Longitudinal stability of HyperSightTM-CBCT based radiomic features in patients with CT guided adaptive SBRT for prostate cancer

Abstract

CT-guided adaptive radiotherapy (aRT) based on HyperSight^TM-CBCT provides high-quality imaging, allowing quantitative radiomic feature analysis as a monitoring tool. This study comprehensively evaluates the stability of radiomic features, as potential imaging biomarkers, in pelvic structures of prostate cancer patients treated with adaptive stereotactic body radiation therapy (SBRT). Between December 2023 and July 2024, 32 patients with localized prostate cancer underwent adaptive SBRT at the Ethos^® linear accelerator (Varian, Siemens Healthineers) with HyperSight-CBCT imaging. Longitudinal stability was assessed by intraclass correlation coefficient (ICC) over five fractions of aRT for target structures and non-hollow organs at risk. In pooled organs at risk, 93.0% of features showed very high stability (ICC > 0.9) compared to 67.4% in pooled target structures, indicating significantly lower stability for target structures (p = 0.00009129). Second-order features demonstrated greater stability than conventional and shape-based features (p = 0.0433, p = 0.0252). Fraction number significantly affected longitudinal prostate feature variability (p = 0.0135). This study comprehensively analyzed HyperSight-CBCT imaging to evaluate longitudinal stability of radiomic features during adaptive SBRT for prostate cancer. The trends observed will provide a framework for future CT-guided aRT studies, facilitating quantitative imaging analysis of radiological biomarkers for clinical translation and improving personalized treatment.

Keywords: Adaptive radiotherapy; Ethos®; HyperSight-CBCT; Prostate cancer; Radiomics.

Fig. 1

Workflow of adaptive radiotherapy in patients with prostate cancer undergoing ultra-hypofractionated stereotactic body radiotherapy. Target delineation is based on pelvic CT (computed tomography) and MRI (magnetic resonance imaging) scans which are used for treatment planning. Adaptive radiotherapy (aRT) is performed based on fractional HyperSight-CBCT (conebeam computed-tomography) scans, with online adaptive segmentation of normal tissue organs and target volumes. Following fractional aRT treatment, HyperSight-CBCT image processing is conducted, and radiomic features (e.g. of shape, intensity/histogram and texture analysis) are extracted using the open-source software LIFEx (version 5.10, https://www.lifexsoft.org). Subsequently, pre-therapeutic robustness, based on multiple segmentations and clinical longitudinal stability analyses across five fractions of adaptive radiotherapy are performed.

Fig. 2

Heatmap displaying the intraclass correlation coefficient (ICC) of radiomic features in different volumes of interest (VOIs) in fractional HyperSight-CBCT scans during adaptive SBRT of prostate cancer: Prostate & Seminal vesicle (pooled), Femora & Penile Bulb (pooled), Prostate, Seminal vesicle, Femora (pooled left &right) and Penile Bulb. Colors ranging from dark red (ICC close to 1.0, indicating very high stability) to blue (ICC near 0.5, indicating low stability). Radiomic features: conventional values (CONVEN), shape-based features (SHAPE), grey-level co-occurrence matrix (GLCM), grey-level run length matrix (GLRLM), neighboring grey-level dependence matrix (NGLDM), and grey-level zone length matrix (GLZLM).

Fig. 3

Intraclass correlation coefficient of radiomic features in fractional HyperSight-CBCT scans during adaptive SBRT of prostate cancer: (a) Bar plot illustrating the percentage of radiomic features with very high (ICC > 0.9), high (ICC = 0.75–0.9), moderate (ICC = 0.5–0.75), and low (ICC < 0.5) stability across different volumes of interest. (b) Bar plot showing the distribution of ICCs of radiomic feature groups (conventional values (CONVEN), shape matrix (SHAPE), grey-level co-occurrence matrix (GLCM), grey-level run length matrix (GLRLM), neighboring grey-level dependence matrix (NGLDM), and grey-level zone length matrix (GLZLM)) across all volumes of interest (VOIs).

Fig. 4

Intraclass correlation coefficient (ICC) of radiomic features in fractional HyperSight-CBCT scans during adaptive SBRT of prostate cancer: Violin plots showing the distribution of ICC values for each volume of interest (VOI). Black points represent individual ICCs for each radiomic feature within each VOI. White points represent the median ICC values. Significant differences between VOIs, as determined by pairwise comparisons, are indicated by * p ≤ 0.05, ** p ≤ 0.01, and **** p ≤ 0.0001.

Fig. 5

Intraclass correlation coefficient (ICC) of radiomic features in fractional HyperSight-CBCT scans during adaptive SBRT of prostate cancer: Violin plots showing the distribution of ICCs in different radiomic feature groups: conventional (CONVEN), second-order (NGLDM, GLZLM, GLCM, GLRLM), shape-based features (SHAPE). Black points represent individual ICCs for each radiomic feature. White points indicate median. Significant differences between feature groups are marked with * p ≤ 0.05 and ** p ≤ 0.01.

Fig. 6

Boxplots showing the distribution of differences in radiomic features in fractional HyperSight-CBCT scans between fractions 2–5 and fraction 1 during adaptive SBRT of prostate cancer: Colored boxes denote the distribution of differences for each volume of interest (VOI). Black lines represent the median, box edges denote the interquartile range (IQR), whiskers represent the range within 1.5 times the IQR. A pronounced difference can be seen longitudinally for the VOI prostate.