Physics and Imaging in Radiation Oncology最新文献

{"title":"Feasibility of longitudinal prostate R2⁎ measurements in patients undergoing stereotactic ablative radiotherapy on an MRI-linac","authors":"Chris Moore,&nbsp;Asher Ezekiel,&nbsp;Vodathi Bamunuarachchi,&nbsp;Ereny Saad,&nbsp;Anubhav Datta,&nbsp;Peter Hoskin,&nbsp;Ananya Choudhury,&nbsp;Marcel van Herk,&nbsp;David L. Buckley,&nbsp;Michael J. Dubec,&nbsp;Damien J. McHugh","doi":"10.1016/j.phro.2026.101103","DOIUrl":"10.1016/j.phro.2026.101103","url":null,"abstract":"<div><h3>Background and Purpose</h3><div>Tissue R₂* (1/T₂*), sensitive to deoxyhaemoglobin, is a proposed hypoxia biomarker in prostate cancer. This exploratory work investigates prostate R₂* and volume throughout and after hypo-fractionated radiotherapy and the effects of androgen deprivation therapy (ADT).</div></div><div><h3>Materials and methods</h3><div>Fifteen patients (seven with ADT) received five-fraction stereotactic-ablative-radiotherapy (SABR) on a 1.5 T Elekta Unity MRI-linac, with R₂* mapping at MR-Simulation and treatment visits 1, 3, 5. Six patients had follow-up imaging post-treatment.</div><div>Treatment-induced whole-prostate and tumour R₂*/volume differences were investigated using repeated-measures analysis-of-variance.</div><div>ADT effects on baseline whole-prostate R₂*/volume were investigated using unpaired <em>t</em>-tests. Spearman correlations between whole-prostate R₂* and volume were calculated before and during treatment.</div></div><div><h3>Results</h3><div>Whole-prostate R₂* (mean difference [95% CIs] = −1.77 [−2.17 to −0.498] s⁻¹, <em>p</em> = 0.003) and tumour R₂* (−3.11 [−5.78 to −1.44] s⁻¹, <em>p</em> = 0.003) decreased, whole-prostate volume (8.05 [4.76 to 11.7] cm³) increased after two fractions.</div><div>Follow-up whole-prostate R₂* (<em>p</em> = 0.62) and volume (<em>p</em> = 0.16) were not significantly different from baseline.</div><div>ADT patients had higher baseline whole-prostate R₂* (<em>p</em> = 0.006) and lower volume (<em>p</em> = 0.003). ADT did not impact treatment-induced changes (<em>p</em> = 0.51).</div><div>Baseline whole-prostate R₂*/volume were negatively correlated (<em>r</em> = −0.75, <em>p</em> = 0.002); treatment-induced R₂*/volume changes were not correlated <em>r</em> = 0.44, <em>p</em> = 0.11).</div></div><div><h3>Conclusions</h3><div>Treatment-induced whole-prostate and tumour R₂* changes were measured after two SABR fractions. Post-treatment whole-prostate R₂* and volume were not significantly different from baseline.</div><div>Further work is required to confirm the mechanism(s) causing observed treatment effects.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101103"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149141481","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning segmentation of the cerebral ventricular system and brainstem for pediatric radiotherapy planning","authors":"Feline Heinzelmann,&nbsp;Fabian Hörst,&nbsp;Lukas Heine,&nbsp;Jana Fragemann,&nbsp;Moritz Rempe,&nbsp;Gijs Luijten,&nbsp;Sarah Peters,&nbsp;Julian Hlouschek,&nbsp;Jan Egger,&nbsp;Beate Timmermann,&nbsp;Jens Kleesiek,&nbsp;Christian Bäumer","doi":"10.1016/j.phro.2026.101086","DOIUrl":"10.1016/j.phro.2026.101086","url":null,"abstract":"<div><h3>Background and Purpose</h3><div>Segmentation of the cerebral ventricular system (VS) is central to radiotherapy, both for whole-ventricular irradiation in intracranial germ cell tumors (GCTs) and protection of the periventricular region. Complex, variable VS anatomy makes manual delineation challenging and hampers auto-segmentation model development. We reported the evaluation and workflow integration of a deep-learning (DL) model for VS and brainstem segmentation on pediatric magnetic resonance imaging (MRI).</div></div><div><h3>Materials and methods</h3><div>An nnU-Net was trained on MRIs from 76 unique brain tumor cases (<em>n</em> = 73/71 for T1/T2). The evaluation used two independent retrospective cohorts: (i) GCTs evaluated on planning computed tomography (CT) images after MRI-to-CT registration mapping (<em>n</em> = 13/14), and (ii) brain tumors evaluated on MRI (<em>n</em> = 14). Performance was assessed using the Dice similarity coefficient (DSC) and Hausdorff distances (HD) against manual contours. Prospective deployment in routine clinical planning assessed usability and workflow integration.</div></div><div><h3>Results</h3><div>In the GCT-CT cohort, brainstem segmentations achieved median DSCs of &gt;0.85 and HDs₉₅ of ≤5.0/6.4 mm, while VS segmentations achieved DSCs of &gt;0.70 and HDs₉₅ of ≤5.3/4.3 mm, after rigid/deformable mapping to planning CT. In the MR cohort, performance on T1−/T2-weighted MRI reached DSCs of ≥0.90 for both structures, and HDs₉₅ of ≤3.0/3.4 mm and ≤ 1.3/2.2 mm for the brainstem and VS, respectively. The DL-segmentation workflow reduced workflow time by &gt;50% compared with manual contouring.</div></div><div><h3>Conclusions</h3><div>Lower performance in the GCT-CT cohort likely reflects noisier CT-based references, interobserver variability, and uncorrected geometric/anatomical mismatch. The model's accuracy supports radiotherapy planning in pediatric brain tumors through risk-structure delineation and target-volume segmentation in GCTs.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101086"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149141483","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Target-volume-based prescription selection for specified adverse-effect levels in automated stereotactic radiosurgery","authors":"Joel A. Pogue,&nbsp;Rex Cardan,&nbsp;Farnaz Rahim Li,&nbsp;Rodney Sullivan,&nbsp;Natalie Viscariello,&nbsp;Luke Moradi,&nbsp;John Fiveash,&nbsp;Richard Popple","doi":"10.1016/j.phro.2026.101093","DOIUrl":"10.1016/j.phro.2026.101093","url":null,"abstract":"<div><div>Automated stereotactic radiosurgery (SRS) planning produced consistent dose falloff, enabling adverse-effect-relevant isodose volumes (<span><math><mi>IDV</mi></math></span>) to be estimated before plan generation. We derived a direct isoeffect prescription dose (<span><math><msub><mi>D</mi><mi>RX</mi></msub></math></span>) model from previously validated target-volume (<span><math><msub><mi>V</mi><mtext>target</mtext></msub></math></span>) power laws for single-isocenter automated SRS. <span><math><mi>IDV</mi></math></span> coefficients and exponents were modelled as continuous functions of relative isodose level (<span><math><mi>RIL</mi></math></span>), allowing <span><math><msub><mi>D</mi><mi>RX</mi></msub></math></span> to be solved from <span><math><msub><mi>V</mi><mtext>target</mtext></msub></math></span>, desired <span><math><mi>IDV</mi></math></span>, and HyTEC brain adverse-effect dose thresholds. Across held-out levels, target-level <span><math><mi>IDV</mi></math></span> residuals had a median of 0.2%, an interquartile range of [−5.2%, +7.0%], and a 95% interval of [−12.8%, +27.0%].</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101093"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149141526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Improved chest wall stability during deep-inspiration breath-hold in breast cancer radiotherapy following quality improvement initiatives","authors":"Kristine Wiborg Høgsbjerg,&nbsp;Anders W. Mølby Nielsen,&nbsp;Esben Yates,&nbsp;Christina Maria Lutz,&nbsp;Per Rugaard Poulsen,&nbsp;Birgitte Vrou Offersen,&nbsp;Harald Spejlborg","doi":"10.1016/j.phro.2026.101079","DOIUrl":"10.1016/j.phro.2026.101079","url":null,"abstract":"<div><div>Intrafractional chest wall motion during radiotherapy in deep-inspiration breath-hold may compromise dose delivery to the internal mammary nodes in patients with breast cancer. The present study evaluated chest wall motion after implementation of four treatment quality improvement initiatives, including repositioning of the gating marker block from the upper abdomen to the chest wall. Geometrical errors were quantified using cine MV images. The occurrence of geometrical errors &gt;3 mm was 4.7%, which was significantly lower than before the improvement initiatives (21.2%, <em>p</em> &lt; 0.0001). However, predominantly mild skin-related side effects were reported by 61% of patients caused by the gating marker block overlapping with the treatment field.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101079"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149052297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dose accuracy evaluation of multi-leaf collimator modeling in single-isocenter multi-target stereotactic radiosurgery","authors":"Farnaz Rahim Li,&nbsp;Rex A. Cardan,&nbsp;Joel A. Pogue,&nbsp;Natalie N. Viscariello,&nbsp;Rodney J. Sullivan,&nbsp;John B. Fiveash,&nbsp;Carlos E. Cardenas,&nbsp;Richard A. Popple","doi":"10.1016/j.phro.2026.101076","DOIUrl":"10.1016/j.phro.2026.101076","url":null,"abstract":"<div><h3>Background and purpose</h3><div>Accurate modeling of multi-leaf collimators is critical for dose calculation in stereotactic radiosurgery (SRS), where steep dose gradients and small target volumes amplify modeling uncertainties. This study compared enhanced leaf modeling and dosimetric leaf gap approaches using two clinically implemented dose calculation algorithms in single-isocenter SRS plans.</div></div><div><h3>Materials and methods</h3><div>Sixty clinical volumetric modulated arc therapy (VMAT) single-isocenter SRS plans (30 single-target, SIST; 30 multi-target, SIMT) were selected and measured with a two-dimensional diode array, normalized to a point measurement. Dose was calculated using both enhanced leaf model (ELM) and dosimetric leaf gap (DLG) models and the anisotropic analytical algorithm (AAA) and Acuros XB (AXB) algorithms across several spot-size configurations. Plan quality assurance was assessed using gamma analysis for several dose-distance criteria. Dose differences were further analyzed as functions of target size and distance from isocenter.</div></div><div><h3>Results</h3><div>Under 2%/1 mm gamma criteria, AXB-DLG having spot size 0.5 × 0.5 mm² showed the highest median and mean gamma pass rate (fraction of points having gamma &gt;1), 98.7% and 97.4% respectively, while AXB-ELM having spot-size 0.5 × 0.7 mm² had the lowest, 98% and 96.3%, respectively. SIST and SIMT plans showed no clinically significant differences. AAA-ELM (spot size: 0 × 0 mm²) achieved the highest median and mean gamma values, while AXB-DLG remained robust across plan types. Dose calculations were accurate for both configurations, with minimal mean differences (0.0%, range − 2.7% to +1.8%) reported.</div></div><div><h3>Conclusions</h3><div>DLG showed slightly better VMAT SRS agreement than ELM, but differences were clinically negligible. ELM's model-based MLC configuration removes user-specific DLG variability, potentially improving dose-calculation precision and consistency across institutions.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101076"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149052287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Type-B uncertainties related to ionization chamber modelling in Monte Carlo based reference dosimetry for clinical proton and carbon ion beams","authors":"Guillaume Houyoux,&nbsp;Nick Reynaert,&nbsp;Kilian-Simon Baumann","doi":"10.1016/j.phro.2026.101068","DOIUrl":"10.1016/j.phro.2026.101068","url":null,"abstract":"<div><h3>Background and Purpose:</h3><div>Monte Carlo simulations are increasingly used to derive beam quality correction factors (<span><math><msub><mrow><mi>k</mi></mrow><mrow><mi>Q</mi><mo>,</mo><msub><mrow><mi>Q</mi></mrow><mrow><mn>0</mn></mrow></msub></mrow></msub></math></span>) for reference dosimetry in clinical proton and carbon ion beams. Type-B uncertainties represent their dominant source of uncertainty, among which the impact of ionization chamber geometry and material properties on absorbed dose calculations has not been fully quantified yet.</div></div><div><h3>Materials and Methods:</h3><div>Seven ionization chamber models have been modelled in <figure><img></figure> , and independent perturbations of the geometrical dimensions, material mass densities and <span><math><mi>I</mi></math></span>-values were performed. For each chamber model, the absorbed dose in the sensitive volume was computed for clinical proton and carbon ion beams. The deviations from unperturbed simulations were used to estimate type-B uncertainties.</div></div><div><h3>Results:</h3><div>All perturbed configurations showed deviations within three standard uncertainties (type-A) of the reference result. Maximum deviations of 0.4% were observed across the three types of perturbation, independently of beam energy or chamber type. Each source of uncertainty contributed to roughly 0.2% individually, with a combined effect of <span><math><mrow><mo>∼</mo><mn>0.4</mn><mtext>%</mtext></mrow></math></span> on the absorbed dose.</div></div><div><h3>Conclusions:</h3><div>The type-B uncertainties associated with ionization chamber modelling do not dominate the overall uncertainty of ionization response functions <span><math><msub><mrow><mi>f</mi></mrow><mrow><mi>Q</mi></mrow></msub></math></span> for clinical proton and carbon ion beams which are the basis of Monte Carlo calculated <span><math><msub><mrow><mi>k</mi></mrow><mrow><mi>Q</mi><mo>,</mo><msub><mrow><mi>Q</mi></mrow><mrow><mn>0</mn></mrow></msub></mrow></msub></math></span> factors.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101068"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148937998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Clinical implementation of intensity modulated proton therapy in breath-hold for mediastinal lymphoma patients using surface monitoring and nasal high-flow therapy","authors":"Gloria Vilches-Freixas,&nbsp;Kim van der Klugt,&nbsp;Richard Canters,&nbsp;Esther Kneepkens,&nbsp;Fleur Vereijken,&nbsp;Maud Cobben,&nbsp;Maud van den Bosch,&nbsp;Indra Lubken,&nbsp;Anne van Engelen,&nbsp;Cissy Stultiens,&nbsp;Marije Velders,&nbsp;Giorgio Cartechini,&nbsp;Mirko Unipan,&nbsp;Maaike Berbée,&nbsp;Bastiaan Ta","doi":"10.1016/j.phro.2026.101071","DOIUrl":"10.1016/j.phro.2026.101071","url":null,"abstract":"<div><h3>Background and purpose</h3><div>We report our first clinical experience treating mediastinal lymphoma patients in breath-hold (BH) with intensity modulated proton therapy (IMPT), using optical surface scanning (OSS) for beam gating and nasal high-flow therapy (NHFT). Moreover, we evaluated BH stability throughout treatment and assessed adequacy of applied treatment margins.</div></div><div><h3>Materials and methods</h3><div>Thirteen patients received IMPT-BH with anterior beams supported by NHFT and OSS. OSS enabled gated treatment delivery with a 3 mm breathing window and a maximum isocenter shift of 5 mm. OSS respiratory traces were synchronized with machine log-file beam delivery data to identify beam-on periods. BH stability, BH duration, and OSS-derived isocenter shifts were assessed during cone beam CT (CBCT) and treatment. Margin adequacy was evaluated using post-treatment CBCTs acquired for 34 fractions (2–3 CBCTs per patient), while plan robustness over time was assessed with weekly repeat-CTs.</div></div><div><h3>Results</h3><div>Analyzing 187 fractions, the median number of BHs was 1 [1–6] per CBCT and 5 [3–20] per treatment delivery. Intra- and inter-BH surface stability were within 1 mm, and the median isocenter deviation during beam delivery was less than 3 mm. Average fraction time was 44 ± 13 min, decreasing as patients became more experienced. One plan adaptation was required in the entire cohort. Residual target shifts during BH treatments were &lt; 8 mm, comparable to motion in free-breathing.</div></div><div><h3>Conclusions</h3><div>IMPT-BH with OSS and NHFT is feasible for treating mediastinal lymphoma, providing stable BHs, and acceptable treatment times. Interfraction motion was slightly higher than intrafraction motion, suggesting that combining bone alignment with soft tissue adjustment could improve setup accuracy.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101071"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148937973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Three-dimensional deep learning nomogram for post-plan quality assurance in high-dose-rate prostate brachytherapy","authors":"Joel Beaudry,&nbsp;Tonghe Wang,&nbsp;Shirin A. Enger,&nbsp;Antonio L. Damato","doi":"10.1016/j.phro.2026.101100","DOIUrl":"10.1016/j.phro.2026.101100","url":null,"abstract":"<div><h3>Purpose</h3><div>Classical nomograms for high-dose-rate (HDR) prostate brachytherapy quality assurance (QA) evaluate agreement between planned and expected total dwell time but do not directly assess the three-dimensional treated volume. This study's aim was to develop a deep learning-based 3D nomogram to predict institution-specific clinical 100% isodose volumes from anatomy and dwell positions for post-plan QA.</div></div><div><h3>Methods</h3><div>A total of 2394 single-fraction 15 Gy HDR prostate brachytherapy plans were included. Input label maps encoded the 3D prostate, urethra, rectum, and dwell positions. The prediction target was the clinical 100% isodose volume. A five-fold cross-validation strategy was used for model development, with 2155 plans used for training and validation and 239 plans held out for testing. A 3D U-Net was trained using combined Dice and binary cross-entropy loss. Agreement was evaluated using prostate V100%, 100% isodose volume, Dice similarity coefficient, and average symmetric surface distance (ASSD). Validation-derived thresholds were used for outlier screening and targeted review.</div></div><div><h3>Results</h3><div>The model predicted prostate V100% with a mean error of 0.73 percentage points. Predicted and clinical 100% isodose volumes showed strong agreement, with R² = 0.990, slope = 0.99, and intercept = 0.77 cm³. Median Dice similarity coefficient was 0.960, and median ASSD was 0.48 mm. A 95% threshold flagged 6.3% of plans as atypical relative to validation data.</div></div><div><h3>Conclusions</h3><div>A deep learning-based nomogram predicted an institution-specific 100% isodose volume with low coverage error and high spatial agreement. This geometry aware post-plan QA tool can complement equation-based nomograms by directly evaluating the spatial treated volume.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101100"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149199856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Longitudinal radiomics of liver metastases during magnetic resonance-guided radiotherapy","authors":"Alina Paunoiu,&nbsp;Maksym Fritsak,&nbsp;Hubert S. Gabryś,&nbsp;Sebastian M. Christ,&nbsp;Matthias Guckenberger,&nbsp;Stephanie Tanadini-Lang,&nbsp;Riccardo Dal Bello","doi":"10.1016/j.phro.2026.101094","DOIUrl":"10.1016/j.phro.2026.101094","url":null,"abstract":"<div><h3>Background and purpose</h3><div>Magnetic resonance-guided radiotherapy (MRgRT) offers high-quality soft tissue contrast and treatment monitoring, supporting longitudinal radiomics analysis. We investigated treatment-related radiomic changes during stereotactic body radiotherapy (SBRT) for liver metastases and healthy liver.</div></div><div><h3>Materials and methods</h3><div>Features were extracted from the gross tumour volume (GTV) and five 2-cm liver rings around the planning target volume (PTV) in 34 patients treated with five-fraction MRgSBRT. Features showing changes below baseline variability or high correlation (Spearman's |r| &gt; 0.9) were excluded, with pre-treatment values averaged into a robust baseline. Linear regression slopes informed uniform manifold approximation and projection (UMAP) followed by hierarchical density-based spatial clustering of applications with noise (HDBSCAN) for patient stratification, with false discovery rate-corrected Kruskal–Wallis testing identifying representative features. Correlations with GTV <em>volume</em> and mean cumulative liver-ring dose were evaluated.</div></div><div><h3>Results</h3><div>Of 160 GTV features, 55 uncorrelated features remained. Three patient groups (<em>n</em> = 4, 10, and 20) with distinct longitudinal patterns were identified. <em>Intensity-based energy</em> showed significant differences across patient groups (adjusted <em>p</em> = 0.001), weakly correlated (|r| &lt; 0.7) with <em>volume</em> in the two larger groups. In the innermost liver ring, 49 uncorrelated features were analysed. Two patient groups (<em>n</em> = 19 and 15) were identified, with <em>intensity-based energy</em>, <em>mean intensity</em>, and <em>maximum intensity</em> showing significant group differences, weakly correlated with mean cumulative dose.</div></div><div><h3>Conclusions</h3><div>This longitudinal MR-based radiomics framework identified GTV and healthy liver patient groups, with distinct treatment-related feature changes. This pipeline could be extended to integrate treatment response assessment in future studies.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101094"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149159855","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning-based metal artifact reduction in CT for head and neck radiotherapy","authors":"Ehsan Mihandoust,&nbsp;Peyman Sheikhzadeh,&nbsp;Saeedeh Navaei Lavasani,&nbsp;Hossein Arabi,&nbsp;Davoud Ahmadvand","doi":"10.1016/j.phro.2026.101098","DOIUrl":"10.1016/j.phro.2026.101098","url":null,"abstract":"<div><h3>Background and purpose</h3><div>Metal artifacts from dental fillings and implants compromise CT image quality, target delineation, and dose calculation accuracy in head and neck (H&amp;N) radiotherapy. Previous reviews have largely focused on traditional metal artifact reduction methods and image quality metrics rather than radiotherapy-specific endpoints. This systematic review evaluates deep learning-based metal artifact reduction (DL-MAR) approaches for H&amp;N radiotherapy, emphasizing technical advances and clinical dosimetric validation.</div></div><div><h3>Materials and methods</h3><div>Following PRISMA guidelines, PubMed, IEEE Xplore, and Scopus were systematically searched (updated April 2026) for studies assessing DL-MAR methods in H&amp;N CT imaging and radiotherapy. Of 227 identified records, 40 studies (11 traditional MAR and 29 DL-MAR studies) met the inclusion criteria.</div></div><div><h3>Results</h3><div>Reported DL architectures included convolutional neural networks, generative adversarial networks, dual-domain networks, diffusion models, and task-oriented loss functions. Among studies with clinical dosimetric validation, GAN-based and task-oriented approaches consistently reported dose deviations below 3%. Gamma pass rates exceeding 95% (3%/3 mm) were reported in a subset of these studies. Dual-domain and diffusion-based methods were evaluated primarily using image quality metrics without direct radiotherapy-specific dosimetric validation.</div></div><div><h3>Conclusions</h3><div>Current evidence suggests that DL-MAR may achieve clinically acceptable dosimetric accuracy while improving artifact suppression and target visualization in H&amp;N radiotherapy planning. Task-oriented loss functions currently demonstrate promising potential for clinical translation based on available retrospective evidence. Broader clinical implementation remains limited by reliance on synthetic training data, restricted access to sinogram data, and the absence of prospective multicenter validation studies.</div></div>","PeriodicalId":36850,"journal":{"name":"Physics and Imaging in Radiation Oncology","volume":"41 ","pages":"Article 101098"},"PeriodicalIF":3.2,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149159856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}