Physica Medica-European Journal of Medical Physics最新文献

{"title":"Radiobiological effect of delivery time in treatments of multiple brain metastases with robotic radiosurgery","authors":"David Sevillano ,&nbsp;Petar Ivaylov ,&nbsp;Eva Fernández-Lizarbe ,&nbsp;Juan David García ,&nbsp;Mercedes Martín ,&nbsp;Rafael Morís ,&nbsp;Belén Capuz ,&nbsp;Rafael Colmenares ,&nbsp;Miguel Cámara ,&nbsp;Rubén Chillida ,&nbsp;Carlos Rodríguez-Manzaneque ,&nbsp;Juan Trujillo ,&nbsp;María José Béjar ,&nbsp;Daniel Prieto ,&nbsp;Feliciano García-Vicente","doi":"10.1016/j.ejmp.2025.105087","DOIUrl":"10.1016/j.ejmp.2025.105087","url":null,"abstract":"<div><h3>Purpose</h3><div>To analyze the dose effect of the delivery time in CyberKnife Stereotactic Radiosurgery treatments. A prediction tool and methods to minimize this effect are presented.</div></div><div><h3>Methods</h3><div>Eighty CyberKnife plans of multiple brain metastases (BMs) were analyzed. The Microdosimetric Kinetic Model was used to assess the biological effectiveness of each treatment at each BM. Results were obtained in terms of relative dose (RD) between delivered and planned treatments.</div><div>A prediction algorithm of RD was developed by predicting beam delivery times of not yet delivered plans. Two strategies to optimize treatments in terms of biological effectiveness were proposed: beam reordering and treatment splitting.</div></div><div><h3>Results</h3><div>An average RD of 92.3 % (SD = 2.6 %) was obtained for our patient population. Differences between treated and predicted RD were −0.12 % (SD = 0.6 %). Beam reordering improved RD by a mean value of 5.5 % (SD = 3.3 %) with 17.3 % mean increase in overall treatment time.</div><div>Splitted treatments yielded a mean increase of 5.8 % in RD with a time increase of 35 %, while healthy brain was equally spared if treatments were delivered in different days. For those five treatments, an 8.0 % increase in RD and a 20 % increase in treatment time were obtained with reordered beams.</div></div><div><h3>Conclusions</h3><div>The effect of treatment time on the dose delivered to BMs in SRS with CyberKnife should be taken into consideration, especially when a large number of BMs are involved. This effect can be predicted and minimized by reordering beams at the cost of increasing overall treatment time.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105087"},"PeriodicalIF":2.7,"publicationDate":"2025-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144887188","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Temporal footprint reduction via neural network denoising in 177Lu radioligand therapy","authors":"Max Celio Nzatsi ,&nbsp;Nicolas Varmenot ,&nbsp;David Sarrut ,&nbsp;Grégory Delpon ,&nbsp;Michel Cherel ,&nbsp;Caroline Rousseau ,&nbsp;Ludovic Ferrer","doi":"10.1016/j.ejmp.2025.105071","DOIUrl":"10.1016/j.ejmp.2025.105071","url":null,"abstract":"<div><h3>Background</h3><div>Internal vectorised therapies, particularly with [177Lu]-labelled agents, are increasingly used for metastatic prostate cancer and neuroendocrine tumours. However, routine dosimetry for organs-at-risk and tumours remains limited due to the complexity and time requirements of current protocols.</div></div><div><h3>Method</h3><div>We developed a Generative Adversarial Network (GAN) to transform rapid 6 s SPECT projections into synthetic 30 s-equivalent projections. SPECT data from twenty patients and phantom acquisitions were collected at multiple time-points.</div></div><div><h3>Results</h3><div>The GAN accurately predicted 30 s projections, enabling estimation of time-integrated activities in kidneys and liver with maximum errors below 6 % and 1 %, respectively, compared to standard acquisitions. For tumours and phantom spheres, results were more variable. On phantom data, GAN-inferred reconstructions showed lower biases for spheres of 20, 8, and 1 mL (8.2 %, 6.9 %, and 21.7 %) compared to direct 6 s acquisitions (12.4 %, 20.4 %, and 24.0 %). However, in patient lesions, 37 segmented tumours showed higher median discrepancies in cumulated activity for the GAN (15.4 %) than for the 6 s approach (4.1 %).</div></div><div><h3>Conclusion</h3><div>Our preliminary results indicate that the GAN can provide reliable dosimetry for organs-at-risk, but further optimisation is needed for small lesion quantification. This approach could reduce SPECT acquisition time from 45 to 9 min for standard three-bed studies, potentially facilitating wider adoption of dosimetry in nuclear medicine and addressing challenges related to toxicity and cumulative absorbed doses in personalised radiopharmaceutical therapy.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105071"},"PeriodicalIF":2.7,"publicationDate":"2025-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867021","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and simulation of a SPECT real time dose monitoring system for BNCT: response at the LENA reactor","authors":"L. Santini ,&nbsp;T. Ferri ,&nbsp;L. Grisoni ,&nbsp;A. Caracciolo ,&nbsp;D. Bortot ,&nbsp;G. Borghi ,&nbsp;A. Pola ,&nbsp;S. Agosteo ,&nbsp;V. Pascali ,&nbsp;N. Protti ,&nbsp;D.Ramos Lòpez ,&nbsp;N. Ferrara ,&nbsp;D. Mazzucconi","doi":"10.1016/j.ejmp.2025.105070","DOIUrl":"10.1016/j.ejmp.2025.105070","url":null,"abstract":"<div><h3>Background</h3><div>Boron Neutron Capture Therapy (BNCT) selectively targets tumor cells while sparing healthy ones, by exploiting neutron capture on boron-10, which accumulates to the cancerous cells. To ensure that the therapy is properly tuned, real-time dose monitoring during treatment plays a fundamental role. A Single Photon Emission Computed Tomography (SPECT) imaging system relying on the 478 keV gamma ray emitted by the neutron capture reaction, can, in principle, detect the boron distribution and allow the 3D reconstruction of the dose inside the patient. However, neutron interactions with all the other elements present in tissues and structures introduce background signals, complicating dose evaluation.</div></div><div><h3>Methods</h3><div>In this study, FLUKA Monte Carlo simulations were applied to a BNCT-SPECT oriented detector to simulate the image reconstruction process. The simulations were conducted by irradiating the system at the LENA (Laboratorio Energia Nucleare Applicata) in Pavia and compared with experimental data. Moreover, a proof-of-concept study on a SPECT acquisition have been performed on different borated samples.</div></div><div><h3>Results</h3><div>The experimental and simulated results are in good agreement for both image acquisition and detected counting rates. The simulated projections, reconstructed with an appropriated iterative algorithm, show that the presented system is capable of distinguishing two separated vials containing boron-10.</div></div><div><h3>Conclusion</h3><div>This study show that the presented system holds a good promise for enhancing the precision of dose monitoring and localization during clinical BNCT treatments. For this reason, the system will be deployed in real BNCT facilities to evaluate and validate its performance under clinical conditions.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105070"},"PeriodicalIF":2.7,"publicationDate":"2025-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852409","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Using HistoQC to predict disagreement on human epidermal growth factor receptor 2 (HER2) assessment","authors":"Laura Verzellesi ,&nbsp;Moira Ragazzi ,&nbsp;Andrea Botti ,&nbsp;Giacomo Santandrea ,&nbsp;Andrew Janowczyk ,&nbsp;Luca Bottazzi ,&nbsp;Alessandra Bisagni ,&nbsp;Ione Tamagnini ,&nbsp;Giorgio Gardini ,&nbsp;Saverio Coiro ,&nbsp;Elisa Gasparini ,&nbsp;Mauro Iori","doi":"10.1016/j.ejmp.2025.105066","DOIUrl":"10.1016/j.ejmp.2025.105066","url":null,"abstract":"<div><h3>Background</h3><div>The human epidermal growth factor receptor 2 (HER2) gene is a significant prognostic and predictive factor for breast cancer therapy response. HER2 assessment is critical for targeted therapy eligibility, but interobserver reproducibility is a well-known issue in HER2 evaluation.</div></div><div><h3>Purpose</h3><div>The goal of our study is to create a machine learning (ML) system able to detect whole slide images (WSIs) that might cause discrepancies among observers.</div></div><div><h3>Methods</h3><div>We collected 132 pathology slides with double-blind HER2 evaluation and defined the agreement between observers as a binary classification: 0 for disagreement and 1 for agreement. We utilized HistoQC software to analyze and characterize the pathology slides based on a series of quality-related features. HistoQC-derived quality metrics were used to train a machine learning model (XGBoost) to predict interobserver disagreement. The dataset was randomly split into training and testing at proportions of 60%/40%, respectively.</div></div><div><h3>Results</h3><div>Our model demonstrated a mean AUC of 0.86 (standard deviation, SD = 0.09) across five cross-validation runs on the training set, highlighting its predictive reliability. The AUC on the testing set was 0.81 (confidence interval, CI = [0.82–0.94]), emphasizing the model’s precision in predicting whether an unseen WSI would lead to discordance.</div></div><div><h3>Conclusions</h3><div>Our study presents a machine learning model built to identify potential diagnostic disagreements in HER2 pathology evaluations. The results demonstrate a correlation between the quality of pathology slides and diagnostic outcomes. Upon proper validation, our tool could be integrated among the existing quality assurance systems used in anatomic pathology departments to improve HER2 diagnostic process.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105066"},"PeriodicalIF":2.7,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144828690","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Radiation protection in proton therapy: Insights from a multi-institutional survey and experimental measurements within the SINFONIA project","authors":"Marijke De Saint-Hubert ,&nbsp;Maite Romero-Expósito ,&nbsp;Malgorzata Liszka ,&nbsp;Olivier Van Hoey ,&nbsp;Sita Pinasti ,&nbsp;Linda Eliasson ,&nbsp;Jan Lillhok ,&nbsp;Carles Domingo ,&nbsp;Lara Struelens ,&nbsp;Alexandru Dasu","doi":"10.1016/j.ejmp.2025.105049","DOIUrl":"10.1016/j.ejmp.2025.105049","url":null,"abstract":"<div><h3>Purpose:</h3><div>This study, part of the EU-funded SINFONIA project, addresses radiation protection concerns in daily PT practice to study potentially under- or overprotection of staff depending on their working environment and tasks.</div></div><div><h3>Methods:</h3><div>A multi-institutional survey was carried out, complemented by specific neutron dose measurements at a PT centre. Besides mono-energetic irradiations, clinical treatments for brain, thorax (Hodgkin’s lymphoma — HL) and pelvis (prostate), were investigated with and without range shifter (RS). The Thermo Fisher Scientific WENDI II and Berthold LB6411 ambient neutron monitors and two DIAMON neutron spectrometers were used for H*(10). Furthermore, BTI BD-PND bubble detectors, LANDAUER Neutrak dosimeters and Intercast CR-39 + plastic converters were also tested for Hp(10).</div></div><div><h3>Results:</h3><div>Initial survey results suggest no considerable risk for staff, with doses registering below 1 mSv/year. Further, site-specific measurements conducted at the Skandion facility, unveil annual staff doses ranging from 5–10 <span><math><mi>μ</mi></math></span>Sv in the staff control room. Accidental exposure scenarios on the other hand could reach up to 0.26 <span><math><mi>μ</mi></math></span>Sv/Gy in the gantry pit. Notably, H*(10) detectors demonstrated good performance, while H_p(10) dosimeters solely captured doses within the treatment room, yet remaining in agreement with reference values.</div></div><div><h3>Conclusion:</h3><div>Findings suggest a minimal neutron exposure risk for staff members within contemporary PT facilities.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105049"},"PeriodicalIF":2.7,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144828689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bragg peak position monitoring using silicon and titanium nanoparticles as prompt-gamma tracers","authors":"Bartosz Klębowski ,&nbsp;Barbara Kołodziej ,&nbsp;Barbara Beus ,&nbsp;Magdalena Garbacz ,&nbsp;Ronja Hetzel ,&nbsp;Jonas Kasper ,&nbsp;Aleksandra Kaszlikowska ,&nbsp;Andrzej Magiera ,&nbsp;Aleksandra Wrońska","doi":"10.1016/j.ejmp.2025.105068","DOIUrl":"10.1016/j.ejmp.2025.105068","url":null,"abstract":"<div><h3>Purpose</h3><div>Spherical silicon and rods-like titanium oxide nanoparticles (NPs) have been analyzed for use in the proton range verification method in proton therapy (main goal), as well as radiosensitizers (second goal) in this therapy due to their physical and biological properties. The method involved the use of tracers emitting prompt-gamma radiation during irradiation with protons. The basic assumption of the method is to selectively deliver the tracer in form of NPs to the tumor. The cytotoxicity of the obtained nanomaterials was also checked against normal and cancer cells.</div></div><div><h3>Methods and Materials</h3><div>Correlation between the Bragg peak (BP) position in the PMMA phantom and the signal emitted by the analyzed tracers were determined on the basis of simulations carried out using the Geant4 toolkit. To determine the cytotoxicity of nanosilicone and nanotitanium, as well as their radiosensitizing properties a classic MTS test and a modified multiple MTS test were performed. The location of both types of NPs was determined using holotomographic microscopy.</div></div><div><h3>Results</h3><div>For silicon NPs, a signal was observed when the BP was located entirely in the structure imitating a tumor and decreased when the BP was entirely outside the structure. In the case of titanium NPs, the signal did not correlate with the position of the structure mimicking a tumor. Both types of NPs at low concentrations turned out to be non-toxic to both cell lines. It has been shown that both types of nanoparticles have promising radiosensitizing properties, in particular towards cancer cells.</div></div><div><h3>Conclusions</h3><div>When it comes to physical properties, silicon appears to be an optimal candidate for use in proton therapy monitoring. Moreover, the silica NPs turned out to be slightly more effective radiosensitizers than titanium NPs.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105068"},"PeriodicalIF":2.7,"publicationDate":"2025-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144810195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Weakly supervised commissioning of externally developed auto-segmentation models and applied to male pelvis MR auto-segmentation","authors":"Bastiaan W.K. Schipaanboord ,&nbsp;Peter J. Koopmans ,&nbsp;Erik van der Bijl ,&nbsp;Charlotte L. Brouwer ,&nbsp;Tomas Janssen","doi":"10.1016/j.ejmp.2025.105057","DOIUrl":"10.1016/j.ejmp.2025.105057","url":null,"abstract":"<div><h3>Introduction:</h3><div>When introducing an auto-segmentation model into clinical practice, assessing the quality of the predicted segmentations and the robustness of the model over a wide range of anatomical variation and/or image quality is difficult. Especially, when the model is provided by an external party and the institution introducing the model does not possess a high-quality dataset to commission the model on.</div></div><div><h3>Materials &amp; Methods:</h3><div>Assuming that a model is more likely to fail for an atypical case as opposed to a more average one, we propose a methodology that selects cases for commissioning using unsupervised anomaly detection. For this, the model supplier provides a set of image/shape features that correlate with model performance on the training data. Next, the receiving hospital can use these features to train an unsupervised anomaly detector on a large dataset of unlabeled cases and use the anomaly scores to select representative cases for commissioning of the model. Since the anomaly detector is trained on unlabeled data, a large, high-quality, curated dataset is not required on the receiving hospital side.</div></div><div><h3>Results:</h3><div>Using the proposed approach, the likelihood of selecting atypical edge cases with low segmentation performance was increased, as compared to random selection. For a selection of 20 cases, an increase of 22% was observed.</div></div><div><h3>Conclusions:</h3><div>The increased performance spread provides a more representative range of expected performance in clinical practice. This approach could be used for model commissioning to increase the confidence that the model performs well over a wide range of expected anatomical variation.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"136 ","pages":"Article 105057"},"PeriodicalIF":2.7,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144769415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Validating an explainable radiomics approach in non-small cell lung cancer combining high energy physics with clinical and biological analyses","authors":"Mariagrazia Monteleone ,&nbsp;Francesca Camagni ,&nbsp;Stefano Percio ,&nbsp;Letizia Morelli ,&nbsp;Guido Baroni ,&nbsp;Simone Gennai ,&nbsp;Pietro Govoni ,&nbsp;Chiara Paganelli","doi":"10.1016/j.ejmp.2025.105054","DOIUrl":"10.1016/j.ejmp.2025.105054","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aims at establishing a validation framework for an explainable radiomics-based model, specifically targeting classification of histopathological subtypes in non-small cell lung cancer (NSCLC) patients.</div></div><div><h3>Methods</h3><div>We developed an explainable radiomics pipeline using open-access CT images from the cancer imaging archive (TCIA). Our approach incorporates three key prongs: SHAP-based feature selection for explainability within the radiomics pipeline, a technical validation of the explainable technique using high energy physics (HEP) data, and a biological validation using RNA-sequencing data and clinical observations.</div></div><div><h3>Results</h3><div>Our radiomic model achieved an accuracy of 0.84 in the classification of the histological subtype. The technical validation performed on the HEP domain over 150 numerically equivalent datasets, maintaining consistent sample size and class imbalance, confirmed the reliability of SHAP-based input features. Biological analysis found significant correlations between gene expression and CT-based radiomic features. In particular, gene <em>MUC21</em> achieved the highest correlation with the radiomic feature describing the10th percentile of voxel intensities (r = 0.46, p &lt; 0.05).</div></div><div><h3>Conclusion</h3><div>This study presents a validation framework for explainable CT-based radiomics in lung cancer, combining HEP-driven technical validation with biological validation to enhance interpretability, reliability, and clinical relevance of XAI models.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"136 ","pages":"Article 105054"},"PeriodicalIF":2.7,"publicationDate":"2025-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144724454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Calibration services for X-ray multimeters in Europe: current situation and future needs","authors":"Ivana Komatina ,&nbsp;Vedrana Makarić ,&nbsp;Miloš Živanović ,&nbsp;Paula Toroi ,&nbsp;Aino Tietäväinen ,&nbsp;Massimo Pinto ,&nbsp;Luigi Rinaldi ,&nbsp;Markus Borowski ,&nbsp;Siarhei Saroka ,&nbsp;Nikola Kržanović ,&nbsp;Alessia Ciccotelli ,&nbsp;Bartel Jansen ,&nbsp;Stefan Pojtinger ,&nbsp;Ana Fernandes ,&nbsp;Vittorio Cannatà ,&nbsp;Mika Kortesniemi","doi":"10.1016/j.ejmp.2025.105055","DOIUrl":"10.1016/j.ejmp.2025.105055","url":null,"abstract":"<div><h3>Introduction</h3><div>Optimization and quality control of the diagnostic and interventional radiology procedures is usually performed with an X-ray multimeter (XMM) based on the non-invasive measurements of different X-ray tube and output parameters obtained from the X-ray beam, such as air kerma, tube voltage and half-value layer. Standardization and metrological support need to be improved, and harmonized calibration procedures are not available for all quantities. There is also a lack of data on performance of XMMs in different measurement conditions relevant for clinical practice.</div></div><div><h3>Methods</h3><div>The needs for calibration of XMMs and current state of the art of calibration services were investigated by performing an overview of the standards, conducting surveys addressed to the clinical medical physicists and calibration laboratories and investigating the key comparison database.</div></div><div><h3>Results</h3><div>There are widely available calibration services for air kerma measurements for a large range of radiation qualities. However, there is a lack of calibration services for all other measured quantities, and very few laboratories besides the manufacturers are able to perform these calibrations. In addition, standardization gaps with non-harmonized calibration and measurement procedures for these quantities were found.</div></div><div><h3>Conclusion</h3><div>New calibration services with harmonized procedures are needed for XMMs, especially for quantities beyond air kerma. There is a need to better understand and reduce measurement uncertainty for some quantities. New procedures will be developed within the TraMeXI project and disseminated to the standardization bodies, metrology and medical physics community.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"136 ","pages":"Article 105055"},"PeriodicalIF":2.7,"publicationDate":"2025-07-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144723463","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Physics-informed machine learning for predicting MLC and gantry errors in VMAT: a feature engineering approach","authors":"Perumal Murugan,&nbsp;Ravikumar Manickam","doi":"10.1016/j.ejmp.2025.105064","DOIUrl":"10.1016/j.ejmp.2025.105064","url":null,"abstract":"<div><h3>Background</h3><div>This study presents a physics-informed, feature-engineered machine learning (ML) framework to predict multileaf collimator (MLC) and gantry positional errors in volumetric modulated arc therapy (VMAT)</div></div><div><h3>Methods</h3><div>Data from 32 VMAT trajectory logs (TrueBeam linac, HD120 MLC) were synchronized with DICOMRT plans to extract delivery dynamics. Novel physics-based parameters were introduced: a friction factor, an enhanced gravity vector, and MLC speed-normalized features. Three ML models XGBoost, LightGBM, and deep neural networks (DNNs) were optimized using Optuna and trained on trajectory log and DICOM-RT-derived datasets. Feature importance was evaluated via Spearman correlation, mutual information, and SHapley Additive Explanations (SHAP).</div></div><div><h3>Results</h3><div>Systematic discrepancies between DICOM-RT and trajectory log data were identified, with mean absolute deviations of 7.0 % (MLC speed), 8.0 % (gantry speed), and 8.5 % (dose rate). MLC speed emerged as the dominant predictor (Spearman: r_s = 0.891), while friction and gravity features exhibited significant correlations (r_s = 0.46 and 0.33, respectively). Mutual information revealed non-monotonic dependencies between gantry error and gantry angle (score: 0.34). LightGBM and XGBoost achieved superior MLC error prediction (MAE: 0.0019 mm, RMSE: 0.0027 mm), capturing &gt; 90 % of observed errors, while DNNs lagged by 30 %. Engineered features reduced residual errors by 30 %. Gantry error predictions showed lower accuracy (MAE: 0.012°–0.015°). SHAP analysis highlighted physics-driven features as top contributors.</div></div><div><h3>Conclusion</h3><div>This work underscores the critical role of domain knowledge in ML for radiotherapy, achieving a 30% error reduction through physics-based feature engineering. The findings advocate for prioritizing feature space exploration alongside model optimization to enhance VMAT quality assurance.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"136 ","pages":"Article 105064"},"PeriodicalIF":2.7,"publicationDate":"2025-07-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144724682","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}