Trial Outcomes & Findings for Evaluation of Novel Cone-Beam CT for Guidance and Adaptation of Precision Radiotherapy (NCT NCT05176860)

Patients who withdraw consent, who are removed from the study by the investigator, or who otherwise end their participation in the study prior to the acquisition of any study imaging could be replaced.

Race and Ethnicity were not collected from any participant.

Artifact Index (AI) is a measurement of the strength of imaging artifact and the degree to which is affects visibility of anatomical structures in the vicinity of the artifact. Artifacts can be produced in CT and CBCT images by a number of factors, such as metal implants, gas, or breathing motion. AI = sqrt((STD\_VOI)\^2 - (STD\_background)\^2), where STD\_VOI is the standard deviation of the image Hounsfield Units in a region of interest at the location of an artifact, and STD\_background is the standard deviation of the Hounsfield Unit values in the background (i.e. in similar tissue but away from the artifact. A lower AI value indicates that the artifact has a lower impact on image quality. Artifacts were identified in all study participants. The median AI across the study population is presented for four imaging modalities.

Nonuniformity (NU) is a measure of the variation of CT image intensity in uniform tissue. NU = (HU\_max - HU\_min)/(HU\_max + HU\_min), where HU\_max and HU\_min are the maximum and minimum Hounsfield Unit values among multiple locations sampled within regions of uniform tissue that were relevant to the anatomy of interest (e.g., a uniform region of breast tissue for patients undergoing breast treatments). A lower NU represents greater uniformity of CT image intensity within a region of interest. Median NU across the study population is presented for four imaging modalities.

Contrast represents the ability to distinguish between two different regions in a CT image (e.g. to distinguish between two adjacent organs). Contrast = \|HU1 - HU2\| where HU1 and HU2 are the mean HU values in two different 100 mm\^2 ROIs, where the ROIs were located in two different tissue types that were relevant to the site being treated (e.g., in the liver and in perihepatic fat for liver treatments). Higher contrast values indicate that it is easier to distinguish between regions (anatomical structures) in a CT image. Median contrast across the study population is presented for four imaging modalities.

Contrast to Noise Ratio (CNR) measures the ability to distinguish an object or lesion from its background. CNR = \|HU1 - HU2\|/\[0.5 (STD1 + STD2)\] where HU1 and HU2 are the mean Hounsfield Unit values in two different 100 mm\^2 ROIs, where the ROIs were located in two different tissue types that were relevant to the site being treated (e.g., in the liver and in perihepatic fat for liver treatments), and STD1 and STD2 are the standard deviations of the HU values in those same ROIs. A higher CNR makes it easier to distinguish an object from its background. CNR analysis was limited to images with similar imaging dose. Median CNR across all study participants treated for lung cancer are presented for three CBCT modalities.

The intensity of a pixel in a CT image is a function of its Hounsfield Unit (HU) value. HU is also directly related to the underlying electron density, which means that the pixel value of a CT image can be used directly in the calculation of dose for a prescribed radiation treatment plan. CT simulation scanners produce images with high HU accuracy and are regularly used for radiation treatment planning. Here, we present the difference in HU between CT simulation images and different CBCT images. ΔHU = HU\_CBCT - HU\_CTSim, where HU\_CBCT and HU\_CTSim are mean values among HU averages at 4 reference points in a CBCT image and the corresponding CT simulation image, respectively. The lower the ΔHU, the greater the HU accuracy of the CBCT image, and the greater the likelihood that CBCT imaging can be used for radiation treatment planning. Median ΔHU across the study population are presented for three different tissue types for three CBCT imaging modalities.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. The overall difference between calculated radiation distributions was evaluated using three different gamma pass criteria: 3% dose difference / 3 mm distance to agreement, 2%/2mm, and 1%/1mm. The gamma pass rate is expressed as a percentage of data points that meet the pass criteria. A gamma pass rate of \> 95% is typically considered acceptable for 3%/3mm. As the gamma pass criteria become stricter, the pass rates decrease. Gamma pass rates were calculated to compare the CT simulation-based dose calculation and the high performance CBCT-based dose calculation. Gamma pass rates were also calculated to compare the CT simulation-based dose calculation and the conventional CBCT-based dose calculation. The median gamma pass rates across the entire study population are presented.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. Dose-volume histograms (DVH) were calculated for individual target structures from all three dose distributions. Individual DVH metrics, such as V90(%) (the percentage of the structure volume receiving 90% of the prescribed radiation dose) were extracted for individual target structures from their DVH. The difference between a DVH metric derived from CT simulation-based dose calculation and the same metric derived from a CBCT-based dose calculation are reported. The smaller the difference, the greater the accuracy of the CBCT-based dose calculation. Median target DVH metric differences across the study population are presented.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. Dose-volume histograms (DVH) were calculated for individual target structures from all three dose distributions. Individual DVH dose metrics, such as D95(%) (the minimum dose covering 95% of the structure, expressed as a % of the prescription dose) were extracted for individual target structures from their DVH. The difference between a DVH metric derived from CT simulation-based dose calculation and the same metric derived from a CBCT-based dose calculation are reported. The smaller the difference, the greater the accuracy of the CBCT-based dose calculation. Median target DVH metric differences across the study population are presented.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. Dose-volume histograms (DVH) were calculated for individual organs at risk (OAR) from all three dose distributions. The key organs at risk for patients being treated for breast cancer are the heart, ipsilateral lung, and contralateral breast. The differences between the D2%(%) (minimum dose received by the "hottest" 2% of the OAR, expressed as a % of the prescription dose) derived from CT simulation-based dose calculation and the same metric derived from a CBCT-based dose calculation are reported. The smaller the difference, the greater the accuracy of the CBCT-based dose calculation. Median differences in OAR D2%(%) across study participants treated for breast cancer are presented.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. Dose-volume histograms (DVH) were calculated for individual organs at risk (OAR) from all three dose distributions. The key organs at risk for patients being treated for lung cancer are the heart, esophagus and spinal cord. The differences between the D2%(%) (minimum dose received by the "hottest" 2% of the OAR, expressed as a % of the prescription dose) derived from CT simulation-based dose calculation and the same metric derived from a CBCT-based dose calculation are reported. The smaller the difference, the greater the accuracy of the CBCT-based dose calculation. Median differences in OAR D2%(%) across study participants treated for lung cancer are presented.

Every trial participant had a radiation treatment plan calculated on their CT simulation image series. That same plan was then re-calculated on both the breath hold high-performance CBCT and conventional CBCT. Dose-volume histograms (DVH) were calculated for individual organs at risk (OAR) from all three dose distributions. The key organs at risk for patients being treated for abdominal cancer are the heart, bowel and kidneys. The differences between the D2%(%) (minimum dose received by the "hottest" 2% of the OAR, expressed as a % of the prescription dose) derived from CT simulation-based dose calculation and the same metric derived from a CBCT-based dose calculation are reported. The smaller the difference, the greater the accuracy of the CBCT-based dose calculation. Median differences in OAR D2%(%) across study participants treated for abdominal cancer are presented.