Medical physics最新文献

{"title":"Experimental investigation of metal artifact reduction using one-step spectral CT reconstruction","authors":"Benjamin M. Rizzo, Emil Y. Sidky, Taly Gilat Schmidt","doi":"10.1002/mp.70672","DOIUrl":"10.1002/mp.70672","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>The constrained one-step spectral CT image reconstruction algorithm (cOSSCIR) is a model-based approach that iteratively estimates basis material maps directly from a set of transmission measurements. By incorporating a model of the imaging system's x-ray beam and detector response, the method can be applied to both dual-kV and photon-counting problems. Since the polyenergetic composition of the x-ray beam is incorporated into the reconstruction, cOSSCIR has the potential to address beam hardening metal artifacts using accurate spectral estimation for both modalities. Furthermore, the use of convex constraints in the inversion procedure can be leveraged to address photon starvation caused by metal using a masking technique.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Purpose</h3>\u0000 \u0000 <p>The purpose of this study was to experimentally investigate metal artifact reduction performance within a one-step framework for spectral CT reconstruction.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>Dual-kV and photon-counting transmission measurements were taken through a step-wedge phantom and were used to form spectral models for an energy-integrating and photon-counting detector for our benchtop micro-CT imaging system. Then experiments were performed using an electron-density rod phantom with and without metal to evaluate the effect of spectral modeling and masking on the cOSSCIR reconstructions for both imaging modalities. Further experiments were performed using a biological specimen to demonstrate cOSSCIR and the artifact reduction techniques on a phantom with more complicated structure. Finally, cOSSCIR was demonstrated on head CT data with dental hardware obtained from a first-generation silicon PCCT clinical prototype.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>The cOSSCIR reconstructions demonstrated a reduction in metal artifacts for both dual-kV and photon counting data. For a region between the metal inserts of the electron-density phantom, the mean absolute errors in the 60 keV dual-kV virtual mono-energetic images were 115 HU for FBP, compared to 87 HU and 41 HU using cOSSCIR and cOSSCIR with a mask, respectively. Likewise, the mean absolute errors in the 60 keV virtual mono-energetic images were 217 HU for FBP, 71 HU using cOSSCIR, and 22 HU using cOSSCIR with a mask in the 60 keV photon-counting virtual mono-energetic images. In a region adjacent to the metal, the dual-kV images had mean absolute errors of 100 HU for FBP, 53 HU for cOSSCIR, and 41 HU for cOSSCIR+Mask and the photon-counting images had errors of 74 HU for FBP, 27 HU for cOSSCIR, and 63 HU for cOSSCIR+Mask. The additional spectral infor","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148972827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Correction to “Improving textural realism in breast phantom images”","authors":"","doi":"10.1002/mp.70644","DOIUrl":"10.1002/mp.70644","url":null,"abstract":"<p>Omena LM, de Oliveira GM, do Rêgo TG, et al. Improving textural realism in breast phantom images. <i>Med Phys</i>. 2026; 53:e70501. doi: 10.1002/mp.70501</p><p>The above listed article was mistakenly first published with the article type of Research Article. The article, however, was intended to publish as a Technical Note. The online version of this article has been corrected accordingly.</p><p>The Publisher apologizes for this error.</p>","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13575301/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148972863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Roadmap for the implementation of particle minibeam therapy","authors":"Reem Ahmad,&nbsp;Oxana Actis,&nbsp;Adam H. Aitkenhead,&nbsp;Richard A. Amos,&nbsp;Niels Bassler,&nbsp;Catherine Burne,&nbsp;Pankaj Chaudhary,&nbsp;Gerd Datzmann,&nbsp;Umberto Deut,&nbsp;Fabio Di Martino,&nbsp;Rebecca Habgood,&nbsp;Simon Jolly,&nbsp;Emmanuel Jouglar,&nbsp;Eva Kasanda,&nbsp;Nigel Lee,&nbsp;Dominic Leiser,&nbsp;Xiangkun Li,&nbsp;Giuliana Milluzzo,&nbsp;Jessica Neubauer,&nbsp;Kelvin Ng Wei Siang,&nbsp;Hugo Palmans,&nbsp;Yolanda Prezado,&nbsp;Judith Reindl,&nbsp;Francesco Romano,&nbsp;Loris Roncali,&nbsp;Aikaterini Rousseti,&nbsp;Thomas E. Schmid,&nbsp;Ileana Silvestre Patallo,&nbsp;Frank Stephan,&nbsp;Anna Subiel,&nbsp;Edward Taylor,&nbsp;Annalisa Walpen,&nbsp;Samuel Flynn","doi":"10.1002/mp.70660","DOIUrl":"10.1002/mp.70660","url":null,"abstract":"<p>Particle minibeam therapy (PMBT) has emerged as an advanced modality within spatially fractionated radiation therapy, using the physical properties of protons, ions, and electrons to achieve superior normal tissue sparing whilst preserving spatial dose distribution at depths that are unachievable in conventional kV or MV X-rays. As with any novel radiotherapy technique, successful translation into routine clinical practice requires robust evidence of efficacy, safety, and technical feasibility. This review presents a coordinated roadmap for the development and future implementation of PMBT, identifying the key challenges that must be addressed to enable safe and effective clinical implementation. This review highlights the interconnected nature of PMBT research and proposes the priority areas for future investigation, and the steps necessary to guide PMBT from experimental concept, towards clinical implementation. As most of the steps are relevant to all particle types, PMBT will be referred to in general, where specific particle types will be discussed where appropriate.</p>","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13575480/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148972881","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design and imaging performance of BPET–DBT – a dedicated scanner for breast imaging","authors":"Srilalan Krishnamoorthy,&nbsp;Emmanuel Morales,&nbsp;William J. Ashmanskas,&nbsp;Matthew E. Werner,&nbsp;Samuel Matej,&nbsp;Juhi Raj,&nbsp;Trevor L. Vent,&nbsp;Chloe J. Choi,&nbsp;Andrew D. A. Maidment,&nbsp;David A. Mankoff,&nbsp;Christine E Edmonds,&nbsp;Joel S Karp,&nbsp;Suleman Surti","doi":"10.1002/mp.70651","DOIUrl":"10.1002/mp.70651","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;A dedicated breast PET (BPET) scanner integrated with digital breast tomosynthesis (DBT) enables three-dimensional functional imaging with quantitative information co-registered to high-resolution anatomical images. With the availability of breast cancer specific PET radiotracers, such a scanner has the potential to improve treatment planning in breast cancer. We recently developed a BPET–DBT scanner utilizing an integrated gantry design to generate intrinsically co-registered BPET–DBT images.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;This paper presents the scanner design, characterizes its imaging performance, and reports on our initial human imaging experience to demonstrate scanner's clinical readiness.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;The PET component is comprised of two detector heads, providing a 20 × 10 × 10 cm&lt;sup&gt;3&lt;/sup&gt; scanner field-of-view. The detector provides time-of-flight (TOF) measurement and is comprised of 32×32 LYSO arrays of 1.5 × 1.5 × 15 mm&lt;sup&gt;3&lt;/sup&gt; crystals coupled to multi-anode PMTs. Custom waveform-sampling electronics ensures high spatial and timing resolution with minimal deadtime. The DBT component is a state-of-the-art system which employs 2D x-ray tube and detector motion for improved contrast and resolution. Phantoms appropriate for breast imaging were imaged to assess spatial resolution, image quantitation, count-rate capability, and co-registration accuracy. First human imaging was performed in volunteers with estrogen-receptor-positive (ER+) breast cancer. A five-minute BPET–DBT scan using &lt;sup&gt;18&lt;/sup&gt;F-fluoroestradiol was acquired and compared with a whole-body PET scan.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Results&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;System timing resolution of 425 ps and energy resolution of 16% were measured. Spatial resolution measurements showed an in-plane (parallel to the detector) resolution of 2 mm and demonstrates the ability to resolve 1.6 mm rods in the micro-deluxe hot rod phantom. TOF reconstruction combined with image-based resolution modeling mitigates artifacts typical in scanners with incomplete angular coverage and enables good out-of-plane (orthogonal to detector) discrimination of the 2.4 mm rods. Count-rate measurements confirm low scanner dead-time, with sufficient count-rate capability for clinical use.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Conclusions&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;We successfully constructed and tested a BPET scanner and integrated it with DBT. BPET scanne","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13575451/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148972829","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Inter-crystal scatter recovery of a light-sharing depth-encoding Prism-PET prototype scanner using a diffusion model","authors":"Wanbin Tan,&nbsp;Saurav Dosi,&nbsp;Soroush Shabani Sichani,&nbsp;Yixin Li,&nbsp;Zipai Wang,&nbsp;Eric Petersen,&nbsp;Xinjie Zeng,&nbsp;Jinyi Qi,&nbsp;Amir H. Goldan","doi":"10.1002/mp.70661","DOIUrl":"10.1002/mp.70661","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Inter-crystal scattering (ICS) is a fundamental limitation in ultra-high resolution positron emission tomography (PET), particularly in light-sharing detectors with small crystal elements, where frequent Compton scatter across neighboring crystals obscures the initial interaction position and degrades the spatial resolution, image contrast, and quantitative accuracy. In our previous work, a convolutional neural network, Recovery-Net, was proposed and evaluated to recover the crystal location and depth-of-interaction (DOI) of the initial interaction in a light-sharing depth-encoding Prism-PET detector using Monte Carlo (MC) simulations.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;This work aims to develop and experimentally validate a scanner-level ICS recovery framework for a light-sharing depth-encoding Prism-PET prototype scanner using a diffusion model.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;We first developed a novel method to prepare the training dataset by combining scanner-level simulations with experimental measurements. The resulting dataset comprises an 8 × 8 array of silicon photomultiplier (SiPM) signals for each ICS event with associated ground-truth information, including crystal location, DOI, and deposited energy of the initial interaction. Using the SiPM signal patterns as input, we proposed a new diffusion-driven and physics-informed tripartite framework, named D3Recovery, to estimate the 3D position and deposited energy of the initial interaction. The previously proposed Recovery-Net was also implemented and evaluated for comparison. An ultra-micro Derenzo phantom and a 3D Hoffman brain phantom from the Prism-PET prototype scanner experiment were used to evaluate the impact of ICS recovery on image quality.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Results&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Recovery-Net and D3Recovery achieved crystal identification accuracies of 64.1% and 70.1% for ICS events, respectively, with D3Recovery further improving DOI estimation to a full width at half maximum (FWHM) of 3.6 mm compared with 5.1 mm for Recovery-Net. Compared with center-of-gravity (CoG) positioning, D3Recovery reduced the mean crystal localization error from 2.18 mm to 0.99 mm and recovered the deposited energy with an error FWHM of 34.0 keV. For the ultra-micro Derenzo phantom images, Recovery-Net and D3Recovery increased peak-to-valley ratios by 13% and 24% for ICS events and by 14% and 17% for all events, respectively. For the Hoffman brain phantom, Recovery-Net and D3Recovery reduced mean absolute error (MAE) by 15.53% and 17.84% for ICS events and by 16.22% and 18.63% for all events, while increasing str","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148972847","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Cross-fraction prior learning for scalable organ-at-risk segmentation in abdominal MR-guided radiotherapy","authors":"Chengyin Li,&nbsp;Doris Rusu,&nbsp;Rafi Ibn Sultan,&nbsp;Hassan Bagher-Ebadian,&nbsp;Anthony Doemer,&nbsp;Dongxiao Zhu,&nbsp;Parag J. Parikh,&nbsp;Kundan Thind","doi":"10.1002/mp.70648","DOIUrl":"https://doi.org/10.1002/mp.70648","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Manual organ-at-risk (OAR) delineation takes 20–40 min per case, a major bottleneck within the 50–90 min treatment window of abdominal MR-guided adaptive radiotherapy (MRgRT). Most deep learning systems adopt single-fraction approaches that discard valuable temporal context from prior treatment fractions.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;This study develops AdaptSeg, a scalable framework leveraging cross-fraction anatomical priors to substantially improve OAR segmentation without per-patient retraining.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;We implemented a dual-path neural architecture conditioning current fraction segmentation on paired image-mask information from supporting fractions. AdaptSeg was instantiated with convolutional (3D UNet) and transformer-based (SwinUNETR) backbones. Evaluation used 104 pancreatic cancer patients across 520 treatment fractions for four abdominal organs (colon, duodenum, small bowel, stomach), with patient-level splitting: 72 training, 10 validation, 22 test patients. Performance metrics included Dice Similarity Coefficient (DSC), 95th percentile Hausdorff Distance (HD95), and Average Symmetric Surface Distance (ASSD); paired comparisons used two-sided Wilcoxon signed-rank tests with Benjamini–Hochberg correction, and 95% bootstrap confidence intervals for the means.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Results&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Cross-fraction priors improved segmentation performance for both tested backbones. The 3D UNet achieved 87.22% mean DSC versus 83.78% baseline, while SwinUNETR reached 85.19% versus 82.49% baseline. For highly deformable organs, improvements included up to 7.0 percentage point DSC gains (small bowel: 77.8% to 84.8%, &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mi&gt;p&lt;/mi&gt;\u0000 &lt;mo&gt;&lt;&lt;/mo&gt;\u0000 &lt;mn&gt;0.001&lt;/mn&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$p &lt; 0.001$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt;) and 62% boundary error reduction (colon HD95: 23.13 to 8.74 mm, &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mi&gt;p&lt;/mi&gt;\u0000 &lt;mo&gt;&lt;&lt;/mo&gt;\u0000 &lt;mn&gt;0.001&lt;/mn&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$p &lt; 0.001$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt;). Compared to nine state-of-the-art methods, AdaptSeg achieved the best overall performance with substantial improvements in mean DSC (4.1%), HD95 (43%), and ASSD (39%) over the strongest baseline. All variants maintained computa","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148942941","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From Skull Density Ratio-based eligibility to efficiency-informed planning: Predictive modeling of thermal response in MR-guided focused ultrasound","authors":"José Angel Pineda-Pardo,&nbsp;Jaime Caballero-Insaurriaga,&nbsp;Marta Castillo-Ortiz,&nbsp;Miguel López-Aguirre,&nbsp;Rafael Rodríguez-Rojas,&nbsp;Jorge U. Máñez-Miró,&nbsp;Michele Matarazzo,&nbsp;Marta Del Alamo,&nbsp;Raul Martínez-Fernández","doi":"10.1002/mp.70652","DOIUrl":"10.1002/mp.70652","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Transcranial MR-guided focused ultrasound (MRgFUS) enables incisionless thermoablation of deep brain targets and has become an established modality in functional neurosurgery. Treatment efficiency is strongly influenced by skull-mediated ultrasound attenuation. Although the skull density ratio (SDR) is currently the primary parameter used for pre-procedural screening, it provides only a partial description of skull-related variability in thermal response and offers limited guidance for intra-procedural energy titration.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;To comprehensively characterize skull-related determinants of thermal efficiency beyond SDR and to develop multivariate models for predicting treatment efficiency and sonication-wise peak temperature to support patient screening and intra-procedural decision-making.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;We retrospectively analyzed 316 MRgFUS thermoablative procedures (214 thalamotomies and 102 subthalamotomies). Skull metrics derived from CT included SDR, skull thickness (ST), diploe thickness (DT), angle of incidence (AOI), and higher-order distributional descriptors. Thermal efficiency was quantified using multiple temperature-energy-based estimators. Two predictive models were developed, each validated on an independent held-out test set (20% of procedures): (1) a forward-selected multivariate linear regression model for procedure-level thermal efficiency (temperature-to-energy ratio at 55°C-TER&lt;sub&gt;55&lt;/sub&gt;), guided by adjusted &lt;i&gt;R&lt;/i&gt;&lt;sup&gt;2&lt;/sup&gt; and AIC, with performance reported as &lt;i&gt;R&lt;/i&gt;&lt;sup&gt;2&lt;/sup&gt;; and (2) a gradient boosting model for sonication-wise peak temperature prediction, integrating sonication parameters, skull metrics, and cumulative treatment history, with performance reported as MAE globally and stratified by temperature range.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Results&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;SDR, ST, and DT were significantly associated with thermal efficiency, and composite metrics such as SDR/DT outperformed SDR alone. A five-feature multivariable linear regression model based on skull-derived metrics achieved an adjusted &lt;i&gt;R&lt;/i&gt;&lt;sup&gt;2&lt;/sup&gt; of 0.63 (cross-validated &lt;i&gt;R&lt;/i&gt;&lt;sup&gt;2&lt;/sup&gt; = 0.61, MAE = 0.54°C/kJ) for predicting TER&lt;sub&gt;55&lt;/sub&gt;, with consistent generalization on an independent held-out test set (&lt;i&gt;R&lt;/i&gt;&lt;sup&gt;2&lt;/sup&gt; = 0.58, MAE = 0.53°C/kJ). The peak temperature prediction model achieved a cross-validated MAE of 1.70°C on the training set and 1.91°C on a fully held-out procedure-level test set, with stable performanc","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/mp.70652","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148922349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Local large language model-guided optimization for prostate radiotherapy planning","authors":"Humza Nusrat,&nbsp;Bing Luo,&nbsp;Hassan Bagher-Ebadian,&nbsp;Ryan Hall,&nbsp;Joshua Kim,&nbsp;Anthony Doemer,&nbsp;Khaled Adil,&nbsp;Mohamed Elsheikh,&nbsp;Aharon Feldman,&nbsp;Benjamin Movsas,&nbsp;Kundan Thind","doi":"10.1002/mp.70654","DOIUrl":"10.1002/mp.70654","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>Radiotherapy planning is time-intensive, iterative, and operator-dependent. Locally deployed large language-model agents may support planning automation while avoiding transmission of patient data to external model providers.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Purpose</h3>\u0000 \u0000 <p>The purpose of this work was to develop and evaluate SAGE, a locally deployed LLM agent for autonomous VMAT optimization-priority tuning, and to investigate the impact of model size (8B vs. 70B) and intra-case memory implemented through retrieval augmentation.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>The SAGE agent (Llama 3.1 8B/70B) was interfaced with a clinical TPS to tune optimization priority values for 17 retrospective prostate cancer patients (60 Gy/20fx). All SAGE priority values were initialized to 1, while the study-defined optimization objective set and beam geometry were held fixed. We compared four agent variants (8B, 70B, 8B-RAG, 70B-RAG) to the clinical plan using paired Wilcoxon tests with FDR control.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>No statistically significant difference in target (PTV V58.5 Gy) coverage was observed between SAGE variants and clinical plans. The 70B-RAG agent significantly improved penile bulb sparing (V22Gy) over clinical plans (<i>p</i> = 0.014). Enabling intra-case memory provided a statistically significant improvement in penile bulb sparing, while no significant effect was observed from model size alone. While clinical plans were statistically better for some OARs, all SAGE 70B-RAG plans remained clinically acceptable. The most complex version of SAGE (70B-RAG) completed ten autonomous iterations in a median compute time of 45 min (IQR: 34–51). For reference, the clinical planning workflow turnaround (ARIA-recorded task duration) median time was 2700 min (IQR: 2040–3060).</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusions</h3>\u0000 \u0000 <p>This proof-of-concept study demonstrates that a local LLM agent can autonomously tune optimization priority values to generate clinically acceptable prostate VMAT plans within a fixed, study-defined objective space. Intra-case memory produced a statistically significant, endpoint-specific improvement (penile bulb V22Gy). These findings suggest that smaller, efficient models may be viable for constrained treatment-planning optimization tasks.</p>\u0000 </section>\u0000 </div>","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/mp.70654","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148922347","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of collimator misalignments on dosimetric robustness in proton minibeam radiotherapy using an automated measurement setup","authors":"Sake Salverda,&nbsp;Ernst van der Wal,&nbsp;Thomas Toet,&nbsp;Mischa Hoogeman,&nbsp;Kelvin Ng Wei Siang","doi":"10.1002/mp.70671","DOIUrl":"10.1002/mp.70671","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Proton minibeam radiation therapy is an emerging technique that combines the tissue-sparing benefits of protons with the spatial modulation used for minibeams to reduce toxicity in healthy tissue. Achieving accurate delivery of planned dose distributions depends on precise collimator-to-beam alignment, for which it is essential to understand the behavior of both multi-slit and single-slit configurations. Understanding single-slit behavior is particularly important, as it also linearly affects the heterogeneity (peaks and valleys) of the overall minibeam. While several studies have investigated misalignment effects with multi-slit collimators, there is limited experimental characterisation of single-slit behavior and its impact on multi-slit arrays.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;This study aims to quantify the dosimetric effects of translational and rotational collimator-to-beam misalignments for single-slit proton minibeams using our in-house automated measurement setup. Additionally, the robustness of lateral dose profiles for minibeam arrays will be investigated under varying collimator misalignments.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;A single slit collimator with &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mn&gt;150&lt;/mn&gt;\u0000 &lt;mspace&gt;&lt;/mspace&gt;\u0000 &lt;mi&gt;M&lt;/mi&gt;\u0000 &lt;mi&gt;e&lt;/mi&gt;\u0000 &lt;mi&gt;V&lt;/mi&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$150 ,mathrm{M}mathrm{emathrm{V}}$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt; protons was used to produce a single minibeam with adjustable collimator slit widths (0.4 and &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mn&gt;0.7&lt;/mn&gt;\u0000 &lt;mspace&gt;&lt;/mspace&gt;\u0000 &lt;mi&gt;m&lt;/mi&gt;\u0000 &lt;mi&gt;m&lt;/mi&gt;\u0000 &lt;/mrow&gt;\u0000 &lt;annotation&gt;$0.7 ,mathrm{m}mathrm{m}$&lt;/annotation&gt;\u0000 &lt;/semantics&gt;&lt;/math&gt;). We introduced rotational and translational misalignments using motorized stages. We developed a quick measurement setup to measure lateral profiles using a microDiamond detector. The lateral profiles were measured at depths of 1 and &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mn&gt;13&lt;/mn&gt;\u0000 &lt;mspace&gt;&lt;/mspace&gt;\u0000 &lt;mi&gt;c&lt;/mi&gt;\u0000 &lt;mi&gt;m&lt;/mi&gt;\u0000 &lt;/mrow&gt;\u0000 ","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13548184/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148901779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dual-branch MRI denoising and motion artifact suppression with uncertainty estimation and segmentation-based evaluation","authors":"Mingfu Jiang,&nbsp;Chi Kin Lam,&nbsp;Zhuoneng Zhang,&nbsp;Poorya MohammadiNasab,&nbsp;Hing-Chiu Chang,&nbsp;Tong Tong,&nbsp;Hualong She,&nbsp;Zhen Yuan,&nbsp;Tao Tan","doi":"10.1002/mp.70662","DOIUrl":"10.1002/mp.70662","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Background&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;Magnetic resonance (MR) images are essential for clinical diagnosis, yet their quality is often compromised by noise and motion artifacts, potentially obscuring anatomical details and reducing diagnostic accuracy. Existing deep learning approaches frequently address denoising and motion artifact suppression as separate tasks, overlook pixel correlations and varying noise levels, require manual parameter tuning, and lack uncertainty evaluation, thus limiting their practical utility in clinical settings.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Purpose&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;This study aims to develop a Dual-branch MRI Denoising and Motion Artifact Suppression with Uncertainty Estimation and Segmentation-based Evaluation (DbUSNet). The model is designed to adaptively handle varying noise and motion artifact levels, preserve complex structural details, and provide uncertainty estimates for the processed outputs. This may offer insights into downstream clinical tasks such as image segmentation.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Methods&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;The DbUSNet integrates two specialized branches: A Heterogeneous Window Transformer Network (HWTNet) for adaptive denoising using both local and global feature modeling, and an improved U-Net based Motion Artifact Suppression Network (ASNet). Each branch is coupled with an estimation subnetwork—Noise Estimation Subnetwork (NES) and Artifact Estimation Subnetwork (AES)—to dynamically assess and adapt to input specific degradation levels. The outputs of both branches are fused via an attention-based fusion module. Uncertainty is quantified during inference using Monte Carlo (MC) dropout. The model was trained and tested on brain MR scans from Brats2021 dataset (2000 samples), OpenNeuro dataset (22 samples), and MR-ART dataset (148 samples), incorporating synthetic Rician noise and motion artifacts. Downstream segmentation performance was evaluated on the Medical Segmentation Decathlon (MSD) brain dataset (1666 samples) and a prostate MR dataset (203 samples).&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Results&lt;/h3&gt;\u0000 \u0000 &lt;p&gt;On the Brats2021, OpenNeuro, and MR-ART testsets, DbUSNet achieved PSNR of 28.23 dB, 27.13 dB, and 27.83 dB, SSIM of 0.93, 0.91, and 0.92, NMSE of 0.08, 0.08, and 0.08, respectively. For the downstream segmentation task, the denoised and motion artifact suppressed outputs led to average Dice score improvements of 3.10%, 3.59%, 2.46%, and 2.15% for T1, T1ce, FLAIR, and T2 modalities in the MSD brain dataset, and 2.16% and 2.24% for T2 and ADC sequences in the prostate dataset.&lt;/p&gt;\u0000 &lt;/section&gt;\u0000 \u0000 &lt;section&gt;\u0000 \u0000 &lt;h3&gt; Conclusion&lt;/h3&gt;\u0000 ","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"53 9","pages":""},"PeriodicalIF":3.2,"publicationDate":"2026-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148901720","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}