Skip to main content

Main menu

  • Home
  • Content
    • Current
    • Ahead of print
    • Past Issues
    • JNM Supplement
    • SNMMI Annual Meeting Abstracts
    • Continuing Education
    • JNM Podcasts
  • Subscriptions
    • Subscribers
    • Institutional and Non-member
    • Rates
    • Journal Claims
    • Corporate & Special Sales
  • Authors
    • Submit to JNM
    • Information for Authors
    • Assignment of Copyright
    • AQARA requirements
  • Info
    • Reviewers
    • Permissions
    • Advertisers
  • About
    • About Us
    • Editorial Board
    • Contact Information
  • More
    • Alerts
    • Feedback
    • Help
    • SNMMI Journals
  • SNMMI
    • JNM
    • JNMT
    • SNMMI Journals
    • SNMMI

User menu

  • Subscribe
  • My alerts
  • Log in
  • My Cart

Search

  • Advanced search
Journal of Nuclear Medicine
  • SNMMI
    • JNM
    • JNMT
    • SNMMI Journals
    • SNMMI
  • Subscribe
  • My alerts
  • Log in
  • My Cart
Journal of Nuclear Medicine

Advanced Search

  • Home
  • Content
    • Current
    • Ahead of print
    • Past Issues
    • JNM Supplement
    • SNMMI Annual Meeting Abstracts
    • Continuing Education
    • JNM Podcasts
  • Subscriptions
    • Subscribers
    • Institutional and Non-member
    • Rates
    • Journal Claims
    • Corporate & Special Sales
  • Authors
    • Submit to JNM
    • Information for Authors
    • Assignment of Copyright
    • AQARA requirements
  • Info
    • Reviewers
    • Permissions
    • Advertisers
  • About
    • About Us
    • Editorial Board
    • Contact Information
  • More
    • Alerts
    • Feedback
    • Help
    • SNMMI Journals
  • View or Listen to JNM Podcast
  • Visit JNM on Facebook
  • Join JNM on LinkedIn
  • Follow JNM on Twitter
  • Subscribe to our RSS feeds
Meeting ReportPhysics, Instrumentation & Data Sciences - Data Sciences

Impact of training dataset size on technical performance of a deep learning model for detection and quantification of lymphomatous disease on 18F-FDG PET/CT

Georgia Ionescu, Russell FROOD, Andrew SCARSBROOK and Julien Willaime
Journal of Nuclear Medicine June 2023, 64 (supplement 1) P1069;
Georgia Ionescu
1Mirada Medical Ltd.
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Russell FROOD
2LEEDS TEACHING HOSPITALS NHS TRUST
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Andrew SCARSBROOK
2LEEDS TEACHING HOSPITALS NHS TRUST
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
Julien Willaime
1Mirada Medical Ltd.
  • Find this author on Google Scholar
  • Find this author on PubMed
  • Search for this author on this site
  • Article
  • Figures & Data
  • Info & Metrics
Loading

Abstract

P1069

Introduction: FDG PET/CT is widely used for staging high-grade lymphoma. Time to evaluate studies vary depending on case complexity. Integrating artificial intelligence (AI) within the reporting workflow has the potential to improve efficiency and enable use of advanced quantification methods in a clinical setting. This study evaluated the impact of amount of data used to train a convolutional neural network (CNN) based deep learning (DL) model on detection and segmentation performance metrics.

Methods: A total of 6150 lymphoma lesions considered as ground truth (GT), were segmented on pre-treatment FDG PET/CT scans of 420 patients with high-grade lymphoma by a radiologist with 7 years’ experience. GT segmentation included nodal and extra-nodal disease. All segmentations were checked by a dual-certified radiologist/nuclear medicine physician with >15 years of experience. A DL model, consisting of an ensemble of patch-based 3D DenseNet, was trained using various dataset sizes, N = 50, 100, 150, 200 and 300, randomly sampled from a total of 300 cases. The same architecture, training strategies and loss function were used for each of the 5 training sets. Technical performance was assessed on a separate evaluation dataset of 120 cases. Per-patient lesion detection performance was assessed by computing true positive rate (TPR) and number of false positive (FP) findings. Voxel-wise detection sensitivity and positive predictive value (PPV) were also calculated. Segmentation and quantification performance were evaluated using DICE, non-parametric Bland Altman, and intraclass correlation coefficient (ICC) for SUVmax and SUVmean per lesion, and total metabolic volume (TMV) and total lesion glycolysis (TLG) per patient. Statistics reported for Bland-Altman were median difference (bias) and lower and upper limits of Agreement (LoA) calculated as 2.5th and 97.5th percentile.

Results: All models demonstrated good lesion detection capability (median TPR: 83-88%) whilst the number of FPs (median) decreased from 9 (N=50) to 3 (N=300). Similarly, per-voxel-analysis demonstrated consistent sensitivity across the 5 models (91-93%) whilst PPV increased (median: 75%, 82%, 83%, 86% and 88%, for N=50-300). Agreement between predicted and GT contours, measured using DICE score improved with larger datasets (median: 0.78, 0.83, 0.84, 0.85, 0.86, respectively). Bland Altman analysis showed significantly better agreement between predicted and GT SUVmax values for N=300 (bias = 0, LoA = [-0.03, 0.0]) versus N<=200 (bias = 0, LoA varying between [-0.19, 0] and [-0.12, 0.1]). However, for N>50, for 95% of cases predicted and GT SUVmax were in perfect agreement. LoA for SUVmean were consistent across the 5 models (between [-1.1, 1.1] and [-1.5, 1.5]) with bias between -0.11 and 0. TMV agreement consistently improved with increasing training dataset size (LoA = [-499, 461] for N=50 gradually decreasing down to LoA = [-345, 281] for N=300), whilst bias decreased from 26 (N=50) to 6 (N=300). TLG showed a similar trend to TMV. ICC for SUVmax significantly increased between N=50 (ICC=0.72) and N > 100 (ICC=0.97-0.99). ICC for other parameters was consistent across all models: 0.93-0.95 for SUVmean, 0.91-0.94 for TMV and 0.97-0.99 for TLG. Visual assessment confirmed that accuracy of lesion segmentation improved with larger training data size.

Conclusions: A deep learning model was relatively unaffected by size of training dataset in its ability to detect lymphoma lesions on PET/CT scans. However, more training data reduced FP rate, and improved agreement between prediction and ground truth segmentations for SUVmax, TMV and TLG.

Figure
  • Download figure
  • Open in new tab
  • Download powerpoint
Figure
  • Download figure
  • Open in new tab
  • Download powerpoint
Previous
Back to top

In this issue

Journal of Nuclear Medicine
Vol. 64, Issue supplement 1
June 1, 2023
  • Table of Contents
  • Index by author
Article Alerts
Sign In to Email Alerts with your Email Address
Email Article

Thank you for your interest in spreading the word on Journal of Nuclear Medicine.

NOTE: We only request your email address so that the person you are recommending the page to knows that you wanted them to see it, and that it is not junk mail. We do not capture any email address.

Enter multiple addresses on separate lines or separate them with commas.
Impact of training dataset size on technical performance of a deep learning model for detection and quantification of lymphomatous disease on 18F-FDG PET/CT
(Your Name) has sent you a message from Journal of Nuclear Medicine
(Your Name) thought you would like to see the Journal of Nuclear Medicine web site.
Citation Tools
Impact of training dataset size on technical performance of a deep learning model for detection and quantification of lymphomatous disease on 18F-FDG PET/CT
Georgia Ionescu, Russell FROOD, Andrew SCARSBROOK, Julien Willaime
Journal of Nuclear Medicine Jun 2023, 64 (supplement 1) P1069;

Citation Manager Formats

  • BibTeX
  • Bookends
  • EasyBib
  • EndNote (tagged)
  • EndNote 8 (xml)
  • Medlars
  • Mendeley
  • Papers
  • RefWorks Tagged
  • Ref Manager
  • RIS
  • Zotero
Share
Impact of training dataset size on technical performance of a deep learning model for detection and quantification of lymphomatous disease on 18F-FDG PET/CT
Georgia Ionescu, Russell FROOD, Andrew SCARSBROOK, Julien Willaime
Journal of Nuclear Medicine Jun 2023, 64 (supplement 1) P1069;
Twitter logo Facebook logo LinkedIn logo Mendeley logo
  • Tweet Widget
  • Facebook Like
  • Google Plus One
Bookmark this article

Jump to section

  • Article
  • Figures & Data
  • Info & Metrics

Related Articles

  • No related articles found.
  • Google Scholar

Cited By...

  • No citing articles found.
  • Google Scholar

More in this TOC Section

  • Persistent Homology-Driven Topological Loss for Robust Lung Lobe Segmentation using nnUNet
  • Unsupervised deep learning improves low dose SV2A PET imaging: a correlation study with motor severity in Parkinson’s disease
  • Deep learning-assisted automatic differentiated diagnosis of acute tubular necrosis from acute rejection in transplanted kidney scintigraphy
Show more Physics, Instrumentation & Data Sciences - Data Sciences

Similar Articles

SNMMI

© 2025 SNMMI

Powered by HighWire