Purpose: Respiratory-induces tumour motion in lung cancer poses challenges for precise radiotherapy delivery. Free-breathing radiotherapy, guided by 4D-CT, helps mitigate these effects by treating the entire breathing cycle. However, standard treatment planning relies on a static, time-averaged CT, prescribing doses to the total movement of the tumour, which may lead to suboptimal dose distributions when considered with intensity-modulated delivery methods. This study investigates interplay effects of 3D-CT-based VMAT SABR planning by comparing it with 4D-CT-based recalculations.

 

Methods: Eight patients with early-stage non-small cell lung cancer (NSCLC) treated with free-breathing VMAT SABR were analyzed. An in-house MATLAB script divided radiation delivery to each phase of their 4D-CT based on the patient’s respiratory trace. Phase-specific doses were recalculated using AcurosXB in Eclipse v15.6 and accumulated onto end-inhale and end-exhale reference CTs. Dose coverage of the tumour was evaluated using D95 and compared to GTV/IGTV ratios and Hounsfield Unit (HU) values.

 

Results and Discussion: Recalculated D95 errors (recalculated D95/original D95) ranged from 0.86 to 1.08. Two cases showed notable underdosage (≤0.86), while others showed dose variations exceeding 7%, highlighting potential under- or overdosing risks with 3D-CT-based planning. GTV-to-IGTV and mean CT density ratios varied from 25%–93% and 0.3%–58% respectively, demonstrating a complex relationship between motion and treatment planning.

 

Conclusion: 4D-CT-based dose analysis highlights interplay effects that contribute to dose discrepancies in VMAT, revealing potential under- and overdosing risks. These findings suggest that leveraging 4D-CT data could improve the identification of dose errors, particularly in tumor coverage, by accounting for respiratory motion

Purpose

18F-fluorodeoxyglucose positron emission tomography (18F-FDG-PET) has been explicitly implicated in improved interobserver variation of the residual GTV in cervical brachytherapy. However, a solution that can be integrated with MR imaging performed at the time of brachytherapy is desired. Chemical exchange saturation transfer (CEST) measures unlabeled D-glucose (D-Glc) using MRI. Using non-radioactive glucose in cervical cancer applications has yet to be investigated. This project aims to assess the potential of glucoCEST-MRI as an alternative to FDG-PET in the context of cervical brachytherapy.

Methods

A study was conducted to optimize continuous wave saturation imaging sequences described in the literature under physiological conditions. NMR tubes containing 0-20mM D-Glc in PBS (pH 6.6) were imaged at 7T (Bruker Biospec MRI; 40 mm inner diameter quadrature RF coil) using a CEST-RARE sequence (TR =10s, 64 x 64 matrix, RARE factor=16, NEX=1, 5s presaturation block pulse at 2µT, 49 frequency offsets between –6 and 6 ppm, 0.25ppm increment, ~30 minutes). Saturation transfer was calculated through asymmetry analysis.

Results

The strongest MR signal modulation was observed at 1.25 ppm, consistent with the known frequency offset of glucose hydroxyl protons. This signal modulation ranged from 9 to 23% across glucose concentrations from 5 to 20 mM. Saturation transfer increased linearly with glucose concentration.

Conclusions

This work aims to facilitate metabolic imaging of cervical tumors at the time of brachytherapy using glucoCEST. The results were sufficiently strong to support feasible translation to murine models bearing patient-derived cervical cancer xenografts.

Purpose: This study aims to extend CT Perfusion (CTP) software for simultaneous imaging of lung ventilation and perfusion, addressing limitations of traditional Ventilation-Perfusion (VQ) SPECT scans such as limited spatial resolution, time inefficiency, time mismatch between ventilation and perfusion scans, and qualitative evaluation.

Methods: Four pigs were examined in the supine position using dynamic CT with contrast. The pigs were anesthetized and mechanically ventilated at 20 breaths per minute (BPM). The lung tissue time density curves (TDCs) were derived, comprising signals from perfusion, cardiac and respiratory motion, and ventilation. Non-rigid registration was applied to remove motion artifacts. Perfusion TDCs and perfusion were computed using the Johnson-Wilson-Lee (JWL) model across different lung segments and subtracted from the motion-corrected lung TDCs to extract ventilation TDCs. Fourier transform was applied to the ventilation curves to determine the dominant ventilation frequency and amplitude.

Results: Fourier transform-derived ventilation frequency was 0.35 Hz in agreement with the ventilation rate of 20 BPM. Ventilation amplitude and perfusion were higher in the posterior lung segments compared to the anterior, consistent with gravitational effects and tissue compression in the supine position.

Conclusion: This study demonstrates the feasibility of using CTP to simultaneously assess lung ventilation and perfusion, eliminating the need for two separate studies. CT is also more accessible than SPECT in emergent settings. Future work will focus on generating comprehensive ventilation maps to evaluate ventilation gradients across the entire lung, with further validation using 68Ga-Galligas for ventilation and 68Ga-MAA for perfusion.

Purpose: This work investigates the application of a joint reconstruction algorithm to MP2RAGE data to accelerate quantitative T1 mapping by sharing k-space information between acquired anatomical volumes.

 

Methods: Reconstructed 1 mm3 MP2RAGE scans from a public database (n = 50) were used for retrospectively undersampling k-space. GRAPPA was used as a reference algorithm and JVC-GRAPPA for joint reconstruction. Image quality was assessed with the root mean square error (RMSE) between the reconstructed and provided magnitude data at a reduction factor R = 6 and as the size of the autocalibration signal (ACS) is lowered. The MP2RAGE anatomical volumes (INV1 and INV2) were used to estimate a T1 map.

Results: At ACS = 32, JVC-GRAPPA reduced the mean RMSE by 38% for INV2 (p < 0.001) and increased it by 18% for INV1 (p < 0.001).  The mean RMSE of the estimated T1 maps was reduced by 31% (p < 0.001). Lowering to ACS = 24, JVC-GRAPPA reduced the mean RMSE by 12% (p < 0.001) for INV2 and increased it by 340% (p < 0.001) for INV1. The mean RMSE of the estimated T1 maps was increased by 1% (p = 0.7479).

Conclusions: At R = 6 and ACS = 32, JVC-GRAPPA demonstrated improvements to estimated T1 maps. This study has provided initial evidence of the efficacy of joint reconstruction being applied to reconstructing MP2RAGE anatomical volumes, allowing for improved T1 mapping at higher reduction factors, ultimately accelerating the acquisition.

Photon-counting (PC) computed tomography (CT) systems have currently been introduced into clinical practice. Due to their high costs, the technology is restricted to 3rd generation CT, which has prominent ring artifacts from PC detector non-linearity. Current image artifact correction techniques can generate artifacts and remove anatomical structures. Our study proposes a novel and simple method of reducing ring artifacts by translating the detector during the scan.

 

In simulation, the detector was translated horizontally and/or vertically during imaging for one to two revolutions. The maximum magnitude above the background of the artifacts was compared between translational (Gen3+) CT with regular (Gen3) CT for an uniform water phantom. In experiment, a benchtop cadmium zinc telluride PC CT system was used with a quantitative CT phantom. The detector was translated horizontally at 3.6mm/360 degrees over a 500-degree acquisition. In addition to ring artifact magnitude, the contrast-to-noise ratio (CNR) was also measured.

 

In simulation, Gen3+ CT reduced ring artifacts magnitude by 78% for a single rotation and up to 90% for two rotations. In experiment Gen3+ CT reduced ring artifact magnitude by between 30% to 80% in comparison to regular CT. Additionally, the CNR was improved by up to 15% depending on the severity and frequency of the ring artifacts and corrected unexpected streaking artifacts.
Adding a small amount of translation to the detector during PCCT acquisition improves image quality while suppressing artifacts. Gen3+ may be implemented in clinical CT systems by desyncing rotation speed between the source and the detector.

Purpose: To evaluate a novel ultra-low dose volumetric 4D-CT protocol designed to achieve a more accurate tumor motion model without compromising the image quality needed for CT simulation utilizing a raw-data projection-based averaging technique.

Methods: Catphan504 and QUASAR Respiratory Motion Phantoms were imaged on a 320-slice Aqullion ONE CT scanner for 60-sec using a 160mm axial field-of-view, fixed table position, 120kVp, 0.275-sec rotation, and tube current of 10mA. A 300mA reference scan was also performed. The 10mA Catphan504 projection data was exported from the scanner and averaged at each gantry angle using an in-house MATLAB script. The averaged projection data was then reimported to the scanner for reconstruction. Image noise and signal-difference-to-noise-ratio (SDNR) were assessed as a function of the number of rotations averaged. The QUASAR phantom was imaged using a real patient waveform, with embedded inserts automatically segmented. The extracted motion data was compared to the ground-truth motion using coefficient of determination (r²).

Results: For a 60-second averaged acquisition, noise reduction was observed to exceed a factor of 5 compared to a single 10mA scan. SDNR increased by a factor of 8 – 10 for each low-contrast insert and relative to the 300mA scan for all inserts. Motion profiles extracted from the 10mA QUASAR scan demonstrated high correlation with the ground-truth motion (r²>0.99, p<0.001), with no motion artifacts detected.

Conclusion: Our novel ultra-low-dose volumetric 4D-CT protocol can effectively generate a more accurate 4D model of tumour motion compared to conventional 4D-CT, while preserving image quality suitable for 4D-CT simulation radiotherapy.

Purpose: The wide-bore Canon Aquilion Exceed LB is capable of helical and volumetric dual-energy (DE) data acquisition. CT number linearity and quality of DE blended 120 kV-equivalent images are comparable to single-energy helical 120 kVp (SE-120) images used for radiotherapy simulation. This study explores the generation of virtual monochromatic images (VMIs) from DE data and their ability to enhance iodine contrast visibility.

Methods: A CIRS Electron Density phantom with cylindrical inserts containing 30, 20, 15, 10, 7.5, 5.0, 1.0 and 0.5 mg iodine/mL was imaged using SE-120 and dual-energy 80/135 kVp volumetric (DEVOL) modes. Synthetic 120 kV-equivalent blended images (DEVOL-120) and VMIs across 35-135 keV in 5 keV increments were reconstructed from the DEVOL data and compared to SE-120 images in terms of contrast and contrast-to-noise ratio (CNR) for each iodine insert.

Results: Contrast was maximized for all iodine concentrations at 35 keV VMI, exceeding SE-120 image contrast by a factor of 4.7–5.2 depending on the concentration considered. Noise was highest in the 35 keV VMI at 108 HU. CNR was maximized for all iodine concentrations at 65 keV VMI, exceeding that quantified in the SE-120 image by a factor of 1.6–1.9 depending on the concentrations considered.

Conclusions: Enhanced iodine contrast visibility has been demonstrated using VMIs derived from DEVOL data on the Canon Aquilion Exceed LB. Iodine contrast and CNR were maximized at 35 and 65 keV, respectively. DECT-based radiotherapy simulation may enhance target volume delineation with auto-registered VMIs relative to conventional SECT when contrast is used.

Purpose: White matter in multiple sclerosis (MS) is highly heterogeneous and changes over time (Ellen et al., 2023; Lucchinetti et al., 2000). This study aimed to investigate whether baseline quantitative magnetic resonance imaging (qMRI) metrics differed in stable and transitioning regions in and around white matter lesions (WML) over the subsequent 2 years.

Methods: Ten baseline qMRI maps and longitudinal WML masks (2-year followed-up MS cohort) were obtained from the Comorbidity and Cognition in MS (CCOMS) Study (Uddin et al., 2022). Three WML subtypes were classified at the individual lesion level: i) newly appearing, ii) vanishing, and iii) enduring (with “expanding”, “contracting”, and “persisting” subregions). “Stable” or “transitioning” perilesional normal appearing white matter regions were also computed. Comparisons were made among primary WML types, among enduring WML subregions, as well as between subtypes/subregions that were longitudinally or cross-sectionally identical. Wilcoxon rank sum and Wilcoxon signed rank tests were used accordingly (all tests two-tailed and FDR-corrected).

Results: Across the full MS cohort, 2123 unique WMLs were identified, with 623 newly appearing WMLs, 519 vanishing WMLs, 978 enduring WMLs. Overall, significant differences were found for 83/110 comparisons across WML subregions with baseline qMRI measures.

Conclusions: Baseline qMRI measures exhibited significant differences between regions defined by longitudinal WML masks over 2-years. The sensitivity of baseline qMRI measures to differences among these subtypes/subregions suggests that they could potentially predict the evolution of WMLs.

Purpose: To compare AUC and ADC’s (diffusion MRI parameters) association with quantifying changes in parotid glands pre- vs. post-radiation therapy.

 

Methods: Anatomical and diffusion MR images of 13 head-and-neck cancer patients were acquired before and after radiotherapy with bilateral parotid glands (n=26) contoured and processed using 3D Slicer and DICOMautomaton. Diffusion parameters were extracted using two models: apparent diffusion coefficient (ADC) using least-squares, and the area-under-the-curve (AUC) of the signal decay curve. Parameter distribution metrics (min, max, mean, median, and Q1 and Q3 percentiles) were evaluated against radiation dose statistics using Spearman rank correlation coefficients, with Bonferroni correction and Wilcoxon rank-sum testing to compare correlations.

 

Results: The strongest correlations were observed between maximum dose and Q3 diffusion parameter changes, with the top four Spearman correlation coefficients (three from AUC metrics, one from ADC) being 0.54 (p=0.005), 0.49 (p=0.012), 0.47 (p=0.015), and 0.47 (p=0.016) . Notably, AUC-derived metrics demonstrated stronger overall correlations at the extremes of dose distribution (minimum and maximum values), compared to ADC. These findings indicate AUC can quantify radiation effects on parotid tissue at dose extremes where distinct biological damage mechanisms occur.

 

Conclusions: AUC diffusion metrics may offer advantages over traditional ADC parameters for quantifying radiation-induced changes in parotid glands, particularly when correlating with minimum and maximum radiation dose distributions. Future research with larger patient cohorts is needed to validate these preliminary findings and determine whether AUC offers meaningful advantages over ADC in specific clinical contexts.