4DCT image artifact detection using deep learning
Medical Physics · 2025-02 · 52 (2) · 1096-1107
DOI: 10.1002/mp.17513 ↗At a glance
Journal article in Medical Physics, published online on 2024-11-14. Written by Joshua W Carrizales and 7 co-authors. Institution: Biomedical Engineering, University of Iowa, Iowa City, Iowa, USA. Major MEDLINE topics: Deep Learning, Artifacts, Four-Dimensional Computed Tomography, Image Processing, Computer-Assisted. Cited 7 times so far (Crossref); 73 references. The full text is freely available (open access, CC BY 4.0). Funded by: NCI NIH HHS, NHLBI NIH HHS, NIH HHS.
Abstract (original language)
Background
Four‐dimensional computed tomography (4DCT) is an es sential tool in radiation therapy. However, the 4D acquisition process may cause motion artifacts which can obscure anatomy and distort functional measurements from CT scans.
Purpose
We describe a deep learning algorithm to identify the location of artifacts within 4DCT images. Our method is flexible enough to handle different types of artifacts, including duplication, misalignment, truncation, and interpolation.
Methods
We trained and validated a U‐net convolutional neural network artifact detection model on more than 23 000 coronal slices extracted from 98 4DCT scans. The receiver operating characteristic (ROC) curve and precision‐recall curve were used to evaluate the model's performance at identifying artifacts compared to a manually identified ground truth. The model was adjusted so that the sensitivity in identifying artifacts was equivalent to that of a human observer, as measured by computing the average ratio of artifact volume to lung volume in a given scan.
Results
The model achieved a sensitivity, specificity, and precision of 0.78, 0.99, and 0.58, respectively. The ROC area‐under‐the‐curve (AUC) was 0.99 and the precision‐recall AUC was 0.73. Our model sensitivity is 8% higher than previously reported state‐of‐the‐art artifact detection methods.
Conclusions
The model developed in this study is versatile, designed to handle duplication, misalignment, truncation, and interpolation artifacts within a single image, unlike earlier models that were designed for a single artifact type.
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- Biomedical Engineering, University of Iowa, Iowa City, Iowa, USA.
- Medical Physics, University of Wisconsin - Madison, Madison, Wisconsin, USA.
- Biological Engineering, Utah State University, Logan, Utah, USA.
- Medicine, University of Florida, Gainesville, Florida, USA.
- Radiation Medicine, Oregon Health and Science University, Portland, Oregon, USA.
- Electrical and Computer Engineering, University of Iowa, Iowa City, Iowa, USA.
Funding
- NCI NIH HHS
- NHLBI NIH HHS
- NIH HHS
- National Institutes of Health
Cite
APA
Carrizales, J. W., Flakus, M. J., Fairbourn, D., Shao, W., Gerard, S. E., Bayouth, J. E., Christensen, G. E., & Reinhardt, J. M. (2025). 4DCT image artifact detection using deep learning. Medical Physics, 52(2), 1096-1107. https://doi.org/10.1002/mp.17513
Vancouver
Carrizales JW, Flakus MJ, Fairbourn D, Shao W, Gerard SE, Bayouth JE, et al. 4DCT image artifact detection using deep learning. Medical Physics. 2025;52(2):1096-1107. doi:10.1002/mp.17513
BibTeX
@article{Carrizales2025,
author = {Carrizales, Joshua W and Flakus, Mattison J and Fairbourn, Dallin and Shao, Wei and Gerard, Sarah E and Bayouth, John E and Christensen, Gary E and Reinhardt, Joseph M},
title = {4DCT image artifact detection using deep learning},
journal = {Medical Physics},
year = {2025},
volume = {52},
number = {2},
pages = {1096--1107},
doi = {10.1002/mp.17513}
}
EUVIMED