Image Guided Intraoperative Radiation Therapy After Surgical Resection of Brain Metastases: A First In-Human Feasibility Report.

Purpose: A correct placement of the applicator during intraoperative radiation therapy for brain metastasis is of paramount importance, to deliver a precise and safe treatment. The applicator-to-surface contact assessment cannot be performed under direct observation because the applicator itself limits the visual range. No image guided verification is currently performed intracranially. We hypothesize that image guided intraoperative radiation therapy would assure a more precise delivery in the target area. We describe our workflow in a first in-human experience. Methods and Materials: Phantom-based measurements were performed to reach the best cone beam computed tomography imaging quality possible. Once defined, a clinical feasibility study was initiated. An in-room cone beam computed tomography device is used to acquire intraoperative images after placing the applicator. Repositioning the applicator is thereafter discussed with the surgeon, according to the imaging outcomes, if required. Results: An optimal image quality was achieved with 120-kV voltage, 20-mA current, and a tube current time product of 150 mAs. An additional 0.51 mSv patient exposure was calculated for the entire procedure. The wide dynamic range (-600 HU to +600 HU) of cone beam computed tomography and a 27 HU mean computed tomography values difference between brain tissue and spherical applicator allows distinguishing both structures. In this first in-human experience, the applicator was repositioned after evidencing air gaps, assuring full applicator-to-surface contact. Conclusions: This first in-human procedure confirmed the feasibility of kilovoltage image guided intraoperative radiation therapy in a neurosurgical setting. A prospective study has been initiated and will provide further dosimetric details.

Artificial Intelligence-Based Autosegmentation: Advantages in Delineation, Absorbed Dose-Distribution, and Logistics.

Purpose: The study's purpose was to compare the performance of artificial intelligence (AI) in auto-contouring compared with a human practitioner in terms of precision, differences in dose distribution, and time consumption. Methods and Materials: Datasets of previously irradiated patients in 3 different segments (head and neck, breast, and prostate cancer) were retrospectively collected. An experienced radiation oncologist (MD) performed organs-at-risk (OARs) and standard clinical target volume delineations as baseline structures for comparison. AI-based autocontours were generated in 2 additional CT copies; therefore, 3 groups were assessed: MD alone, AI alone, and AI plus MD corrections (AI+C). Differences in Dice similarity coefficient (DSC) and person-hour burden were assessed. Furthermore, changes in clinically relevant dose-volume parameters were evaluated and compared. Results: Seventy-five previously treated cases were collected (25 per segment) for the analysis. Compared with MD contours, the mean DSC scores were higher than 0.7 for 74% and 80% of AI and AI+C, respectively. After corrections, 17.1% structures presented DSC score deviations higher than 0.1 and 10.4% dose-volume parameters significantly changed in AI-contoured structures. The time consumption assessment yielded mean person-hour reductions of 68%, 51%, and 71% for breast, prostate, and head and neck cancer, respectively. Conclusions: In great extent, AI yielded clinically acceptable OARs and certain clinical target volumes in the explored anatomic segments. Sparse correction and assessment requirements place AI+C as a standard workflow. Minimal clinically relevant differences in OAR exposure were identified. A substantial amount of person-hours could be repurposed with this technology.

Intraoperative or postoperative stereotactic radiotherapy for brain metastases: time to systemic treatment onset and other patient-relevant outcomes.

PURPOSE: Intraoperative radiotherapy (IORT) has become a viable treatment option for resectable brain metastases (BMs). As data on local control and radiation necrosis rates are maturing, we focus on meaningful secondary endpoints such as time to next treatment (TTNT), duration of postoperative corticosteroid treatment, and in-hospital time. METHODS: Patients prospectively recruited within an IORT study registry between November 2020 and June 2023 were compared with consecutive patients receiving adjuvant stereotactic radiotherapy (SRT) of the resection cavity within the same time frame. TTNT was defined as the number of days between BM resection and start of the next extracranial oncological therapy (systemic treatment, surgery, or radiotherapy) for each of the groups. RESULTS: Of 95 BM patients screened, IORT was feasible in 84 cases (88%) and ultimately performed in 64 (67%). The control collective consisted of 53 SRT patients. There were no relevant differences in clinical baseline features. Mean TTNT (range) was 36 (9 - 94) days for IORT patients versus 52 (11 - 126) days for SRT patients (p = 0.01). Mean duration of postoperative corticosteroid treatment was similar (8 days; p = 0.83), as was mean postoperative in-hospital time (11 versus 12 days; p = 0.97). Mean total in-hospital time for BM treatment (in- and out-patient days) was 11 days for IORT versus 19 days for SRT patients (p < 0.001). CONCLUSION: IORT for BMs results in faster completion of interdisciplinary treatment when compared to adjuvant SRT, without increasing corticosteroid intake or prolonging in-hospital times. A randomised phase III trial will determine the clinical effects of shorter TTNT.

Outcome assessment of intraoperative radiotherapy for brain metastases: results of a prospective observational study with comparative matched-pair analysis.

PURPOSE: Intraoperative radiation therapy (IORT) is an emerging alternative to adjuvant stereotactic external beam radiation therapy (EBRT) following resection of brain metastases (BM). Advantages of IORT include an instant prevention of tumor regrowth, optimized dose-sparing of adjacent healthy brain tissue and immediate completion of BM treatment, allowing an earlier admission to subsequent systemic treatments. However, prospective outcome data are limited. We sought to assess long-term outcome of IORT in comparison to EBRT. METHODS: A total of 35 consecutive patients, prospectively recruited within a study registry, who received IORT following BM resection at a single neuro-oncological center were evaluated for radiation necrosis (RN) incidence rates, local control rates (LCR), distant brain progression (DBP) and overall survival (OS) as long-term outcome parameters. The 1 year-estimated OS and survival rates were compared in a balanced comparative matched-pair analysis to those of our institutional database, encompassing 388 consecutive patients who underwent adjuvant EBRT after BM resection. RESULTS: The median IORT dose was 30 Gy prescribed to the applicator surface. A 2.9% RN rate was observed. The estimated 1 year-LCR was 97.1% and the 1 year-DBP-free survival 73.5%. Median time to DBP was 6.4 (range 1.7-24) months in the subgroup of patients experiencing intracerebral progression. The median OS was 17.5 (0.5-not reached) months with a 1 year-survival rate of 61.3%, which did not not significantly differ from the comparative cohort (p = 0.55 and p = 0.82, respectively). CONCLUSION: IORT is a safe and effective fast-track approach following BM resection, with comparable long-term outcomes as adjuvant EBRT.

Comprehensive investigation of lateral dose profile and output factor measurements in small proton fields from different delivery techniques.

BACKGROUND AND PURPOSE: As a part of the commissioning and quality assurance in proton beam therapy, lateral dose profiles and output factors have to be acquired. Such measurements can be performed with point detectors and are especially challenging in small fields or steep lateral penumbra regions as the detector's volume effect may lead to perturbations. To address this issue, this work aims to quantify and correct for such perturbations of six point detectors in small proton fields created via three different delivery techniques. METHODS: Lateral dose profile and output measurements of three proton beam delivery techniques (pencil beam scanning, pencil beam scanning combined with collimators, passive scattering with collimators) were performed using high-resolution EBT3 films, a PinPoint 3D 31022 ionization chamber, a microSilicon diode 60023 and a microDiamond detector 60019 (all PTW Freiburg, Germany). Detector specific lateral dose response functions K(x,y) acting as the convolution kernel transforming the undisturbed dose distribution D(x,y) into the measured signal profiles M(x,y) were applied to quantify perturbations of the six investigated detectors in the proton fields and correct the measurements. A signal theoretical analysis in Fourier space of the dose distributions and detector's K(x,y) was performed to aid the understanding of the measurement process with regard to the combination of detector choice and delivery technique. RESULTS: Quantification of the lateral penumbra broadening and signal reduction at the fields center revealed that measurements in the pencil beam scanning fields are only compromised slightly even by large volume ionization chambers with maximum differences in the lateral penumbra of 0.25 mm and 4% signal reduction at the field center. In contrast, radiation techniques with collimation are not accurately represented by the investigated detectors as indicated by a penumbra broadening up to 1.6 mm for passive scattering with collimators and 2.2 mm for pencil beam scanning with collimators. For a 3 mm diameter collimator field, a signal reduction at field center between 7.6% and 60.7% was asserted. Lateral dose profile measurements have been corrected via deconvolution with the corresponding K(x,y) to obtain the undisturbed D(x,y). Corrected output ratios of the passively scattered collimated fields obtained for the microDiamond, microSilicon and PinPoint 3D show agreement better than 0.9% (one standard deviation) for the smallest field size of 3 mm. CONCLUSION: Point detector perturbations in small proton fields created with three delivery techniques were quantified and found to be especially pronounced for collimated small proton fields with steep dose gradients. Among all investigated detectors, the microSilicon diode showed the smallest perturbations. The correction strategies based on detector's K(x,y) were found suitable for obtaining unperturbed lateral dose profiles and output factors. Approximation of K(x,y) by considering only the geometrical averaging effect has been shown to provide reasonable prediction of the detector's volume effect. The findings of this work may be used to guide the choice of point detectors in various proton fields and to contribute toward the development of a code of practice for small field proton dosimetry.

Validating a double Gaussian source model for small proton fields in a commercial Monte-Carlo dose calculation engine.

PURPOSE: The primary fluence of a proton pencil beam exiting the accelerator is enveloped by a region of secondaries, commonly called "spray". Although small in magnitude, this spray may affect dose distributions in pencil beam scanning mode e.g., in the calculation of the small field output, if not modelled properly in a treatment planning system (TPS). The purpose of this study was to dosimetrically benchmark the Monte Carlo (MC) dose engine of the RayStation TPS (v.10A) in small proton fields and systematically compare single Gaussian (SG) and double Gaussian (DG) modeling of initial proton fluence providing a more accurate representation of the nozzle spray. METHODS: The initial proton fluence distribution for SG/DG beam modeling was deduced from two-dimensional measurements in air with a scintillation screen with electronic readout. The DG model was either based on direct fits of the two Gaussians to the measured profiles, or by an iterative optimization procedure, which uses the measured profiles to mimic in-air scan-field factor (SF) measurements. To validate the DG beam models SFs, i.e. relative doses to a 10 × 10 cm2 field, were measured in water for three different initial proton energies (100MeV, 160MeV, 226.7MeV) and square field sizes from 1×1cm2 to 10×10cm2 using a small field ionization chamber (IBA CC01) and an IBA ProteusPlus system (universal nozzle). Furthermore, the dose to the center of spherical target volumes (diameters: 1cm to 10cm) was determined using the same small volume ionization chamber (IC). A comprehensive uncertainty analysis was performed, including estimates of influence factors typical for small field dosimetry deduced from a simple two-dimensional analytical model of the relative fluence distribution. Measurements were compared to the predictions of the RayStation TPS. RESULTS: SFs deviated by more than 2% from TPS predictions in all fields <4×4cm2 with a maximum deviation of 5.8% for SG modeling. In contrast, deviations were smaller than 2% for all field-sizes and proton energies when using the directly fitted DG model. The optimized DG model performed similarly except for slightly larger deviations in the 1×1cm2 scan-fields. The uncertainty estimates showed a significant impact of pencil beam size variations (±5%) resulting in up to 5.0% standard uncertainty. The point doses within spherical irradiation volumes deviated from calculations by up to 3.3% for the SG model and 2.0% for the DG model. CONCLUSION: Properly representing nozzle spray in RayStation's MC-based dose engine using a DG beam model was found to reduce the deviation to measurements in small spherical targets to below 2%. A thorough uncertainty analysis shows a similar magnitude for the combined standard uncertainty of such measurements.