We propose a novel cone-beam CT (CBCT) reconstruction method to address internal patient motion in image-guided radiation therapy. The method assumes that the region of interest (ROI) remains static over short time intervals during scanning, and the projection data can thus be grouped into subsets, each covering a limited angular range (LAR) for the ROI. In contrast, the region outside the ROI is assumed to remain static throughout the entire scan and is covered by a full angular range (FAR). To reconstruct images from this hybrid data set, we develop a directional total variation-total variation (DTV-TV) algorithm, which applies DTV constraints within the ROI and conventional TV constraint outside it. The reconstruction is formulated as a convex optimization problem and solved using a primal-dual algorithm. We utilize the DTV-TV algorithm for image reconstruction from simulated data, demonstrating its ability to suppress artifacts and preserve anatomical structures within ROI. This approach provides insights for CBCT imaging in the presence of internal patient motion and has the potential to improve target localization accuracy in clinical radiation therapy.

Hypoxic resistance of living tissue to cytotoxin, most notably radiation, has been known for over a century but was believed in early trials to be a universal characteristic of all tumors. Oxygen electrode measurements of tumor pO2 demonstrated that many tumors had insufficient hypoxia to affect tumor resistance. Electron paramagnetic resonance oxygen images have been shown to accurately quantify and locate tumor hypoxia. Using novel radiation delivery for preclinical models with isocentric tumor boosts using 3D printed radiation apertures, we compared tumor control after equivalent integral dose boosts-either to hypoxic tumor or to well-oxygenated tumor-in preclinical models. For the first time, in three preclinical tumor models, we demonstrated a significant advantage for hypoxic boosts relative to well-oxygenated boosts.

Accurate urine collection is essential for preclinical studies involving metabolism, toxicology, and disease modeling. However, conventional methods such as metabolic caging and manual restraint are often invasive, labor intensive, prone to contamination, and not scalable for large cohorts, while also raising animal welfare concerns. This study presents a noninvasive and practical technique for collecting spot urine samples from healthy and tumor-bearing mice developed as a refinement to existing collection approaches. The method involves positioning the mouse in a natural voiding posture by allowing the hindlimbs to hang freely while the forelimbs grip a wire bar feeder, thereby stabilizing the animal and reducing excessive movement. Gentle abdominal pressure is applied only when needed. The method was validated in 80 C3H/HeJ and athymic nude mice, including both healthy and tumor-bearing (fibrosarcoma or prostate cancer [PC3]) models. Using this approach, spot urine volumes ranging from 13 to 200 µL were reliably collected from all mice, with each successful collection completed within 1 minute. Urine was successfully collected on the first attempt in ∼70% of cases. The mean urine yield was 51.7 ± 38.9 µL for healthy females, 53.7 ± 28.5 µL for healthy males, 50.8 ± 20.2 µL for tumor-bearing females, and 52.2 ± 28.1 µL for tumor-bearing males, with no statistically significant difference (P > 0.05) in urine volume distributions between groups. No visible fecal or particulate contamination was observed during sample collection. The collected volumes are sufficient for a wide range of molecular and metabolic analyses, ensuring the method's applicability in diverse research settings. Unlike other techniques that involve excessive handling, this method requires no restraint devices, making it well suited for large cohort studies and repeated sampling. The proposed technique represents a practical refinement in preclinical research by offering an optimal balance between sample quality, animal welfare, and experimental efficiency.

There is widespread consensus that hypoxia limits the effectiveness of cancer therapy. This has led to interventions to increase oxygen (O2) levels in tumors in patients, but success in clinical trials has been very limited and therefore clinical practice has not incorporated such interventions. The limiting step for successful intervention is the need to identify which tumors are hypoxic, whether they respond to interventions to increase O2, and the timing of the response. Consequently, many techniques have been advanced to measure O2 in tumors, but to date, none has been able to measure O2 directly in the tumor repeatedly under clinically applicable conditions (i.e., without perturbing clinical flow). Initial efforts at Dartmouth demonstrated that in vivo electron paramagnetic resonance (EPR) spectroscopy, using three types of injected or implanted O2 sensors, could provide the desired data under the desired conditions. Two types, injected paramagnetic India ink and an implanted coated derivative of lithium phthalocyanine, were successfully tested in clinical studies. However, their use is limited to tumors <1 cm of the surface. Consequently, Dartmouth developed a third O2 sensor, an “implantable resonator” (IR), to allow measuring in tumors at any depth; the IR has been successfully tested in preclinical studies. However, because the IR requires implanting at greater depth than the other types, its invasiveness was considered to be a drawback for clinical studies. Therefore, Clin-EPR and colleagues at the University of Chicago made additional technical improvements to the IR and proposed a new approach, called the multisite oxygen sensor (MOS), that allows its use in clinical studies without adding any invasiveness to therapy the patient is already undergoing. Specifically, the MOS is being designed to use in conjunction with a frequently used therapeutic approach (HDR brachytherapy delivered with an afterloader), applied initially to cervical cancer. HDR brachytherapy requires the invasive placement of multiple needles into the tumor and leaves them in situ for days during the course of treatment. Using these same needles, modified to be permeable to O2, would allow the MOS, inserted inside each needle, to simultaneously measure O2 at multiple locations throughout the tumor. This O2 measurement session could be repeated periodically during the course of therapy. We report preliminary technical studies of the modified MOS and the proposed modified brachytherapy needles, demonstrating in vitro the feasibility of our new approach to provide important information about tumor hypoxia during the course of radiation therapy without needing any additional invasiveness beyond standard of care therapy.

We present a Physics-guided deep learning framework to address common limitations in Confocal Laser Scanning Microscopy (CLSM), including diffraction-limited resolution, noise, and under sampling due to low laser power conditions. The optical system’s point spread function and primary CLSM image degradation mechanisms, namely photon shot noise, dark current noise, motion blur, speckle noise, and under sampling are explicitly incorporated into the model as physics-based constraints. A convolutional autoencoder is trained with a custom loss function that integrates these optical degradation processes, ensuring that the reconstructed images adhere to physical image formation principles. The model is evaluated on simulated CLSM datasets generated based on experimentally observed CLSM noise characteristics. Statistical comparisons, including intensity histograms, spatial frequency distributions, and structural similarity metrics, confirm that the synthetic dataset closely matches accurate CLSM data. The proposed approach is compared with traditional image reconstruction methods, including Richardson-Lucy deconvolution, non-negative least squares, and total variation regularization. Results indicate that the physics-constrained autoencoder improves structural detail recovery while maintaining consistency with known CLSM imaging physics. This study demonstrates that Physics-guided deep learning can provide an alternative computational approach to CLSM enhancement, complementing existing optical correction methods. Future work will focus on further validation using experimental CLSM acquisitions.

The intensity of the conditioning regimen in hematopoietic stem cell transplantation (HSCT) correlates with the risk of relapse, however its potential benefit may be outweighed by the associated risk of toxicity. The addition of total marrow irradiation (TMI) to myeloablative conditioning provides an opportunity to increase intensity with minimal additional toxicity. In this phase II clinical trial, 30 patients with high-risk myeloid malignancies underwent allogeneic HSCT using myeloablative TMI at 9 Gy in combination with standard myeloablative fludarabine/intravenous busulfan (FluBu4) chemotherapy. The study included patients with matched related donors (N=10) receiving TMI/FluBu4 and patients with matched unrelated (N=14) or one-antigen mismatched unrelated (N=6) donors receiving TMI/FluBu4 and rabbit anti-thymocyte globulin. All patients achieved sustained engraftment. Grade 3-4 extramedullary toxicities were mucositis in 59% (N=17), nausea/vomiting in 10% (N=3) and diarrhea in 7% (N=2) of the patients. Acute graft-versus-host disease (GvHD) grade 3 or 4 was seen in four patients (13.3%). Moderate/severe chronic GvHD was observed in 11 patients (36.7%). With a median follow-up of 1,483 days (range, 63-2,260 days) for patients alive, the overall survival and disease-free survival at 1 year were 72.4% and 65.5%, respectively. GvHD-free relapse-free survival at 1 year was 41.4%. Of 30 patients in the study, six relapsed/progressed (20%) and five of them died of the disease (16.7%), whereas six patients (20%) died of transplant-related causes. We conclude that a myeloablative regimen with TMI at 9 Gy and FluBu4 was well tolerated and achieved encouraging results in patients with myeloid malignancies at high risk of relapse (clinicaltrials.gov identifier: NCT03121014).

A recent parallel-opposed 3D-conformal radiotherapy (3D-CRT) study in mice compared dose escalation (boost) in hypoxic (pO2 ≤ 10 torr) and non-hypoxic tumor subvolumes. They found a hypoxic boost led to significantly greater (p < 1e-4) tumor control probability than an equivalent non-hypoxic boost. We imported imaging and treatment data from this study for 31 SCC7 squamous carcinoma murine leg tumor cases-16 hypoxic boost and 15 non-hypoxic boost plans into a commercial treatment planning system for preclinical radiotherapy. Treatments were retrospectively recalculated with a fast Monte Carlo dose engine. We replanned cases with 3-field IMRT using an analogous uncertainty budget as 3D-CRT. Comparing both treatment groups, the hypoxic boost treatments had a significantly higher hypoxic fraction receive the boost prescription as planned in 3D-CRT (p < 1e-4) and IMRT (p < 1e-4). Surprisingly, retrospective 3D-CRT non-hypoxic boost treatments had a significantly lower non-hypoxic fraction receive the boost prescription (p < 1e-4). 3D-CRT non-hypoxic boost also substantially underdosed the entire tumor between 48-68 Gy compared to the "equivalent" hypoxic boost. In IMRT, the non-hypoxic volume receiving boost prescription was significantly higher in the non-hypoxic boost (p = 0.0215) and dosing in the entire tumor was identical between boost groups. This study displays IMRT's potential to advance the quality of preclinical dose painting studies.

The emergence of foundational models represents a paradigm shift in medical imaging, offering extraordinary capabilities in disease detection, diagnosis, and treatment planning. These large-scale artificial intelligence systems, trained on extensive multimodal and multi-center datasets, demonstrate remarkable versatility across diverse medical applications. However, their integration into clinical practice presents complex ethical challenges that extend beyond technical performance metrics. This study examines the critical ethical considerations at the intersection of healthcare and artificial intelligence. Patient data privacy remains a fundamental concern, particularly given these models' requirement for extensive training data and their potential to inadvertently memorize sensitive information. Algorithmic bias poses a significant challenge in healthcare, as historical disparities in medical data collection may perpetuate or exacerbate existing healthcare inequities across demographic groups. The complexity of foundational models presents significant challenges regarding transparency and explainability in medical decision-making. We propose a comprehensive ethical framework that addresses these challenges while promoting responsible innovation. This framework emphasizes robust privacy safeguards, systematic bias detection and mitigation strategies, and mechanisms for maintaining meaningful human oversight. By establishing clear guidelines for development and deployment, we aim to harness the transformative potential of foundational models while preserving the fundamental principles of medical ethics and patient-centered care.

Gold nanoparticles (AuNPs) have been targeted as novel contrast agent for computerized tomography (CT). However, AuNPs suffer from low-contrast factor in the X-ray regime. Functionalization of AuNPs with folic acid or sugar-based molecules to induce selective uptake have displayed contrast enhancement with improved image brightness and CT signal intensity. However, it was not clear what the basic mechanism for the contrast enhancement was and whether it was related to the uptake enhancement or to a fundamental electromagnetic interaction effect. In this work, we conducted near-field Mie as well as finite-difference time-domain (FDTD) field distribution of the scattering to discern the effect of a thin dielectric coating layer on the contrast functionality of AuNPs. Our results show that upon the incorporation of the dielectric shell (thin film or nanoparticle layer), the cross section of X-ray scattering is enhanced, with silicon being more effective than silica coating, with multiresonance spectral response. The directionality and range and strength of the near field increase for silicon coating (high electron density or high k material in the visible). The effect may be understood in terms of several features. Even though the refractive indices of all materials in the X-ray regime are similar to 1.0, the wavelength dependence of their approach may exhibit sizeable differences The enhancement is understood in terms of high densities of polarization charge especially in silicon, which allows multipole resonances. The multiplicity of resonances leads to enhanced scattering and directionality (angular distribution) with reduced range. A silicon-coating layer on AuNP may not only alleviate the contrast limitation, but it may afford synergistic integration of luminescence and scattering functionalities in the visible and X-ray regimes.

Techniques for preclinical intensity modulated radiation therapy are being developed to improve translation by replicating the clinical paradigm. This study presents the first treatment planning comparison between small animal IMRT (SA-IMRT) and three-dimensional conformal radiotherapy (CRT) in a model application, oxygen-guided dose painting of tumor hypoxia, using actual mouse data. A novel compensator-based platform was employed to generate SA-IMRT and CRT plans with 2-15 beam angles for seventeen mice with fibrosarcoma tumors. The whole tumor received a dose of 22.5 Gy, with a simultaneous integrated boost of 13 Gy to hypoxic voxels identified via electron paramagnetic resonance imaging. Plan quality was assessed using the Paddick conformity index (CI), uniformity, and dose volume histograms. For 3-angles, SA-IMRT yielded significantly improved dose conformity (median hypoxic CI = 0.45 vs. 0.17), tumor dose uniformity (11.0% vs. 14.3%), and dosimetric spread between boost and non-boost targets (D50% difference = 13.0 Gy [ideal], 13.1 Gy [SA-IMRT], 7. 3 Gy [CRT]). No significant improvement in CI was associated with > 3 beam angles (Wilcoxon signed-rank test, p < 0.05). This study demonstrates that SA-IMRT provides significant improvements in radiation plan quality and yields dose distributions that more closely mimic the clinical setting relative to current CRT approaches.

Objectives: Based on a previous phase 1 study, total marrow irradiation (TMI) at 9Gy was added to a myeloablative FluBu4 conditioning regimen in allogeneic hematopoietic stem cell transplantation (HSCT) for myeloid malignancies. Here, we report on the long-term toxicity of TMI combined with FluBu4 and compare it to patients who received only FluBu4. Methods: We retrospectively analyzed 38 consecutive patients conditioned with FluBu4/TMI (n = 15) or FluBu4 (n = 23, control group) who had at least 1 year follow-up post-transplant. The rate of long-term adverse events that have been previously associated with total body irradiation (TBI) was analyzed in the two groups. Results: The baseline characteristics did not differ between the two groups. The control group had a longer median follow-up (71.2 mo) than the TMI group (38.5 mo) (p = .004). The most common adverse events were xerostomia, dental complications, cataracts, or osteopenia and did not differ between the two groups. Cognitive dysfunction or noninfectious pneumonitis, often detected after high dose TBI, were also not different in the two groups (p = .12 and p = .7, respectively). There was no grade 4 adverse event. Conclusion: Our results suggest that a conditioning regimen with TMI 9Gy and FluBu4 does not increase long-term adverse events after allogeneic HSCT.

Background and purposeWe proposed an artificial neural network model to predict radiobiological parameters for the head and neck squamous cell carcinoma patients treated with radiation therapy. The model uses the tumor specification, demographics, and radiation dose distribution to predict the tumor control probability and the normal tissue complications probability. These indices are crucial for the assessment and clinical management of cancer patients during treatment planning.MethodsTwo publicly available datasets of 31 and 215 head and neck squamous cell carcinoma patients treated with conformal radiation therapy were selected. The demographics, tumor specifications, and radiation therapy treatment parameters were extracted from the datasets used as inputs for the training of perceptron. Radiobiological indices are calculated by open-source software using dosevolume histograms from radiation therapy treatment plans. Those indices were used as output in the training of a single-layer neural network. The distribution of data used for training, validation, and testing purposes was 70, 15, and 15%, respectively.ResultsThe best performance of the neural network was noted at epoch number 32 with the mean squared error of 0.0465. The accuracy of the prediction of radiobiological indices by the artificial neural network in training, validation, and test phases were determined to be 0.89, 0.87, and 0.82, respectively. We also found that the percentage volume of parotid inside the planning target volume is the significant parameter for the prediction of normal tissue complications probability.ConclusionWe believe that the model has significant potential to predict radiobiological indices and help clinicians in treatment plan evaluation and treatment management of head and neck squamous cell carcinoma patients.

Purpose/Objective(s) A second allogeneic stem cell transplantation (allo-SCT) is an option following relapse after an initial SCT for hematologic malignancies. Conditioning with intensity-modulated total marrow irradiation (IM-TMI) is feasible and shows promise in optimizing the therapeutic ratio. We report on clinical outcomes and to identify IM-TMI dose to the oral cavity that would be associated with lower incidence of mucositis to help guide planning. Materials/Methods We conducted a retrospective analysis of patients undergoing second allo-SCT enrolled between Dec 2015 and Nov 2023 on a phase I dose-escalation trial of IM-TMI with fludarabine and melphalan. TMI doses were given twice daily, 1.5 Gy per fraction with total dose of 6, 9, or 12 Gy. The clinical target volume consisted of bones excluding mandible (except in 1 case), arms and lower extremities mid-femur down. We collected baseline patient and treatment characteristics and performed univariate analysis reported as mean (SD) or median [interquartile range]. Logistic regression (LR) was performed to evaluate potential predictors of oral mucositis incidence at 1 week following allo-SCT. Results Of 31 patients, 18 (58%) were male, predominantly white (83%), with median age of 49 [41, 64]. The mean BMI was 30.62 (22.33) and with a Karnofsky >90 (68%). Majority being treated for AML (74%). Most common disease status at time of 2nd allo-SCT included 1st relapse in 8 (26%) followed by 2nd complete remission in 5 (16.1%), and 3rd complete remission in 4 (13%). TMI dose of 9 Gy was delivered to 15 patients (48%), 12 Gy to 10 (32%), and 6 Gy to 5 (16%). The average dose to the oral cavity was 298.05 cGy (95.7), with the oral cavity receiving on average 32% of the prescribed dose. 16 patients (52%) experienced mucositis within a month following allo-SCT with IM-TMI; 14 occurred within 7 days, half grade 1 and grade 3 toxicities. Pearson correlation showed moderate association between oral cavity dose and mucositis grade at day 7 (r = 0.284). On LR analysis of predictors, the oral cavity mean dose (OR = 1.032) and age (OR = 0.878) were found to be significant predictors of mucositis incidence at day 7 (p = 0.04 and p = 0.03 respectively). In LR model with oral cavity mean dose and age categorized by the median value, an oral cavity mean dose greater than the median of 264.7 cGy was associated with an OR of 44.8 (p = 0.023). 5 of 31 patients (16%) had relapsed, 3 in the bone marrow and 2 with solid masses, at an average of 197 days (77.8) post allo-SCT. At a median follow up of 11 months, 11 patients (36%) were alive. Most common cause of death included persistence/progression of disease in 8 (40%) and infection in 9 patients (45%). Conclusion Our study adds to the evolving literature on the integration of TMI into conditioning regimens for allo-SCT, specifically in the second transplant setting. Given that the oral cavity mean dose and age are significant predictors of mucositis incidence at day 7, caution must be applied in the delivery of TMI to the oral cavity, specifically in this population.

In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs’ boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs.

[This corrects the article DOI: 10.3389/fmed.2023.1269689.].

Proton therapy is widely used for treating various tumor types due to its favorable dosimetric characteristics compared with conventional radiotherapy. However, a small error in dose calculation may lead to a substantial misadministration of planned radiation dose. This work aims to create a simplified and easy-to-use Monte Carlo (MC) simulation model, based on the multi-source model principle, for spot scanning proton radiotherapy. Our multi-source model contains a set of physical parameters acquired from a Varian ProBeam compact system at the South Florida Proton Therapy Institute (SFPTI). The source model input parameters are mean energy (s(E)) and beam spot (s(s)) standard deviations, which directly affect the integrated depth-dose (IDD) dosimetric characteristics such as beam width, proton range, and distal fall-off. Despite the simplicity of the presented model, all simulated results matched the corresponding experimental values within 2.5% and were found to be within the acceptable clinical limits, for a wide nominal proton energy range (i.e., 90-220 MeV). Additionally, the comparison of the simulated IDDs with the experimental IDDs was found to be highly conformal with 100% of the points passing the 2%/2 mm gamma index test. The model presented in this work can be efficiently used to model high energy (>80 MeV) in a central axis clinical scanning proton beam in a real clinical setting. The small number of input parameters selected allows for a more efficient and user-friendly MC modeling than previously developed models, while off-axis beam characteristics can be studied with the addition of a new sub-source parameter file.

Although radiotherapy is one of the most important curative treatments for cancer, its clinical application is associated with undesired therapeutic effects on normal or healthy tissues. The use of targeted agents that can simultaneously achieve therapeutic and imaging functions could constitute a potential solution. Herein, we developed 2-deoxy-d-glucose (2DG)-labeled poly(ethylene glycol) (PEG) gold nanodots (2DG-PEG-AuD) as a tumor-targeted computed tomography (CT) contrast agent and radiosensitizer. The key advantages of the design are its biocompatibility and targeted AuD with excellent sensitivity in tumor detection via avid glucose metabolism. As a consequence, CT imaging with enhanced sensitivity and remarkable radiotherapeutic efficacy could be attained. Our synthesized AuD displayed linear enhancement of CT contrast as a function of its concentration. In addition, 2DG-PEG-AuD successfully demonstrated significant augmentation of CT contrast in both in vitro cell studies and in vivo tumor-bearing mouse models. In tumor-bearing mice, 2DG-PEG-AuD showed excellent radiosensitizing functions after intravenous injection. Results from this work indicate that 2DG-PEG-AuD could greatly potentiate theranostic capabilities by providing high-resolution anatomical and functional images in a single CT scan and therapeutic capability.

EDITORIAL article Front. Oncol., 19 July 2023Sec. Radiation Oncology Volume 13 - 2023 | https://doi.org/10.3389/fonc.2023.1240530

TPS2080 Background: Stereotactic radiosurgery (SRS) is a commonly utilized treatment strategy for brain metastases which offers a relatively low rate of morbidity and high rate of local control. SRS delivers an ablative dose of radiation in a single session to a limited target volume while minimizing dose to surrounding normal tissue. Image-guided linear accelerator (LINAC)-based frameless SRS has enabled effective, safe delivery of high dose radiation without the discomfort and logistical complications associated with use of a rigid frame. Localization errors during frameless SRS vary depending on the specific LINAC configuration and immobilization technique. The clinical relevance of potential inaccuracies in localization with frameless SRS remains controversial. When employing frameless systems, many centers will account for set-up uncertainties by adding a circumferential margin of 1-3 mm around the GTV to create a planning target volume (PTV). This PTV expansion ensures adequate tumor coverage if target motion occurs, but exposes more normal brain tissue to radiation and may increase the risk of adverse events, chiefly radionecrosis. There is limited literature available to guide radiation oncologists on the appropriate choice of margin size, if any, for frameless SRS. There are no randomized data evaluating the safety and efficacy of omitting the margin entirely. Given the paucity of data examining margin extent and the potentially significant consequences for treatment efficacy and morbidity, we propose a phase II randomized prospective clinical trial evaluating 0 vs 2 mm marginal GTV to PTV expansions for treating patients with intact brain metastases with frameless SRS. Methods: Eligible patients will have brain metastases from solid tumors with ECOG performance status of 0-2 and life expectancy of >3 months. Patients must have 1 - 5 newly diagnosed well-circumscribed, measurable intraparenchymal brain metastases with maximum tumor diameter ≤3.0 cm. At least one lesion must be ≥ 0.5 cm in maximum diameter to be considered measurable. SRS will be delivered in a single fraction to either 20 Gy or 18 Gy dependent on tumor size. Importantly, PTV margin randomization is blinded and occurs after delineation of tumor volume by radiation oncologist to not bias contouring based on knowledge of margin randomization. This is two-armed multi-center phase II randomized controlled trial. The trial is powered to demonstrate non- inferiority of the experimental arm with regard to the primary endpoint of local PFS at 6 months and the superiority of the experimental arm with regard to the secondary endpoint of radionecrosis or pseudoprogression. We expect the total study duration, from the start of screening for the first participant until the end of follow-up for the last one, to be approximately six years. We plan to accrue 166 patients to this trial at all sites. Clinical trial information: NCT02747303 .