eduzhai > Applied Sciences > Engineering >

Benchmarking features from different radiomics toolkits / toolboxes using Image Biomarkers Standardization Initiative

  • Save

... pages left unread,continue reading

Document pages: 21 pages

Abstract: There is no consensus regarding the radiomic feature terminology, theunderlying mathematics, or their implementation. This creates a scenario wherefeatures extracted using different toolboxes could not be used to build orvalidate the same model leading to a non-generalization of radiomic results. Inthis study, the image biomarker standardization initiative (IBSI) establishedphantom and benchmark values were used to compare the variation of the radiomicfeatures while using 6 publicly available software programs and 1 in-houseradiomics pipeline. All IBSI-standardized features (11 classes, 173 in total)were extracted. The relative differences between the extracted feature valuesfrom the different software and the IBSI benchmark values were calculated tomeasure the inter-software agreement. To better understand the variations,features are further grouped into 3 categories according to their properties:1) morphology, 2) statistic histogram and 3)texture features. While a goodagreement was observed for a majority of radiomics features across the variousprograms, relatively poor agreement was observed for morphology features.Significant differences were also found in programs that use different graylevel discretization approaches. Since these programs do not include all IBSIfeatures, the level of quantitative assessment for each category was analyzedusing Venn and the UpSet diagrams and also quantified using two ad hoc metrics.Morphology features earns lowest scores for both metrics, indicating thatmorphological features are not consistently evaluated among software programs.We conclude that radiomic features calculated using different software programsmay not be identical and reliable. Further studies are needed to standardizethe workflow of radiomic feature extraction.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...