Automated detection of dental artifacts for large-scale radiomic analysis in radiation oncology

Abstract

Background and purpose: Computed tomography (CT) is one of the most common medical imaging modalities in radiation oncology and radiomics research, the computational voxel-level analysis of medical images. Radiomics is vulnerable to the effects of dental artifacts (DA) caused by metal implants or fillings and can hamper future reproducibility on new datasets. In this study we seek to better understand the robustness of quantitative radiomic features to DAs. Furthermore, we propose a novel method of detecting DAs in order to safeguard radiomic studies and improve reproducibility. Materials and methods: We analyzed the correlations between radiomic features and the location of dental artifacts in a new dataset containing 3D CT scans from 3211 patients. We then combined conventional image processing techniques with a pre-trained convolutional neural network to create a three-class patient-level DA classifier and slice-level DA locator. Finally, we demonstrated its utility in reducing the correlations between the location of DAs and certain radiomic features. Results: We found that when strong DAs were present, the proximity of the tumour to the mouth was highly correlated with 36 radiomic features. We predicted the correct DA magnitude yielding a Matthews correlation coefficient of 0.73 and location of DAs achieving the same level of agreement as human labellers. Conclusions: Removing radiomic features or CT slices containing DAs could reduce the unwanted correlations between the location of DAs and radiomic features. Automated DA detection can be used to improve the reproducibility of radiomic studies; an important step towards creating effective radiomic models for use in clinical radiation oncology.

SCARF: Auto-Segmentation Clinical Acceptability & Reproducibility Framework for Benchmarking Essential Radiation Therapy Targets in Head and Neck Cancer

“…TCIA-HNSCC (TCHN) had the least overlapping categories (5 out of 19 OARs). Results for each external dataset can be found in Appendix B (Supplementary Figure 8, Supplementary Tables 4 and 5) [5,20,43–52].…”

“…TCIA-HNSCC (TCHN) had the least overlapping categories (5 out of 19 OARs). Results for each external dataset can be found in Appendix B (Supplementary Figure 8, Supplementary Tables 4 and 5) [5,20,[43][44][45][46][47][48][49][50][51][52]. By providing open access to all data and methods used in the analysis, SCARF addresses some of the key challenges faced in the field of radiation therapy and AI integration.…”

Identification of radiomic biomarkers in a set of four skeletal muscle groups on Dixon MRI of the NAKO MR study

“…Recently, radiomics have been applied to extract and leverage quantitative CT [ 17 , 43 ], or MRI [ 44 ] texture features to analyze imaging data at large scales. It has also been shown to be extensible to multi-parametric studies [ 45 ] or be applicable for the classification of dental artifacts [ 46 ]. Motivated by these achievements, we proposed a feature extraction and selection pipeline from raw data to radiomics features of muscle tissue depicted in a two-point Dixon MRI sequence.…”

Assessing the generalisability of radiomics features previously identified as predictive of radiation-induced sticky saliva and xerostomia

“…To avoid analysing CT intensity values that do not correspond to tissue densities [37] , [38] , van Dijk et al excluded 33 % of their patients as they presented with metal artefacts on their CT scans. Implementing the same approach in our cohort, would have resulted in the exclusion of 95 % (104/109) of the patients and was therefore not undertaken.…”