Journal article
4DCT image artifact detection using deep learning
Medical physics (Lancaster), Vol.52(2), pp.1096-1107
02/2025
DOI: 10.1002/mp.17513
PMCID: PMC11788241
PMID: 39540716
Appears in UI Libraries Support Open Access
Abstract
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.BACKGROUNDFour-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.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.PURPOSEWe 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.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.METHODSWe 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.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.RESULTSThe 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.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.CONCLUSIONSThe 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.
Details
- Title: Subtitle
- 4DCT image artifact detection using deep learning
- Creators
- Joshua W Carrizales - University of IowaMattison J Flakus - University of Wisconsin–MadisonDallin Fairbourn - Utah State UniversityWei Shao - University of FloridaSarah E Gerard - University of IowaJohn E Bayouth - Oregon Health & Science UniversityGary E Christensen - University of IowaJoseph M Reinhardt - University of Iowa
- Resource Type
- Journal article
- Publication Details
- Medical physics (Lancaster), Vol.52(2), pp.1096-1107
- DOI
- 10.1002/mp.17513
- PMID
- 39540716
- PMCID
- PMC11788241
- NLM abbreviation
- Med Phys
- ISSN
- 2473-4209
- eISSN
- 2473-4209
- Publisher
- Wiley
- Grant note
- National Institutes of Health
This work was supported in part by the National Institutes of Health grants T32 HL144461 and R01 CA166703.
- Language
- English
- Electronic publication date
- 11/14/2024
- Date published
- 02/2025
- Academic Unit
- Roy J. Carver Department of Biomedical Engineering; Radiology; Electrical and Computer Engineering; Radiation Oncology; Radiation Research Laboratory; The Iowa Institute for Biomedical Imaging; Advanced Pulmonary Physiomic Imaging Laboratory; Holden Comprehensive Cancer Center
- Record Identifier
- 9984747819602771
Metrics
61 Record Views