Aarhus University Seal

Jan Wolff

Comparison of convolutional neural network training strategies for cone-beam CT image segmentation

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review



  • Jordi Minnema, Vrije Universiteit Amsterdam, Netherlands
  • Jan Wolff
  • Juha Koivisto, University of Helsinki, Finland
  • Felix Lucka, Centrum Wiskunde & Informatica (CWI), Amsterdam, University College London, Netherlands
  • Kees Joost Batenburg, Centrum Wiskunde & Informatica (CWI), Amsterdam, Netherlands
  • Tymour Forouzanfar, Vrije Universiteit Amsterdam, Netherlands
  • Maureen van Eijnatten, Centrum Wiskunde & Informatica (CWI), Amsterdam, Netherlands

BACKGROUND AND OBJECTIVE: Over the past decade, convolutional neural networks (CNNs) have revolutionized the field of medical image segmentation. Prompted by the developments in computational resources and the availability of large datasets, a wide variety of different two-dimensional (2D) and three-dimensional (3D) CNN training strategies have been proposed. However, a systematic comparison of the impact of these strategies on the image segmentation performance is still lacking. Therefore, this study aimed to compare eight different CNN training strategies, namely 2D (axial, sagittal and coronal slices), 2.5D (3 and 5 adjacent slices), majority voting, randomly oriented 2D cross-sections and 3D patches.

METHODS: These eight strategies were used to train a U-Net and an MS-D network for the segmentation of simulated cone-beam computed tomography (CBCT) images comprising randomly-placed non-overlapping cylinders and experimental CBCT images of anthropomorphic phantom heads. The resulting segmentation performances were quantitatively compared by calculating Dice similarity coefficients. In addition, all segmented and gold standard experimental CBCT images were converted into virtual 3D models and compared using orientation-based surface comparisons.

RESULTS: The CNN training strategy that generally resulted in the best performances on both simulated and experimental CBCT images was majority voting. When employing 2D training strategies, the segmentation performance can be optimized by training on image slices that are perpendicular to the predominant orientation of the anatomical structure of interest. Such spatial features should be taken into account when choosing or developing novel CNN training strategies for medical image segmentation.

CONCLUSIONS: The results of this study will help clinicians and engineers to choose the most-suited CNN training strategy for CBCT image segmentation.

Original languageEnglish
Article number106192
JournalComputer Methods and Programs in Biomedicine
Number of pages10
Publication statusPublished - Aug 2021

    Research areas

  • Cone-beam computed tomography, Convolutional neural networks, Deep learning, Medical image segmentation, Training strategies

See relations at Aarhus University Citationformats

Download statistics

No data available

ID: 220441178