Multi-institutional PET/CT image segmentation using federated deep transformer learning.

Humans Algorithms Deep Learning Image Processing, Computer-Assisted / methods Neoplasms / diagnostic imaging Positron Emission Tomography Computed Tomography / methods
Deep transformers Federated learning PET/CT Privacy Segmentation

Journal

Computer methods and programs in biomedicine
ISSN: 1872-7565
Titre abrégé: Comput Methods Programs Biomed
Pays: Ireland
ID NLM: 8506513

Informations de publication

Date de publication:
Oct 2023
Historique:
received: 09 01 2023
accepted: 02 07 2023
medline: 29 8 2023
pubmed: 29 7 2023
entrez: 28 7 2023
Statut: ppublish

Résumé

Generalizable and trustworthy deep learning models for PET/CT image segmentation necessitates large diverse multi-institutional datasets. However, legal, ethical, and patient privacy issues challenge sharing of datasets between different centers. To overcome these challenges, we developed a federated learning (FL) framework for multi-institutional PET/CT image segmentation. A dataset consisting of 328 FL (HN) cancer patients who underwent clinical PET/CT examinations gathered from six different centers was enrolled. A pure transformer network was implemented as fully core segmentation algorithms using dual channel PET/CT images. We evaluated different frameworks (single center-based, centralized baseline, as well as seven different FL algorithms) using 68 PET/CT images (20% of each center data). In particular, the implemented FL algorithms include clipping with the quantile estimator (ClQu), zeroing with the quantile estimator (ZeQu), federated averaging (FedAvg), lossy compression (LoCo), robust aggregation (RoAg), secure aggregation (SeAg), and Gaussian differentially private FedAvg with adaptive quantile clipping (GDP-AQuCl). The Dice coefficient was 0.80±0.11 for both centralized and SeAg FL algorithms. All FL approaches achieved centralized learning model performance with no statistically significant differences. Among the FL algorithms, SeAg and GDP-AQuCl performed better than the other techniques. However, there was no statistically significant difference. All algorithms, except the center-based approach, resulted in relative errors less than 5% for SUV The developed FL-based (with centralized method performance) algorithms exhibited promising performance for HN tumor segmentation from PET/CT images.

Sections du résumé

BACKGROUND AND OBJECTIVE OBJECTIVE
Generalizable and trustworthy deep learning models for PET/CT image segmentation necessitates large diverse multi-institutional datasets. However, legal, ethical, and patient privacy issues challenge sharing of datasets between different centers. To overcome these challenges, we developed a federated learning (FL) framework for multi-institutional PET/CT image segmentation.
METHODS METHODS
A dataset consisting of 328 FL (HN) cancer patients who underwent clinical PET/CT examinations gathered from six different centers was enrolled. A pure transformer network was implemented as fully core segmentation algorithms using dual channel PET/CT images. We evaluated different frameworks (single center-based, centralized baseline, as well as seven different FL algorithms) using 68 PET/CT images (20% of each center data). In particular, the implemented FL algorithms include clipping with the quantile estimator (ClQu), zeroing with the quantile estimator (ZeQu), federated averaging (FedAvg), lossy compression (LoCo), robust aggregation (RoAg), secure aggregation (SeAg), and Gaussian differentially private FedAvg with adaptive quantile clipping (GDP-AQuCl).
RESULTS RESULTS
The Dice coefficient was 0.80±0.11 for both centralized and SeAg FL algorithms. All FL approaches achieved centralized learning model performance with no statistically significant differences. Among the FL algorithms, SeAg and GDP-AQuCl performed better than the other techniques. However, there was no statistically significant difference. All algorithms, except the center-based approach, resulted in relative errors less than 5% for SUV
CONCLUSIONS CONCLUSIONS
The developed FL-based (with centralized method performance) algorithms exhibited promising performance for HN tumor segmentation from PET/CT images.

Identifiants

DOI: 10.1016/j.cmpb.2023.107706 PMID: 37506602
pubmed: 37506602
pii: S0169-2607(23)00371-1
doi: 10.1016/j.cmpb.2023.107706
pii:
doi:

Types de publication

Journal Article Multicenter Study

Langues

eng

Sous-ensembles de citation

IM

Pagination

107706

Informations de copyright

Copyright © 2023 The Author(s). Published by Elsevier B.V. All rights reserved.

Déclaration de conflit d'intérêts

Declaration of Competing Interest None

Auteurs

Isaac Shiri (I)

Division of Nuclear Medicine and Molecular Imaging, Geneva University Hospital, Geneva, Switzerland.

Behrooz Razeghi (B)

Department of Computer Science, University of Geneva, Geneva, Switzerland.

Alireza Vafaei Sadr (A)

Institute of Pathology, RWTH Aachen University Hospital, Aachen, Germany; Department of Public Health Sciences, College of Medicine, The Pennsylvania State University, Hershey, PA 17033, USA.

Mehdi Amini (M)

Division of Nuclear Medicine and Molecular Imaging, Geneva University Hospital, Geneva, Switzerland.

Yazdan Salimi (Y)

Division of Nuclear Medicine and Molecular Imaging, Geneva University Hospital, Geneva, Switzerland.

Sohrab Ferdowsi (S)

Department of Computer Science, University of Geneva, Geneva, Switzerland.

Peter Boor (P)

Institute of Pathology, RWTH Aachen University Hospital, Aachen, Germany.

Deniz Gündüz (D)

Department of Electrical and Electronic Engineering, Imperial College London, UK.

Slava Voloshynovskiy (S)

Department of Computer Science, University of Geneva, Geneva, Switzerland.

Habib Zaidi (H)

Division of Nuclear Medicine and Molecular Imaging, Geneva University Hospital, Geneva, Switzerland; Geneva University Neurocenter, University of Geneva, Geneva, Switzerland; Department of Nuclear Medicine and Molecular Imaging, University of Groningen, Groningen, The Netherlands; Department of Nuclear Medicine, University of Southern Denmark, Odense, Denmark. Electronic address: habib.zaidi@hcuge.ch.

Articles similaires

[Redispensing of expensive oral anticancer medicines: a practical application].

Lisanne N van Merendonk, Kübra Akgöl, Bastiaan Nuijen
1.00
Humans Antineoplastic Agents Administration, Oral Drug Costs Counterfeit Drugs

Smoking Cessation and Incident Cardiovascular Disease.

Jun Hwan Cho, Seung Yong Shin, Hoseob Kim et al.
1.00
Humans Male Smoking Cessation Cardiovascular Diseases Female

Evaluation of Low-Value Services Across Major Medicare Advantage Insurers and Traditional Medicare.

Ciara Duggan, Adam L Beckman, Ishani Ganguli et al.
1.00
Humans United States Aged Cross-Sectional Studies Medicare Part C

Effectiveness of Virtual Yoga for Chronic Low Back Pain: A Randomized Clinical Trial.

Hallie Tankha, Devyn Gaskins, Amanda Shallcross et al.
1.00
Humans Yoga Low Back Pain Female Male

Classifications MeSH

questionsmedicales.fr › Eucaryotes › Animaux › Chordés › Vertébrés › Mammifères › Eutheria › Primates › Haplorhini › Catarrhini › Hominidae › Humains › Multi-institutional PET/CT image segmentation...
questionsmedicales.fr › Sciences de l'information › Méthodologies informatiques › Algorithmes › Multi-institutional PET/CT image segmentation...
questionsmedicales.fr › Sciences de l'information › Méthodologies informatiques › Algorithmes › Intelligence artificielle › Réseaux neuronaux (ordinateur) › Apprentissage profond › Multi-institutional PET/CT image segmentation...
questionsmedicales.fr › Sciences de l'information › Méthodologies informatiques › Traitement d'image par ordinateur › Multi-institutional PET/CT image segmentation...
questionsmedicales.fr › Tumeurs › Multi-institutional PET/CT image segmentation...
questionsmedicales.fr › Diagnostic › Techniques et procédures diagnostiques › Techniques de diagnostic radioisotopique › Scintigraphie › Tomoscintigraphie › Tomographie par émission de positons › Tomographie par émission de positons couplée à la tomodensitométrie › Multi-institutional PET/CT image segmentation...