Publications

PURPOSE: This study investigated the effect of bone metastasis on the biomechanical environment of human vertebrae in patients with metastatic spine disease through the metric of load-to-strength ratio (LSR). Specifically, we compared the patients' LSRs to age and sex-similar noncancer controls from the Framingham Heart Study.

METHODS: Derived from clinical CT data of 135 metastatic spine disease patients planned for radiotherapy and 246 normative controls from the Framingham Heart Study, individualized spinal musculoskeletal models and vertebral strength estimates were used to compute level-specific LSR under natural standing and three weight-holding conditions (standing + weight, flexion + weight, and lateral bending + weight).

RESULTS: Adjusted for age, BMI, and spinal region, osteosclerotic and mixed lesion vertebrae had higher strength than osteolytic and control vertebrae. The musculoskeletal models suggested breast, prostate, and male lung cancer patients had higher compressive vertebral loading, and female lung cancer patients had lower compressive vertebral loading than controls. Male patients had higher standardized LSRs in natural standing, while female patients had lower LSRs for all activities than controls. Independent of sex, vertebrae with osteosclerotic and mixed bone metastasis had lower LSRs than controls, while, for osteolytic bone lesions, males had higher and females lower LSRs than controls. Vertebrae with no observed lesion on CT had higher LSRs than controls in males and lower LSRs in females.

DISCUSSION: Our findings highlighted that primary cancer and lesion type differentially affected task-specific vertebral loading and strength, thus modifying the vertebral LSRs. Sex-mediated differences in LSRs between FHS controls and vertebrae with no observed metastatic lesions suggest that considering the latter as "normal" should be taken with care. Our initial assessment supports further examination of whether vertebral LSR measurements are associated with vertebral risk and, if so, what threshold values indicate risk.

LEVEL OF EVIDENCE: 3.

OBJECTIVE: This study aims to assess the diagnostic value of the coronary artery calcium score (CACS) to predict significant coronary artery stenosis (SCAS) in presolid organ transplant patients with a moderate and high risk of coronary artery disease.

METHODS: In this retrospective HIPAA-compliant study, all pre-liver/kidney transplant patients with intermediate or high risk of coronary artery disease who underwent cardiac CT angiogram (CCTA) and CACS between January 1, 2018 and December 31, 2022 were reviewed. CACS was assessed according to the Agatston score. SCAS was defined as ≥50% diameter stenosis on CCTA. The potential to predict SCAS was assessed by computing the area under the receiver curve (AUC), and the sensitivity and specificity of CACS in diagnosing SCAS were calculated at various CACS thresholds.

RESULTS: A total of 291 patients (81; 28% female) with a mean age of 57.7±9.9 years, were included. The median CACS was 772 (IQR: 320 to 1892) in patients with SCAS versus 23 (IQR: 0 to 187) in those without SCAS (P<0.01). CACS demonstrated efficacy in predicting SCAS, with an AUC of 0.88. In the subgroup analysis, mean CACS differed significantly between pre-liver and pre-kidney solid organ transplant patients, but the difference between AUC curves was not significant. With the use of traditional CACS thresholds, the sensitivity dropped below 95%, but with optimized thresholds of CACS ≤62 or ≥869 for 95% sensitivity and specificity, 63% of pretransplant patients could be attributed either to a low-risk or high-risk for SCAS, respectively.

CONCLUSIONS: Our study shows that CACS can accurately predict SCAS in a large proportion of pre-kidney and liver transplant patients with intermediate or high risk of coronary artery disease. CACS can be considered as a screening tool to identify patients with low likelihood of SCAS, who might not require CCTA, as well as those with a high likelihood in whom a proactive approach should be considered.

BACKGROUND: Preliminary work suggests that glymphatic transit is abnormal and dynamic in idiopathic intracranial hypertension (IIH), although its incomplete characterization across the disease course has led to debate as to its pathophysiologic relevance. We seek to clarify whether glymphatic transit varies across IIH's disease course, correlates with intracranial pressure, and might be measured radiographically to aid diagnosis.

METHODS: Diffusion tensor imaging along the perivascular space (ALPS) was used to measure perivascular diffusivity and generate indices of glymphatic transit (ALPS-indices). We studied healthy controls and participants with IIH cross-sectionally, the latter stratified into untreated, treated, or cured IIH. Participants with untreated IIH were substratified as "acute" if imaged <6 months from onset or "chronic" if imaged ≥6 months from onset.

RESULTS: Forty participants were studied. We identified a positive, nonlinear correlation between ALPS-indices and disease duration in patients with untreated IIH. ALPS-indices differed between participants with chronic, treated, and acute IIH, in descending order. Healthy controls exhibited lower ALPS-indices than participants with chronic IIH and higher ALPS-indices than participants with acute IIH. ALPS-indices correlated positively with lumbar puncture opening pressures in participants with chronic IIH. Receiver-operating-characteristic curves demonstrated high areas-under-the curve in distinguishing between participants with untreated IIH and controls.

CONCLUSIONS: These data support the hypothesis that changes in glymphatic transit are likely a result rather than a cause of IIH, that radiographic indices of glymphatic transit may be leveraged diagnostically, and that ALPS-indices measuring perivascular diffusivity are likely a physiologically valid reflection of glymphatic transit in humans.

BACKGROUND: Information derived from lumbar puncture (LP) remains fundamental in the diagnosis of idiopathic intracranial hypertension (IIH). However, almost all patients with IIH are obese, rendering LPs technically difficult. The number of attempts required to achieve success and the likelihood of post-LP headache due to cerebrospinal fluid (CSF) leak has not been well characterized in this specific patient population or by LP modality.

METHODS: We conducted a retrospective chart review of IIH patients seen at a single neuro-ophthalmology clinic who underwent LP, either guided by conventional anatomic landmarks or via fluoroscopy. Chi-squared analysis was performed to assess differences in the likelihood of post-LP CSF leak by LP method. Unpaired t-tests were performed to assess between differences in the number of attempts required, BMI, and opening pressure.

RESULTS: 49 patients underwent 74 LPs. 31 (42%) were performed at the bedside, while 43 (58%) were performed fluoroscopically. Incidence of post-LP CSF leak was 14% in patients who underwent fluoroscopically-guided LPs and 45% in patients receiving bedside LPs (p-value = .0029). 8 bedside LPs (26%) required multiple attempts to complete, while only 1 fluoroscopically guided procedure required more than 1 attempt (2%).

CONCLUSION: Fluoroscopically-guided LPs required fewer attempts and were less likely to be complicated by CSF leak compared to bedside LPs. Body mass index and opening pressure were not associated with the risk of a post-LP CSF leak. The information derived from this study may help to avoid excess time, cost, and morbidity in the evaluation of patients with suspected IIH.

PURPOSE: To develop expert consensus recommendations for patient selection and procedural techniques in cryoneurolysis for chronic pain management using a Delphi process.

MATERIALS AND METHODS: A panel of 22 international interventionists participated in a two-round Delphi process. Participants rated 42 statements on a 10-point Likert scale (1 = strongly disagree, 10 = strongly agree). Consensus was predefined as ≥ 75% of ratings ≥ 7. Descriptive statistics (n, mean ± SD, % ≥ 7) were calculated.

RESULTS: High agreement supported cryoneurolysis for localized chronic pain refractory to conservative therapies (mean 8.55 ± 0.60; 100% ≥ 7) and cases with identifiable peripheral nerve targets (8.45 ± 0.74; 100% ≥ 7). Imaging guidance was deemed essential for nerve identification and probe placement (8.91 ± 0.29; 100% ≥ 7). The panel endorsed individualized freeze-thaw cycles to achieve Sunderland II injury (8.27 ± 0.83; 91% ≥ 7) and emphasized thorough patient education (9.0 ± 0.0; 100% ≥ 7). Experts recommended repeat cryoneurolysis within weeks if initial response was incomplete (8.64 ± 0.65; 100% ≥ 7) and alternative therapies after two unsuccessful sessions (8.36 ± 0.73; 95% ≥ 7). No consensus was reached on restricting treatment to a single anatomical region (5.82 ± 2.91; 50% ≥ 7) or routine prophylactic antibiotics (6.27 ± 2.57; 54% ≥ 7).

CONCLUSION: This Delphi study establishes expert-derived consensus standards for cryoneurolysis, highlighting careful patient selection, mandatory imaging guidance, and flexible freeze protocols while identifying areas requiring further research.

In 2005, three emergent strategic themes for academic radiology were identified to help deal with growing issues in the clinical, research, and educational missions. However, over the past two decades, these emergent strategic themes have actually evolved at different speeds with different levels of adoption by academic radiology departments. By considering these strategic themes as analogs to innovations, we have been able to apply concepts from the study of the diffusion of innovations and demonstrate the relevant "S-curves" so as to analyze the evolution of these strategic themes and their likely future courses. We have found that the emergent strategic theme of "throughput" has been fully adopted among academic radiology departments throughout the 2005 to 2025 period and is in the mature phase-at least with respect to the conventional approach of utilizing human radiologists without AI assistance to perform the key tasks of interpreting and reporting imaging studies. On the other hand, the emergent strategic themes of "teams" and "self-directed learning" for the research and educational missions, respectively, have been adopted only by a subset of academic radiologists and radiology departments during this period, with particularly low rates of adoption in the 2005 to 2015 period. Future development of each of these emergent strategic themes will be significantly impacted by the parallel development, introduction, and evolution of AI within academic radiology in the years to come.

Multiple barriers have been identified to developing a learning health care systems (LHSs) including organizational culture, data systems and interoperability, funding and workforce limitations, and regulatory challenges. Artificial intelligence (AI) is being explored both inside and outside of health care, with varying degrees of scientific rigor in the testing of AI applications. LHSs and AI face similar implementation challenges, which presents an opportunity for synergy. By reviewing AI use cases from the lens of how it can be used to reduce previously identified barriers to progressing toward an LHS, opportunities for facilitating this journey can be identified. AI tools can impact both clinical and nonclinical business processes. The process of testing and implementing AI tools based on high-quality evidence or signal should prespecify thresholds and expectations of incremental effectiveness (marginal risk-benefit) improvement compared with current standards of care, as is standard in health services research, quality improvement, process improvement, and best practices of comparative health care research. Business process examples to improve workflow using AI tools may adhere to less rigorous evidentiary standards compared with tools guiding patient-centered clinical decision scenarios, such as with AI-based diagnostic applications. This review indicates that AI tools provide tremendous opportunities for radiology to improve health care systems, workflow processes, and patients' health outcomes.

PURPOSE: To evaluate the impact of capsular tension ring (CTR) implantation on refractive outcomes, intraocular lens (IOL) stability, and postoperative complications in high myopic patients undergoing cataract surgery.

DESIGN: Systematic review and meta-analysis.

METHODS: A systematic search of PubMed, EMBASE, Cochrane Library, and Google Scholar through March 2025 identified studies comparing cataract surgery with and without CTR implantation in high myopic eyes. The primary outcome was prediction refractive error (PE). Secondary outcomes included absolute refractive error (AE), postoperative corrected distance visual acuity (CDVA), IOL decentration, anterior chamber depth (ACD), posterior capsular opacification (PCO), and Nd:YAG capsulotomy rates. Random-effects meta-analysis, sensitivity analysis, and meta-regression were performed.

RESULTS: Nine studies (4 randomized controlled trials and 5 cohort studies; 846 eyes) were included. CTR implantation did not significantly affect PE (mean difference [MD], 0.00; 95% confidence interval [CI], -0.07 to 0.08; P = .93) or postoperative CDVA (MD, 0.00; 95% CI, -0.06 to 0.06; P = 1.00). However, CTR significantly reduced AE (MD, -0.12; 95% CI, -0.20 to -0.05; P = .001) and IOL decentration (MD, -0.04; 95% CI, -0.07 to -0.01; P = .004).No covariates significantly influenced PE in meta-regression.

CONCLUSIONS: CTR implantation appears to modestly improve refractive predictability and IOL stability in high myopic eyes without compromising postoperative visual acuity. However, given the limited evidence base and small effect sizes, these findings should be interpreted with caution. CTR implantation may be considered in selected myopic cases with suspected zonular weakness or large capsular bags where additional capsular stability is desired.

Extending our validated benchmarking work, GPT-5 showed no improvement in sociodemographic-linked decision variation compared with GPT-4o and seemed to be worse on several endpoints. We re-tested GPT-5 with a fixed pipeline: 500 physician-validated emergency vignettes, each replayed across 32 sociodemographic labels plus an unlabeled control, answering the same four questions (triage, further testing, treatment level, and need for mental-health assessment). This design holds clinical content constant to isolate the effect of the label. GPT-5 reproduced subgroup-linked variation, with higher assigned urgency and less advanced testing for several historically marginalized and intersectional groups. Notably, several LGBTQIA+ labels were flagged for mental-health screening in 100% of cases, versus  41-73% for comparable groups with GPT-4o. Additionally, in an adversarial re-run that inserted one fabricated medical detail into otherwise standard clinical cases, GPT-5 adopted or elaborated on the fabrication in 65% of runs (vs 53% for GPT-4o). A single mitigation prompt reduced this to 7.67%.