Citations

Purpose Automating quality assurance (QA) for contours generated by automatic algorithms is critical in radiotherapy treatment planning. Manual QA is tedious, time-consuming, and prone to subjective experiences. Automatic segmentation reduces physician workload and improves consistency. However, an effective QA process for these automatic contours remains an unmet need in clinical practice. Materials and Methods The patient data used in this study was derived from the AAPM Thoracic Auto-Segmentation Challenge dataset, including left and right lungs, heart, esophagus, and spinal cord. Two groups of organ-at-risk (OAR) were generated. A ResNet-152 network was used as a feature extractor, and a one-class support vector machine (OC-SVM) was employed to classify contours as ‘high’ or ‘low’ quality. To evaluate the generalizability, we generated low-quality contours using translation and resizing techniques and assessed correlations between detection limits and metrics such as volume, Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), and mean surface distance (MSD). Results The proposed OC-SVM model outperformed binary classifiers n metrics such as balanced accuracy and area under the receiver operating characteristic curve (AUC) . It demonstrated superior performance in detecting various types of contour errors while maintaining high interpretability. Strong correlations were observed between detection limits and contour metrics. Conclusion Our proposed model integrates an attention mechanism with a one-class classification framework to automate QA for OAR delineations. This approach effectively detects diverse types of contour errors with high accuracy, significantly reducing the burden on physicians during radiotherapy planning.

Objective: To establish benchmark cutoffs for robotic liver resection (R-LR), encompassing both major and minor resections, and to determine the impact of patient selection on outcomes. Background: R-LR is a key advancement in minimally invasive liver surgery but lacks standardized benchmarks, especially for minor resections. While guidelines endorse R-LR, its role in optimizing outcomes remains unclear. This study establishes the first benchmarks for R-LR, enabling comparisons across surgical modalities and refining patient selection. Methods: This retrospective, multicenter study analyzed consecutive adult patients undergoing R-LR at 30 international centers (2020–2023). Benchmark centers had an annual case volume of ≥15 R-LR. Benchmark criteria included ASA ≤2, no major comorbidities, no prior liver resections, and Child-Pugh A status. Benchmark cutoffs for 14 key outcomes were set at the 50th and 75th percentiles of median values across benchmark centers. Multivariable logistic regression identified predictors of textbook outcomes. Results: Eighteen high-volume centers contributed 4028 cases with 2632 (65.3%) meeting benchmark criteria. Malignancy was the indication in 29.6%, most commonly hepatocellular carcinoma followed by colorectal liver metastases. Major liver resection was performed in 42.6%. The distribution of Iwate difficulty scores was low (25.4%), intermediate (53.9%), advanced (15.9%), and expert (4.9%). Benchmark cutoffs were established for minor and major resection, and stratified by Iwate low (0-3), intermediate (4-6) and high (7-12): Open conversion (≤6.5% minor, ≤10.5% major), major complications within 90 days (≤5.2% minor, ≤16.7% major), R1 resection (≤9.2% minor, ≤6.7% major). Benchmark cases performed at low-volume centers were able to achieve outcomes within the corresponding benchmark cutoffs. In fact, patient selection reflected by the proportion of benchmark patients, rather than case volume, was associated with textbook outcomes. Conclusions: This study defines R-LR benchmarks, emphasizing patient selection over center volume for optimal outcomes. Benchmark cutoffs guide training and support the expansion of R-LR.

Abstract The establishment of guidelines and curriculum standards for medical physics residency training is a critical component of setting expectations and competencies for the profession. Since the last publication of these standards, residency training has become integrated into the eligibility criteria for most medical physics certification bodies. The rapid growth of medical physics requires periodic review of the curriculum to remove outdated approaches, update established knowledge, and add emerging technologies. The goal of this document is to provide a reference standard for residency training program directors, training mentors, and residents to guide creation and/or optimization of their program's curriculum. The document is intentionally forward‐looking to capture emerging technological advances including hybrid diagnostic and therapeutic delivery systems and applications of image processing. This document is an update of the current residency training curriculum. Version history of residency training guidelines Publication Title Date 2013 Essentials and Guidelines for Clinical Medical Physics Residency Training Programs—AAPM Report No. 249 2006 Essentials and Guidelines for Hospital‐Based Medical Physics Residency Training Programs—Report No. 90 1992 Essentials and Guidelines for Hospital‐Based Medical Physics Residency Training Programs—Report No. 36 The Work Group on Periodic Review of Medical Physics Residency Training (WGMPRT) would like to acknowledge the authors of past versions of this document as they form the basis for this update. It is the sincere hope of the WGMPRT that this report will continue to be used by the medical physics community to guide training in the profession.

ABSTRACT Objective There is limited consensus on endoscopic skull base surgery (ESBS) reconstruction principles. This study aims to generate comprehensive themes regarding ESBS reconstruction by pooling the experiences of ESBS experts, with comparison to a literature review of current published evidence. Methods Structured qualitative interviews of ESBS experts regarding postoperative management and reconstruction of various defect locations were conducted. Results A total of 68 experts comprising 40 academic teams across 13 countries with an average of 18 years of ESBS experience were included. We propose 10 stepwise algorithms for common skull base reconstruction scenarios based on these expert interviews. When available, the nasoseptal flap is used for all high_flow cerebrospinal leak defects. Multilayered reconstruction is favored at all anatomical subsites with increasing number of layers for increasing defect size and complexity. Heterogeneity exists in terms of inlay technique and materials, free grafting versus various pedicled flap options for low‐flow defects or in the absence of a nasal septum, nasal packing, tissue sealant, lumbar drain use, and postoperative management. Commonalities and discrepancies between experts were summarized. Conclusion Skull base reconstruction and post‐ESBS management is highly complex with a wide variety of practice patterns and expert strategies. Further research of higher quality evidence is warranted to identify optimal management patterns, though the current work aims to inform surgeons on these controversial areas by drawing from numerous experiences.

Abstract INTRODUCTION There is a need to understand the clinical meaningfulness of symptom score changes in treatment trials of dementia‐related agitation. We estimated minimal clinically important differences (MCIDs) for commonly employed agitation scales and contextualized their clinical application. METHODS We employed anchor‐ and distribution‐based approaches to determine changes in scores corresponding to minimal symptom improvement. An opinion‐based approach assessed expert clinicians’ agreement on the meaningfulness of score changes through three clinical vignettes. RESULTS Minimal symptom improvement for Cohen‐Mansfield Agitation Inventory total score ranged from −4 (over <1 month) to −11 (over 1 to 3 months) points. Greater symptom severity correlated with higher MCID estimates. The clinical importance of score changes was influenced by treatment duration, pharmacological side effects, and impacts on caregiver distress/time resources. DISCUSSION The clinical meaningfulness of agitation scale MCIDs is influenced by trial‐specific and clinical factors. Shorter trial durations and measuring caregiver distress/time resources enhance the clinical interpretation of agitation treatment outcomes. Highlights For the CMAI total score, the MCID was −4 points over shorter time scales and −11 points for longer time scales. Worse agitation severity was associated with higher MCID estimates. There was high expert consensus that a noticeable treatment benefit was not worthwhile if it occurred after 12 weeks or had no impact on caregiver/staff distress/time resources.

Abstract Objective Unlocking clinical information embedded in clinical notes has been hindered to a significant degree by domain-specific and context-sensitive language. Identification of note sections and structural document elements has been shown to improve information extraction and dependent downstream clinical natural language processing (NLP) tasks and applications. This study investigates the viability of a dynamic example selection prompting method to section classification using lightweight, open-source large language models (LLMs) as a practical solution for real-world healthcare clinical NLP systems. Materials and Methods We develop a dynamic few-shot prompting approach to classifying sections where section samples are first embedded using a transformer-based model and deposited in a vector store. During inference, the embedded samples with the most similar contextual embeddings to a given input section text are retrieved from the vector store and inserted into the LLM prompt. We evaluate this technique on two datasets comprising two section schemas, including varying levels of context. We compare the performance to baseline zero-shot and randomly selected few-shot scenarios. Results The dynamic few-shot prompting experiments yielded the highest F1 scores in each of the classification tasks and datasets for all seven of the LLMs included in the evaluation, averaging a macro F1 increase of 39.3% and 21.1% in our primary section classification task over the zero-shot and static few-shot baselines, respectively. Discussion and Conclusion Our results showcase substantial performance improvements imparted by dynamically selecting examples for few-shot LLM prompting, and further improvement by including section context, demonstrating compelling potential for clinical applications.

ImportanceVestibular migraine has no established treatments.ObjectiveTo test the efficacy of rizatriptan for treating vestibular migraine attacks.Design, Setting, and ParticipantsThis double-blind, randomized clinical trial of rizatriptan vs placebo was conducted from December 2014 through July 2020 (data analyses in 2021 and sensitivity analyses in 2022 and 2024) at 2 tertiary neurotologic centers. Adults with vestibular migraine were included.InterventionAll participants underwent prospective observation to confirm diagnosis and illness activity and were then randomized in a 2:1 ratio to receive rizatriptan 10 mg or placebo to treat up to 3 vestibular migraine attacks per participant.Main Outcomes and MeasuresParticipants rated symptoms as absent, mild, moderate, or severe at scheduled intervals. Primary outcomes were the percentage of attacks with reductions in vertigo and unsteadiness/dizziness from moderate or severe to absent or mild at 1 hour. Secondary outcomes were the percentage of attacks with complete resolution of vestibular symptoms at 1 hour; reductions in headache and associated symptoms at 1 hour; use of rescue medications after 1 hour; reductions in vestibular, headache, and associated symptoms at 24 hours without rescue medications; treatment satisfaction and quality of life at 48 hours; and rates of serious adverse effects and discontinuation due to adverse effects.ResultsOf 222 total participants (mean [SD] age, 42.3 [11.7] years; 70.7% were women), 134 (60.4%) with active illness treated 307 attacks. Efficacy was tested using 240 attacks with vestibular symptoms rated as moderate or severe when participants took study drug. At 1 hour, rizatriptan did not differ from placebo for reducing vertigo (73/151 [48.3%] vs 50/88 [56.8%] attacks; odds ratio [OR], 0.71 [95% CI, 0.42-1.21]), unsteadiness/dizziness (29/151 [19.2%] vs 11/89 [12.4%] attacks; OR, 1.69 [95% CI, 0.80-3.57]), or any secondary outcomes. Similar percentages of participants in rizatriptan and placebo groups (26.4% for both groups) added rescue remedies after 1 hour. At 24 hours, rizatriptan had medium effects over placebo for unsteadiness/dizziness (OR, 2.65) and motion sensitivity (OR, 3.58). Post hoc analyses of all treated attacks found a medium effect favoring rizatriptan for headache and photophobia or phonophobia at 24 hours. Treatment satisfaction was equivocal. Quality of life was mixed. No participants experienced serious adverse effects or discontinued treatment for adverse effects.Conclusions and RelevanceIn this study, rizatriptan was ineffective at 1 hour for treating vestibular migraine attacks and had limited benefit on symptoms at 24 hours. Findings do not support using rizatriptan for vestibular migraine attacks.Trial RegistrationClinicalTrials.gov Identifier: NCT02447991

Abstract Introduction Resilient smokers, who maintain respiratory health despite smoking, remain poorly characterized. Airway epithelial cell gene expression in resilient smokers can provide insight into mechanisms and pathways that may protect smokers from developing smoking-related lung disease. The objective of this study is to identify genes and potential pathways associated with lung-related resilience. Methods Resilient smokers were defined based on a previously published multi-dimensional definition which includes the absence of: 1) spirometric airflow limitation, 2) symptoms (cough/sputum, dyspnea), 3) radiographic (HRCT) abnormalities, 4) exacerbations, and 5) an accelerated FEV1 decline. “Non-resilient” smokers had preserved spirometry but showed abnormalities in at least one domain of resilience. We analyzed airway epithelial brush samples from 21 resilient and 34 non-resilient smokers from SPIROMICS. Differential gene expression was performed using DESeq2 package in R, which estimates differential expression using generalized linear models. We adjusted for sequencing batch, age (continuous), race, gender, and current smoking status. While controlling for these covariates we evaluated differential gene expression between resilient and non-resilient smokers, the interaction between resilience and smoking status, and in each sub-category of resilient (absence of symptoms, radiographic abnormalities, exacerbations, and accelerated decline in FEV1). Results We identified 98 genes that were differentially expressed (FDR ≤0.1) when comparing resilient to non-resilient smokers, 20 of those genes were up regulated in resilient smokers (Figure 1). Notable genes included cadherin-12 (CDH12), aspartate beta-hydroxylase (ASPH), and protein kinase AMP-activated alpha 2 (PRKAA2). CDH12 is involved in cell adhesion and epithelial integrity. ASPH influences cell adhesion and signaling. PRKAA2 plays role in cellular homeostasis and stress response, which may be altered in non-resilient smokers, affecting cellular metabolism and inflammation. BPI folding containing family A member (BPIFA1) was significantly associated with smoking status in the interaction analysis, suggesting its role in modulating the impact of smoking on resilience. BPIFA1 has anti-microbial properties, regulates immune responses, and contributes mucosal surface protection. Sub-category analysis identified 17 genes associated with resilience to symptoms and 2 genes related to physiologic resilience (FEV1 decline), though no significant genes were associated with radiographic or exacerbation resilience. Conclusion Differential gene expression of airway epithelial cells identified distinct expression patterns between resilient and non-resilient smokers. Specifically, genes associated with epithelial cell adhesion, signaling, and cellular homeostasis were up regulated in resilient smokers. These findings suggest that resilient smokers may have transcriptomic profiles and potential mechanisms that mitigate the effects of cigarette smoke.

Abstract RATIONALE: Chronic Obstructive Pulmonary Disease (COPD) is a leading cause of mortality worldwide. We previously published the polygenic transcriptome risk score (PTRS) which uses the cumulative effect of predicted gene expression to construct genetic predictors for complex diseases, and we demonstrated the value of PTRS for COPD built on Genotype-Tissue Expression (GTEx) Lung tissue with significantly improved cross-ancestry portability. However, the postmortem collection and lack of lung disease status on GTEx samples likely affect results. We hypothesized that the performance of PTRS will be improved by using disease-relevant expression quantitative trait loci (eQTLs) to construct the score. Here, we aimed to improve performance of PTRS by leveraging disease-relevant eQTLs. METHODS: We constructed two transcriptome prediction models, Elastic Net (EN) and Prediction Using Models Informed by Chromatin conformation and Epigenomics (PUMICE), using TOPMed lung RNA-seq data from the Lung Tissue Research Consortium (LTRC), which included COPD patients undergoing lung surgery. We compared their performance with two published models built on RNA-seq from GTEx Lung tissue, GTEx-EN and GTEx-PUMICE. We first integrated each of four transcriptome models with multi-ancestry GWAS of FEV1/FVC ratio (Shrine et al. 2023) to generate transcriptome-wide association study (TWAS) results, which produced trait-associated genes. Pathway analysis was then conducted on significant TWAS genes. Finally, the PTRS was computed on TWAS genes that were included in the top 20 significantly enriched pathways. We tested performance of PTRS in race-stratified analysis for moderate-to-severe and severe COPD in COPDGene. RESULTS: Our models produced more significant TWAS genes (FDR&lt;0.05) and had more overlaps with genes identified for FEV1/FVC ratio from Shrine et al. 2023 compared to two GTEx models (Fig. A). Additionally, our models had significantly higher AUC for predicting both moderate-to-severe and severe COPD in race-stratified analysis. The mean AUC from LTRC-EN for predicting severe COPD was 0.57 and 0.55 respectively for EUR and AFA, which was significantly higher than mean AUC of 0.54 and 0.51 from GTEx-EN (Delong p-value=8.15x10-10 for EUR and 1.42x10-3 for AFA) (Fig. B). Similarly, the PTRS derived from our models produced higher Odds Ratios (OR) per standard deviation. The mean OR from LTRC-PUMICE for moderate-to-severe COPD was 1.25 and 1.18 respectively for EUR and AFA, while it was 1.17 and 1.12 from GTEx-PUMICE (Fig. C). CONCLUSIONS: Our study demonstrates the value of leveraging disease-relevant RNA-seq to construct transcriptome prediction models and their improvement on identifying lung function genes and predicting COPD across ancestries.