Objective: Conversion of multidimensional data obtained with Computed Tomography (CT) in the field of medical imaging with radiomics into digitized data via mathematical methods are shown to contribute to the diagnosis process, determination of prognosis and treatment processes. However, there are a limited number of studies on using radiomics in Cone Beam Computed Tomography (CBCT) utilized in maxillofacial radiology. The differences in the working principles and the factors affecting image acquisition for CT and CBCT modalities may pose drawbacks for radiomics studies for to be conducted CBCT. Accordingly, the aim of this study is to investigate whether radiomics is a reliable measurement tool in CBCT by repeatability measurement of radiomics parameters and to investigate the specific radiomics features to be used in CBCT studies.

Method: In this study, radiomic analysis was performed using the 3D Slicer program on the data of 33 cases which were taken at different times and with the same CBCT device retrieved from the archive of Kocaeli University Faculty of Dentistry, Department of Oral and Maxillofacial Radiology. A total of 107 radiomics parameters were investigated with on segmented C2 (Axis) data in all cases. The results of the radiomic analysis were evaluated using the Intraclass Correlation Coefficient (ICC) in the SPSS program to assess repeatability.

Results: As a result of the radiomics analysis, good repeatability was observed on 11 out of 14 features associated with morphological features; 16 out of 18 features associated with first-order features; 14 out of 24 features associated with Gray Level Cooccurrence Matrix; 5 out of 14 features associated with Gray Level Dependence Matrix; 5 out of 16 features associated with the Gray Level Run Length Matrix; 10 out of 16 features associated with the Gray Level Size Zone Matrix and 3 out of 5 features associated with the Neighbouring Gray Tone Difference Matrix (ICC>0. 80).

Conclusion: This study presents a alternative parameters for radiomics analysis of CBCT images. To our knowledge, it is the first study to test the repeatability of radiomic features of CBCT in vivo. The poor repeatability of grayscale parameters should be taken into account when performing radiomics analysis of CBCT images.

Keywords: Radiomics; Cone beam computed tomography; Feature selection; Texture analysis