Advanced Reliability of Radiomic Features in Ex Vivo HypersightTM-CBCT Imaging by iCBCT-Acuros Reconstruction.

BACKGROUND/AIM: High-quality cone-beam CT (CBCT) imaging, such as HyperSight™-CBCT on the ETHOS® O-ring linear accelerator, has significantly improved adaptive radiotherapy by providing high imaging precision. Advanced reconstruction techniques like the iCBCT-Acuros reconstruction mode have the potential to enhance reliability of quantitative image evaluations, especially regarding interpretation of radiomic features. This study evaluated how the iCBCT-Acuros reconstruction mode improves reliability of radiomic features compared to standard iCBCT in HyperSight™-CBCT imaging.

MATERIALS AND

METHODS

An ex vivo phantom analysis of 16 fruits and vegetables was conducted with HyperSight™-CBCT using iCBCT and iCBCT-Acuros reconstruction modes. Test-retest reliability was assessed at 0° and 90° rotations, along with inter- and intrareader reliability, inter-reconstruction reproducibility, and interphantom discrimination. A total of 110 radiomic features were extracted with LIFEx software, and reliability was analyzed using concordance and intraclass correlation coefficients. Boruta Random Forest analysis evaluated interphantom discrimination.

RESULTS

Mean test-retest reliability (CCC) was higher for iCBCT-Acuros (0.936±0.137) than iCBCT (0.816±0.199, p<0.0001). "Morphological" features showed the highest reliability for both modes. Interreader and intrareader reliability were significantly higher for iCBCT-Acuros (p<0.0001). Reproducibility was moderate overall (mean 0.832±0.224), with "Morphological" features being most reproducible. Interphantom separability was increased with iCBCT-Acuros.

CONCLUSION

The iCBCT-Acuros reconstruction mode of HyperSight™-CBCT significantly improves radiomic feature reliability compared to iCBCT. These advancements likely extend to other quantitative parameters, enhancing their potential as radiological biomarkers. The iCBCT-Acuros reconstruction enables a significantly more robust imaging analysis, facilitating the integration of advanced quantitative imaging techniques into high-precision radiotherapy.