Purpose: The advent of novel therapeutic techniques and augmentations to high dose rate brachytherapy (HDR-BT) requires pre-clinical research to understand and quantify their effectiveness. This research aims to develop a platform for in vivo irradiations from an HDR-BT afterloader to study novel therapeutic techniques.

Methods: An irradiation jig was constructed as a flat treatment bed and an upright 7cm diameter semi-circle with eight channels for needle insertion from the afterloader to encompass the lateral side of the mouse. A CT scan was acquired of the jig along with silicone phantom of a mouse with a flank tumor. The scan was imported into OncentraTM for planning using 500cGy prescribed to the tumour. EBT-4 radiochromic film and OSLD measurements were employed to verify dose distributions in the axial and coronal directions along with the entrance and exit dose to the flank tumour. Five female immunodeficient mice inoculated with HEC-1A cervical cancer tumours were then irradiated and monitored for 3 weeks for tumour growth, body weight, and signs of toxicity.

Results: Dosimetry measurements before irradiation agreed within 5% of the tumour's entrance and exit dose as reported by OncentraTM using both computational formalisms. Compared to unirradiated mice, a significant reduction in tumour growth post-irradiation was observed in all irradiated mice with no observable signs of toxicity or changes in body weight.

Conclusion: We have successfully developed and piloted a radiation platform for safe and effective in vivo HDR-BT irradiations. Future research will implement this platform to evaluate novel therapeutic techniques in pre-clinical settings.

Purpose:
Prostate stereotactic body radiotherapy simultaneous in-field boost (SBRT-SIB) allows for boosting dominant intra-prostatic lesions (DILs). The DIL boost dose is often limited due to surrounding organs-at-risk (OARs). Our study investigates employing high-dose-rate brachytherapy (HDR-BT) with SBRT to enhance DIL boosting.

Methods:
Ten SBRT-SIB patients (35 Gy in 5 fractions to whole-prostate, up to 50 Gy DIL boost) were selected for retrospective SBRT replanning to a homogenous 30 or 35 Gy whole-prostate dose (two additional plans per patient). DIL contours were transferred to trans-rectal ultrasound (TRUS) via deformable registration to create BT DIL boost plans based on the SBRT-SIB dose constraints with the uniform SBRT contribution subtracted (up to 15 Gy in one fraction). Dose metrics were summed between the uniform SBRT and BT plans to create simulated “SBRT 35 Gy+BT” and “SBRT 30 Gy+BT” treatments, which were compared dosimetrically to the “SBRT-SIB” plan using the EQD2 formulism.

Results:
“SBRT 35 Gy+BT” achieved near-equivalent DIL boosts compared to “SBRT-SIB”, with higher whole-prostate coverage for many patients and similar OAR metrics. “SBRT 30 Gy+BT” achieved higher DIL boost doses (median DIL GTV D95 across “SBRT 30 Gy+BT” plans 27.1 Gy EQD2 higher than “SBRT-SIB” plans, p<0.001), while reducing rectum and bladder doses. Median prostate PTV D95 was lower by 13.7 Gy EQD2 for the “SBRT 30 Gy+BT” plans compared to “SBRT-SIB”.

Conclusions:
Employing the “SBRT 30 Gy+BT” strategy provided greater boost doses while reducing rectum and bladder dose. This required uniform SBRT de-escalation, which may marginally reduce whole-prostate coverage for some patients.

Purpose: High-dose-rate brachytherapy (HDR-BT) for prostate cancer requires accurate and precise 3D transrectal ultrasound (TRUS) imaging. Sagittally reconstructed 3D (SR3D) ultrasound imaging reduces dosimetric uncertainty compared to axially reconstructed approaches for HDR-BT treatment planning but suffers from image degradation caused by manual rotation of the TRUS probe. This study aimed to develop a motorized attachment for the clinically used CIVCO EX3 TRUS stepper enabling motorized SR3D acquisition, and to compare the quality of images acquired using manual and motorized methods.

Methods: An attachment that interfaces with the CIVCO EX3 stepper was developed to enable motorized rotation of the ultrasound probe. Geometric accuracy was validated by imaging a grid-phantom suspending wires spaced by known distances. Image quality of motorized and manual SR3D images was evaluated using peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and multi-scale SSIM (MS-SSIM) using a slowly acquired (4.2°/s) manual reference image.

Results: Geometric validation confirmed the attachment’s accuracy, with a maximum average error of 0.15±0.25mm (1.5%). Motorized acquisitions achieved superior image quality versus manual scans at comparable speeds, with higher PSNR, SSIM, and MS-SSIM values. Despite shorter acquisition time, the fastest motorized scan (10°/s) exceeded all manual scans in each metric.

Conclusions: The motorized attachment for the CIVCO EX3 stepper enabled improved SR3D image quality compared to manually acquired methods, overcoming a critical limitation in implementing SR3D imaging as a clinical standard. By enabling reliable SR3D imaging, our attachment could aid in reducing dosimetric uncertainties during HDR prostate BT, potentially improving patient outcomes.

Purpose  Purpose  
Brachytherapy (BT) is an essential component of curative-intent treatment for locally advanced cervical cancers; however, it is technically challenging with limited training opportunities. An educational simulator can be used to train professionals in a low-stress environment, promoting professional proficiency. We aimed to build and evaluate a BT simulator that is anatomically accurate, has realistic tactility, and is ultrasound and BT-applicator compatible.  

 

Methods  
Realistic organ molds of the bladder, cervix, vulva, and uterus were created based on images from an MRI scan with an implanted ring and tandem applicator. The rectum was modeled as a cylinder compatible with a transrectal ultrasound probe. All organs were cast in a tissue-mimicking plastisol mixture with differing concentrations of glass microbeads to provide ultrasound contrast.  Participants at a BT workshop completed surveys, rating skill and confidence of performing BT procedures on a ten-point scale before and after training on the simulator. They also answered a series of 11 questions evaluating the simulator.  

 

Results  
Five participants completed pre- and post-intervention surveys and skill practice with the simulator. Confidence in performing ultrasound-guided intracavitary BT procedures increased by two points on average. All participants agreed or strongly agreed that the pelvis simulator had anatomically correct size, shape and organ placement and should be promoted as a standard component of BT training. 

 

Conclusions 
Development of this new ultrasound-compatible, anatomically accurate BT simulator can act as a tool for training professionals. Feedback from residents, fellows, and practicing oncologists confirm its applicability and potential as a standard training tool.

Purpose: To determine the change in geometric performance of a high-risk clinical target volume (HR-CTV) auto-contouring model for cervical cancer (CC) high-dose-rate (HDR) brachytherapy when using different loss functions for training.

Methods: T2-weighted MRIs with HR-CTV contours from patients with CC receiving HDR brachytherapy between 2016 and 2024 were used to develop and test two auto-contouring models. Each model was built using PyTorch and the Segresnet architecture available in Medical Open Network for AI (MONAI). The Sorensen-Dice similarity coefficient (DSC) cross-entropy loss or Hausdorff distance (HD) loss were optimized during training. Each model was trained using the Adam optimizer for 100 epochs and tuned using the validation DSC and HD95% for each loss function. Average validation and testing DSC, HD95%, added path length (APL), and surface dice (SD) were calculated using 5-fold cross-validation and soft-voting, respectively, and compared using a paired Student’s t-test.

Results: The retrospective dataset included 120 patients with CC: 115 (184 MRIs) for development and 5 (11 MRIs) for testing. In validation, the mean DSC, HD95%, APL, SD for each model (DSC cross-entropy loss; HD loss) was (0.73±0.02; 0.72±0.02), (8.0±1.2mm; 7.7±0.7mm), (213±41mm; 165±47mm, p<0.05), and (0.79±0.03; 0.78±0.03). In testing, models achieved (0.69±0.09; 0.72±0.10, p<0.05), (11.4±4.2mm; 10.6±4.2mm, p<0.05), (412±301mm; 304±243mm, p<0.05), (0.71±0.14; 0.75±0.14, p<0.05) using DSC cross-entropy loss and HD loss, respectively.

Conclusions: Optimization with HD loss during training significantly improved testing DSC, HD95%, APL, and SD compared to DSC cross-entropy loss, likely leading to reduced manual edits required to achieve a clinically acceptable HR-CTV contour.

Purpose: To quantify and understand the clinical risks involved when using custom channel 3D printed templates for gynecologic interstitial brachytherapy.

Methods: The Brachify software (Kudla et al.) allows brachytherapy users to create 3D printed cylindrical templates (3DCTs) for patient-specific interstitial implants. In the current implementation, a pre-plan is generated in Oncentra Brachy (Elekta, Stockholm) placing desired needles within a dummy cylinder, then the DICOM files are exported into Brachify which generates a 3DCT based on needle trajectories. A failure modes and effects analysis (FMEA) was conducted for all clinical processes that needed modification due to the use of 3DCTs. These processes included cylinder design, 3D printing, machining, sterilization, implantation, pretreatment imaging, treatment planning, and treatment delivery. Failure modes were scored based on the product of severity, occurrence, and detection with a maximum score of 1000, and minimum score of 1.

Results: 42 failure modes were identified with RPN scores ranging from 1 to 400. The largest score corresponded to potential issues with immobilization which could lead to undesirable dosimetry. Other high risk failure modes involve improper cylinder rotations (RPN=150), improperly locked collets (RPN=150), short cylinder length leading to immobilization issues (RPM=320), and misidentification of cylinder edge in CT due to tissue equivalence of 3DCT material (RPN=200). Further controls will be implemented to mitigate these risks.

Conclusions: A comprehensive FMEA was performed on the workflow processes that will change in gynecologic interstitial brachytherapy when using custom channel 3D printed cylindrical templates. This type of analysis could be helpful to other Brachytherapy programs that would benefit from using this technology.