Stine Sofia Korreman
Jasper Nijkamp
author_facet Zixiang Wei
Jintao Ren
Jesper Grau Eriksen
Kenneth Jensen
Hanna Rahbek Mortensen
Stine Sofia Korreman
Jasper Nijkamp
author_sort Zixiang Wei
collection DOAJ
description Background and purpose: Deep learning (DL)-based auto-segmentation of head and neck cancer (HNC) gross tumour volumes remains challenging due to anatomical complexity and limited accuracy. We propose an interactive DL (iDL) workflow that integrates clinician input to enhance segmentation performance and clinical usability. The iDL approach was evaluated using simulations on two datasets and an observer study. Materials and methods: Two iDL approaches were developed and integrated into a workflow: (1) for primary tumour (GTVt) segmentation, clinicians marked the tumour centre and delineated three orthogonal slices to enable patient-specific fine-tuning of a pre-trained 3D UNet; (2) for lymph nodes (GTVn), clinician-provided clicks identified involved lymph nodes used as attention maps in a separate UNet. Methods were evaluated using independent simulations on an internal dataset (n = 204) and the HECKTOR 2022 dataset (n = 524), using aggregated dice-similarity-coefficient (DSCagg). An additional observer study with three radiation oncologists assessed usability and efficiency, using normalized added path length (APL) and the System Usability Scale (SUS). Results: The iDL workflow achieved high segmentation accuracy, with a DSCagg of 0.84 (internal) and 0.88 (HECKTOR) for GTVt, and 0.83 and 0.85 for GTVn. GTVn required minimal correction (mean APL: 4 % vs. 6 % vs. 11 %); two observers made limited corrections to GTVt (mean APL: 11 % vs. 6 % vs. 39 %). Mean segmentation time was 12 min per case. SUS scores ranged from 87.5 to 100, indicating high usability. Conclusion: The iDL workflow achieved high accuracy and usability with limited correction time, offering a practical and efficient solution for HNC segmentation in radiotherapy.
format Article
id doaj-art-cc8af8984e40468c91911964762f9a5a
institution Kabale University
issn 2405-6316
language English
publishDate 2025-07-01
publisher Elsevier
record_format Article
series Physics and Imaging in Radiation Oncology
spelling doaj-art-cc8af8984e40468c91911964762f9a5a2025-08-20T03:40:31ZengElsevierPhysics and Imaging in Radiation Oncology2405-63162025-07-013510082010.1016/j.phro.2025.100820An interactive deep-learning workflow for head and neck gross tumour volume segmentationZixiang Wei0Jintao Ren1Jesper Grau Eriksen2Kenneth Jensen3Hanna Rahbek Mortensen4Stine Sofia Korreman5Jasper Nijkamp6Department of Clinical Medicine, Aarhus University, Aarhus, Denmark; Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, DenmarkDepartment of Clinical Medicine, Aarhus University, Aarhus, Denmark; Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, DenmarkDepartment of Oncology, Aarhus University Hospital, Aarhus, Denmark; Department of Experimental Oncology, Aarhus University Hospital, DenmarkDanish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, DenmarkDepartment of Clinical Medicine, Aarhus University, Aarhus, Denmark; Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, DenmarkDepartment of Clinical Medicine, Aarhus University, Aarhus, Denmark; Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, DenmarkDepartment of Clinical Medicine, Aarhus University, Aarhus, Denmark; Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, Denmark; Corresponding author at: Palle Juul-Jensens Blvd. 25, 8200 Aarhus, Denmark.Background and purpose: Deep learning (DL)-based auto-segmentation of head and neck cancer (HNC) gross tumour volumes remains challenging due to anatomical complexity and limited accuracy. We propose an interactive DL (iDL) workflow that integrates clinician input to enhance segmentation performance and clinical usability. The iDL approach was evaluated using simulations on two datasets and an observer study. Materials and methods: Two iDL approaches were developed and integrated into a workflow: (1) for primary tumour (GTVt) segmentation, clinicians marked the tumour centre and delineated three orthogonal slices to enable patient-specific fine-tuning of a pre-trained 3D UNet; (2) for lymph nodes (GTVn), clinician-provided clicks identified involved lymph nodes used as attention maps in a separate UNet. Methods were evaluated using independent simulations on an internal dataset (n = 204) and the HECKTOR 2022 dataset (n = 524), using aggregated dice-similarity-coefficient (DSCagg). An additional observer study with three radiation oncologists assessed usability and efficiency, using normalized added path length (APL) and the System Usability Scale (SUS). Results: The iDL workflow achieved high segmentation accuracy, with a DSCagg of 0.84 (internal) and 0.88 (HECKTOR) for GTVt, and 0.83 and 0.85 for GTVn. GTVn required minimal correction (mean APL: 4 % vs. 6 % vs. 11 %); two observers made limited corrections to GTVt (mean APL: 11 % vs. 6 % vs. 39 %). Mean segmentation time was 12 min per case. SUS scores ranged from 87.5 to 100, indicating high usability. Conclusion: The iDL workflow achieved high accuracy and usability with limited correction time, offering a practical and efficient solution for HNC segmentation in radiotherapy.http://www.sciencedirect.com/science/article/pii/S2405631625001253Interactive deep-learningHead and neck cancerObserver study
spellingShingle Zixiang Wei
Jintao Ren
Jesper Grau Eriksen
Kenneth Jensen
Hanna Rahbek Mortensen
Stine Sofia Korreman
Jasper Nijkamp
An interactive deep-learning workflow for head and neck gross tumour volume segmentation
Physics and Imaging in Radiation Oncology
Interactive deep-learning
Head and neck cancer
Observer study
title An interactive deep-learning workflow for head and neck gross tumour volume segmentation
title_full An interactive deep-learning workflow for head and neck gross tumour volume segmentation
title_fullStr An interactive deep-learning workflow for head and neck gross tumour volume segmentation
title_full_unstemmed An interactive deep-learning workflow for head and neck gross tumour volume segmentation
title_short An interactive deep-learning workflow for head and neck gross tumour volume segmentation
title_sort interactive deep learning workflow for head and neck gross tumour volume segmentation
topic Interactive deep-learning
Head and neck cancer
Observer study
url http://www.sciencedirect.com/science/article/pii/S2405631625001253
work_keys_str_mv AT zixiangwei aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jintaoren aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jespergraueriksen aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT kennethjensen aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT hannarahbekmortensen aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT stinesofiakorreman aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jaspernijkamp aninteractivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT zixiangwei interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jintaoren interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jespergraueriksen interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT kennethjensen interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT hannarahbekmortensen interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT stinesofiakorreman interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation
AT jaspernijkamp interactivedeeplearningworkflowforheadandneckgrosstumourvolumesegmentation