Identification of reproducible radiomic features from on-board volumetric images: A multi-institutional phantom study.

BACKGROUND: Radiomics analysis using on-board volumetric images has attracted research attention as a method for predicting prognosis during treatment; however, the lack of standardization is still one of the main concerns. PURPOSE: This study investigated the factors that influence the reproducibility of radiomic features extracted from on-board volumetric images using an anthropomorphic radiomics phantom. Furthermore, a phantom experiment was conducted with different treatment machines from multiple institutions as external validation to identify reproducible radiomic features. METHODS: The phantom was designed to be 35 × 20 × 20 cm with eight types of heterogeneous spheres (⌀ = 1, 2, and 3 cm). On-board volumetric images were acquired using 15 treatment machines from eight institutions. Of these, kilovoltage cone-beam computed tomography (kV-CBCT) image data acquired from four treatment machines at one institution were used as an internal evaluation dataset to explore the reproducibility of radiomic features. The remaining image data, including kV-CBCT, megavoltage-CBCT (MV-CBCT), and megavoltage computed tomography (MV-CT) provided by seven different institutions (11 treatment machines), were used as an external validation dataset. A total of 1,302 radiomic features, including 18 first-order, 75 texture, 465 (i.e., 93 × 5) Laplacian of Gaussian (LoG) filter-based, and 744 (i.e., 93 × 8) wavelet filter-based features, were extracted within the spheres. The intraclass correlation coefficient (ICC) was calculated to explore feature repeatability and reproducibility using an internal evaluation dataset. Subsequently, the coefficient of variation (COV) was calculated to validate the feature variability of external institutions. An absolute ICC exceeding 0.85 or COV under 5% was considered indicative of a highly reproducible feature. RESULTS: For internal evaluation, ICC analysis showed that the median percentage of radiomic features with high repeatability was 95.2%. The ICC analysis indicated that the median percentages of highly reproducible features for inter-tube current, reconstruction algorithm, and treatment machine were decreased by 20.8%, 29.2%, and 33.3%, respectively. For external validation, the COV analysis showed that the median percentage of reproducible features was 31.5%. A total of 16 features, including nine LoG filter-based and seven wavelet filter-based features, were indicated as highly reproducible features. The gray-level run-length matrix (GLRLM) was classified as containing the most frequent features (N = 8), followed by the gray-level dependence matrix (N = 7) and gray-level co-occurrence matrix (N = 1) features. CONCLUSIONS: We developed the standard phantom for radiomics analysis of kV-CBCT, MV-CBCT, and MV-CT images. With this phantom, we revealed that the differences in the treatment machine and image reconstruction algorithm reduce the reproducibility of radiomic features from on-board volumetric images. Specifically, the most reproducible features for external validation were LoG or wavelet filter-based GLRLM features. However, the acceptability of the identified features should be examined in advance at each institution before applying the findings to prognosis prediction.

Multi-Institutional Study of End-to-End Dose Delivery Quality Assurance Testing for Image-Guided Brachytherapy Using a Gel Dosimeter.

PURPOSE: To quantify dose delivery errors for high-dose-rate image-guided brachytherapy (HDR-IGBT) using an independent end-to-end dose delivery quality assurance test at multiple institutions. The novelty of our study is that this is the first multi-institutional end-to-end dose delivery study in the world. MATERIALS AND METHODS: The postal audit used a polymer gel dosimeter in a cylindrical acrylic container for the afterloading system. Image acquisition using computed tomography, treatment planning, and irradiation were performed at each institution. Dose distribution comparison between the plan and gel measurement was performed. The percentage of pixels satisfying the absolute-dose gamma criterion was reviewed. RESULTS: Thirty-five institutions participated in this study. The dose uncertainty was 3.6% ± 2.3% (mean ± 1.96σ). The geometric uncertainty with a coverage factor of k = 2 was 3.5 mm. The tolerance level was set to the gamma passing rate of 95% with the agreement criterion of 5% (global)/3 mm, which was determined from the uncertainty estimation. The percentage of pixels satisfying the gamma criterion was 90.4% ± 32.2% (mean ± 1.96σ). Sixty-six percent (23/35) of the institutions passed the verification. Of the institutions that failed the verification, 75% (9/12) had incorrect inputs of the offset between the catheter tip and indexer length in treatment planning and 17% (2/12) had incorrect catheter reconstruction in treatment planning. CONCLUSIONS: The methodology should be useful for comprehensively checking the accuracy of HDR-IGBT dose delivery and credentialing clinical studies. The results of our study highlight the high risk of large source positional errors while delivering dose for HDR-IGBT in clinical practices.

Evaluation of technical performance of optical surface imaging system using conventional and novel stereotactic radiosurgery algorithms.

The Catalyst HD (C-RAD Positioning AB, Uppsala, Sweden) optical surface imaging (OSI) system is able to manage interfractional patient positioning, intrafractional motion monitoring, and non-contact respiratory gating without x-ray exposure for radiation therapy. In recent years, a novel high-precision surface registration algorithm for stereotactic radiosurgery (SRS algorithm) has been released. This study aimed to evaluate the technical performance of the OSI system using rigid phantoms, by comparing the conventional and SRS algorithms. To determine the system's technical performance, isocenter displacements were calculated by surface image registration via the OSI system using head, thorax, and pelvis rigid phantoms. The reproducibility of positioning was evaluated by the mean value calculated by repeating the registration 10 times, without moving each phantom. The accuracy of positioning was evaluated by the mean value of the residual error, where the 10 offset values given to each phantom were subtracted from the isocenter displacement values. The stability of motion monitoring was evaluated by measuring isocenter drift during 20 min and averaging it over 10 measurements. For the head phantom, all tests were compared with the mask types and algorithms. As a result, for all sites and both algorithms, the reproducibility, accuracy, and stability for translation and rotation were <0.1 mm and <0.1°, <1.0 mm and <1.0°, and <0.1 mm and <0.1°, respectively. In particular, the SRS algorithm had a small absolute error and standard deviation of calculated isocenter displacement, and a significantly higher reproducibility and accuracy than the conventional algorithm (P < 0.01). There was no difference in the stability between the algorithms (P = 0.0280). The SRS algorithm was found to be suitable for the treatment of rigid body sites with less deformation and small area, such as the head and face.