Elekta (Sweden)

Digital Breast Tomosynthesis (DBT) has a diagnostic accuracy higher than 2D-FFDM, with a slightly higher Average Glandular Dose (AGD). Photon-counting technology was demonstrated to reduce AGD of 2D-FFDM, while preserving a high image quality; it therefore represents an interesting solution to significantly reduce the radiation dose of DBT. Our purpose is to compare the image quality and the AGD of a photon-counting DBT prototype with a conventional DBT unit in the detection of breast carcinoma. In this preliminary prospective study, 17 women with a suspicious breast finding (BI-RADS 4c/5) were enrolled. Before biopsy, they underwent bilateral 2-view DBT with a conventional unit and a photon-counting prototype. The AGD delivered by the two DBT units were compared. Three readers independently reviewed and compared the images (DBT and synthesized 2D) giving a density, BI-RADS and conspicuity score focusing on suspicious findings. All women underwent core-needle biopsy with imaging guidance and histology was used as standard reference. In 15 examinations at least one suspicious finding was detectable by both systems. Two exams were falsely negative with both units due to high density and small cancer size. All suspicious findings were histologically confirmed as malignant. There was no significant difference in density and BI-RADS scores. The conspicuity of findings was scored as equal or better for photon-counting DBT in 94% observations. The mean AGD delivered by photon-counting DBT was 55% lower than the conventional DBT. According to these preliminary data, photon-counting DBT provides a high image quality, like a conventional DBT unit, with a drastic reduction of the AGD delivered. The clinical trial "Low DoseTomosynthesisCompared to TraditionalTomosynthesis" has an identifier number NCT02771444 on the ClinicalTrials.gov website. It was first registered on 11.05.2016.

Abstract The purpose of this document is to provide the qualified medical physicist (QMP) with guidance on the critical evaluation and independent validation of “closed” or “black box” systems in a radiation oncology setting. Radiotherapy delivery systems and their associated subsystems are highly sophisticated and complicated. In recent years, vendors have worked closely with QMPs, information technology (IT) personnel, and clinical engineers to develop and ultimately provide an entire package of resources at the time of purchase of radiotherapy delivery equipment. Commissioning and routine quality assurance (QA) on these new, black‐box systems is not necessarily more difficult or time‐consuming; however, there are different factors to consider with black‐boxes. Independence in the vendor's own validation techniques now becomes critical to establish. For vendor‐provided tools, the user must understand the implications of working with a service tool or a true QA tool. Lastly, the user must determine which components of the system are unknown and differ from a classic white‐box system. Understanding these characteristics of the black‐box will guide the user in determining which quality management (QM) tools are applicable and which verification and validation (V&V) procedures to follow. Black‐box systems are becoming more and more prevalent in the radiation oncology setting. For example, true black‐box systems in the form of machine learning (ML) algorithms are already widely used within common treatment planning systems (TPS). These systems present a unique challenge to the QMP who is responsible for conducting independent V&V performance measurements on such systems. Although these systems require a different approach from classic treatment delivery systems, we present new terms for characterizing black‐box systems and a methodology for using alternative methods of independent validation.

Intensity-modulated radiotherapy (IMRT) in combination with magnetic resonance imaging (MRI)-guided gated delivery represents the latest development in the treatment of abdominothoracic tumours on MR-linac. In contrast, volumetric-modulated arc therapy (VMAT) is typically used on conventional linacs due to its superior delivery efficiency and speed. Non-inferior VMAT plans were created in a research treatment planning system for eight lung cancer patients previously treated on an MR-linac. VMAT plans were delivered on a moving dosimeter using respiratory multi-leaf collimator (MLC) tracking. VMAT with MLC tracking achieved an average 2%/2 mm local gamma pass rate of 93% relative to planned dose with a delivery efficiency of 83%.

Cardiac radioablation is a noninvasive stereotactic body radiation therapy (SBRT) technique to treat patients with refractory ventricular tachycardia (VT) by delivering a single high-dose fraction to the VT isthmus. Cardiorespiratory motion induces position uncertainties resulting in decreased dose conformality. Electocardiograms (ECG) are typically used during cardiac MRI (CMR) to acquire images in a predefined cardiac phase, thus mitigating cardiac motion during image acquisition. We demonstrate real-time cardiac physiology-based radiotherapy beam gating within a preset cardiac phase on anMR-linac. MR images were acquired in healthy volunteers (n = 5, mean age = 29.6 years, mean heart-rate (HR) = 56.2 bpm) on the 1.5 T Unity MR-linac (Elekta AB, Stockholm, Sweden) after obtaining written informed consent. The images were acquired using a single-slice balance steady-state free precession (bSSFP) sequence in the coronal or sagittal plane (TR/TE = 3/1.48 ms, flip angle = 48 , SENSE = 1.5, , voxel size = , partial Fourier factor = 0.65, frame rate = 13.3 Hz). In parallel, a 4-lead ECG-signal was acquired using MR-compatible equipment. The feasibility of ECG-based beam gating was demonstrated with a prototype gating workflow using a Quasar MRI4D motion phantom (IBA Quasar, London, ON, Canada), which was deployed in the bore of the MR-linac. Two volunteer-derived combined ECG-motion traces (n = 2, mean age = 26 years, mean HR = 57.4 bpm, peak-to-peak amplitude = 14.7 mm) were programmed into the phantom to mimic dose delivery on a cardiac target in breath-hold. Clinical ECG-equipment was connected to the phantom for ECG-voltage-streaming in real-time using research software. Treatment beam gating was performed in the quiescent phase (end-diastole). System latencies were compensated by delay time correction. A previously developed MRI-based gating workflow was used as a benchmark in this study. A 15-beam intensity-modulated radiotherapy (IMRT) plan ( Gy) was delivered for different motion scenarios onto radiochromic films. Next, cardiac motion was then estimated at the basal anterolateral myocardial wall via normalized cross-correlation-based template matching. The estimated motion signal was temporally aligned with the ECG-signal, which were then used for position- and ECG-based gating simulations in the cranial-caudal (CC), anterior-posterior (AP), and right-left (RL) directions. The effect of gating was investigated by analyzing the differences in residual motion at 30, 50, and 70% treatment beam duty cycles. ECG-based (MRI-based) beam gating was performed with effective duty cycles of 60.5% (68.8%) and 47.7% (50.4%) with residual motion reductions of 62.5% (44.7%) and 43.9% (59.3%). Local gamma analyses (1%/1 mm) returned pass rates of 97.6% (94.1%) and 90.5% (98.3%) for gated scenarios, which exceed the pass rates of 70.3% and 82.0% for nongated scenarios, respectively. In average, the gating simulations returned maximum residual motion reductions of 88%, 74%, and 81% at 30%, 50%, and 70% duty cycles, respectively, in favor of MRI-basedgating. Real-time ECG-based beam gating is a feasible alternative to MRI-based gating, resulting in improved dose delivery in terms of high rates, decreased dose deposition outside the PTV and residual motion reduction, while by-passing cardiac MRIchallenges.

Objective: Automated medical image segmentation (MIS) using deep learning has traditionally relied on models built and trained from scratch, or at least fine-tuned on a target dataset. The Segment Anything Model (SAM) by Meta challenges this paradigm by providing zero-shot generalisation capabilities. This study aims to develop and compare methods for refining traditional U-Net segmentations by repurposing them for automated SAM prompting. Approach: A 2D U-Net with EfficientNet-B4 encoder was trained using 4-fold cross-validation on an in-house brain metastases dataset. Segmentation predictions from each validation set were used for automatic sparse prompt generation via a bounding box prompting method (BBPM) and novel implementations of the point prompting method (PPM). The PPMs frequently produced poor slice predictions (PSPs) that required identification and substitution. A slice was identified as a PSP if it (1) contained multiple predicted regions per lesion or (2) possessed outlier foreground pixel counts relative to the patient’s other slices. Each PSP was substituted with a corresponding initial U-Net or SAM BBPM prediction. The patients’ mean volumetric dice similarity coefficient (DSC) was used to evaluate and compare the methods’ performances. Main results: Relative to the initial U-Net segmentations, the BBPM improved mean patient DSC by 3.93 ± 1.48% to 0.847 ± 0.008 DSC. PSPs constituted 20.01–21.63% of PPMs’ predictions and without substitution performance dropped by 82.94 ± 3.17% to 0.139 ± 0.023 DSC. Pairing the two PSP identification techniques yielded a sensitivity to PSPs of 92.95 ± 1.20%. By combining this approach with BBPM prediction substitution, the PPMs achieved segmentation accuracies on par with the BBPM, improving mean patient DSC by up to 4.17 ± 1.40% and reaching 0.849 ± 0.007 DSC. Significance: The proposed PSP identification and substitution techniques bridge the gap between PPM and BBPM performance for MIS. Additionally, the uniformity observed in our experiments’ results demonstrates the robustness of SAM to variations in prompting style. These findings can assist in the design of both automatically and manually prompted pipelines.

Methods described in the literature for designing shielding for treatment rooms for radiotherapy systems often involve assumptions that lead to overestimations of the wall thicknesses required to meet dose rate constraints outside the room. The Leksell Gamma Knife (LGK) has built-in shielding that results in primarily scattered photons leaking into the room. The field of leakage radiation, therefore, has a wide spectrum of energies, up to the primary energies of cobalt-60, and is highly anisotropic, making standard site planning methods difficult to adapt to the LGK. Monte Carlo (MC) simulations are an alternative for dose calculations but are computationallyexpensive. The aim was to develop a dose calculation algorithm for fast estimations of dose rates outside the barriers of a treatment room. The algorithm could then be employed in iterative methods to optimize treatment room parameters such as wall thicknesses and position of the LKGunit. The algorithm uses pre-calculated MC simulation data in two steps. First, it uses phase spaces that describe single photons in the radiation field around the LGK. Using each photon's position and direction, they are raytraced to the outside of the room. Based on their energy, angle of incidence to the barrier, and the barrier's thickness, a depth-dose profile is sampled from a pre-calculated library of profiles. The contribution from each photon is added to the space outside the barriers, giving the total dose distribution. The barrier thicknesses are optimized by iteratively running the algorithm and finding the minimum thickness required to keep the resulting maximum dose rate outside below a given upperlimit. Comparisons between dose distributions from the dose algorithm and full MC simulations of two typical rooms show good agreement, with the maximum dose rate outside each barrier being estimated within what represents a 2 cm error in concrete barrier thickness. The algorithm is successfully used in optimizing the barrier thicknesses and the results show potential decreases in most barriers' thickness, compared to a conventionalmethod. The algorithm is a promising alternative to conventional barrier design methods, eliminating several shortcomings of the latter whilst being significantly faster than a full MC simulation. The flexibility of the algorithm would further enable solving more complex cases, such as optimizing the LGK position or barrier material to further reduce material use andcost.