2017
DOI: 10.1371/journal.pone.0178524
|View full text |Cite|
|
Sign up to set email alerts
|

Harmonizing the pixel size in retrospective computed tomography radiomics studies

Abstract: Consistent pixel sizes are of fundamental importance for assessing texture features that relate intensity and spatial information in radiomics studies. To correct for the effects of variable pixel sizes, we combined image resampling with Butterworth filtering in the frequency domain and tested the correction on computed tomography (CT) scans of lung cancer patients reconstructed 5 times with pixel sizes varying from 0.59 to 0.98 mm. One hundred fifty radiomics features were calculated for each preprocessing an… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

3
139
0

Year Published

2018
2018
2024
2024

Publication Types

Select...
7

Relationship

0
7

Authors

Journals

citations
Cited by 134 publications
(142 citation statements)
references
References 27 publications
3
139
0
Order By: Relevance
“…Down sampling (1.0 × 1.0 × 1.0 mm 3 ) requires inference and can introduce artificial information to the data, in contrast, up sampling to the largest dimension (3.0 × 3.0 × 3.0 mm 3 ) can incur information loss. Up sampling may also introduce image aliasing artifacts as well that requires anti‐aliasing filters prior to filtering . Multiple‐scaling strategies potentially offer a good trade‐off .…”
Section: Discussionmentioning
confidence: 99%
“…Down sampling (1.0 × 1.0 × 1.0 mm 3 ) requires inference and can introduce artificial information to the data, in contrast, up sampling to the largest dimension (3.0 × 3.0 × 3.0 mm 3 ) can incur information loss. Up sampling may also introduce image aliasing artifacts as well that requires anti‐aliasing filters prior to filtering . Multiple‐scaling strategies potentially offer a good trade‐off .…”
Section: Discussionmentioning
confidence: 99%
“…As an example, while numerous studies utilized a fixed number of bins for textural analysis, recent studies suggested that a fixed bin size with variable number of bins per lesion provides better comparability and reproducibility of textural features [10,55,184]. Furthermore, image resolution normalization [228], or normalization of already extracted radiomic features in the feature space [229] have been proposed. In addition, guidelines are available focusing on imaging, feature extraction, analysis and cross-validation standardization of radiomic studies [230][231][232].…”
Section: Feature Engineeringmentioning
confidence: 99%
“…found increased feature variability following voxel resampling on computed tomography images to produce consistent pixel size, compared to no preprocessing . Studies have used methods such as image resizing followed by Butterworth filtering, low‐pass filters, band‐pass filters, or Gaussian filters in order to harmonize images prior to feature calculation . In this study, we chose not to apply these steps to force consistent pixel size as the presented method seeks to address image heterogeneities through feature selection as opposed to image processing.…”
Section: Discussionmentioning
confidence: 99%
“…One challenge faced in developing widely generalizable imaging phenotypes is the sensitivity of individual texture features to imaging conditions. Imaging conditions such as manufacturer, kVp, and processing algorithms may each affect radiomic feature values . Studies have been performed to evaluate repeatability (test–retest) and reproducibility of radiomic features in cancer imaging.…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation