Citations

Abstract Cutaneous squamous cell carcinoma (cSCC) poses significant clinical challenges due to its rising incidence and potential for metastasis. Histopathologic risk stratification is further limited by substantial inter-observer variability. Unsupervised AI approaches based on content-based image retrieval offer scalable and interpretable decision support for diagnostic pathology. The objective of this study was to evaluate the use of image retrieval within histopathology atlases to stratify cSCC tumor differentiation from whole-slide images (WSIs), while comparing different patch selection and feature extraction strategies. This retrospective study included 552 archived WSIs comprising 385 well-differentiated, 102 moderately differentiated, and 66 poorly differentiated cases collected across Mayo Clinic sites in Arizona, Florida, and Minnesota. Image atlases were constructed using multiple patch aggregation strategies (Mosaic, Collage, and Montage) and deep learning encoders (KimiaNet, PathDino, and H-Optimus-0). A leave-one-WSI-out evaluation framework was used to assess differentiation classification performance using accuracy, specificity, sensitivity, and F1 score. Mosaic combined with KimiaNet achieved the highest Top-1 accuracy (74.9%) and specificity (92.6%), while Mosaic with H-Optimus-0 yielded the best Top-5 accuracy (79.0%) and macro-F1 score (62.6%). Collage combined with KimiaNet produced the highest Top-5 specificity (99.5%). The generalizability of the evaluated AI models varied across hospitals, reflecting differences in imaging protocols, staining practices, and patient populations. Overall, unsupervised image search and retrieval provides effective, annotation-free support for cSCC differentiation and has the potential to enhance dermatopathology workflows when appropriate combinations of patch selection and feature ex-traction methods are employed.

Abstract Background OnabotulinumtoxinA (BoNT‐A) is an established preventive treatment for chronic migraine but involves 31 to 40 injections per session, often causing discomfort and post‐procedural headaches. Remote electrical neuromodulation (REN) is a noninvasive therapy with efficacy in migraine treatment via conditioned pain modulation but has not been evaluated for procedural pain. This study evaluated REN's effectiveness in reducing acute procedural pain and postprocedural headache associated with BoNT‐A injections. Methods This was an investigator‐initiated single‐center, randomized, single‐blind, sham‐controlled crossover study enrolled 80 adults (aged 22 to 74 years) with chronic migraine undergoing routine BoNT‐A treatment. Each participant received one injection session without a device, followed by sessions using active REN and sham in randomized order. REN was applied to the upper arm 10 min prior to injections and removed after injection completion. Pain intensity was measured using a visual analog scale (0 to 100) at pre‐procedure, intra‐procedure, and post‐procedure time points. The primary outcome was procedural pain intensity, and secondary outcomes included post‐procedural headache incidence and adverse events. Due to clear benefit, the study was terminated early based on predefined stopping criteria. Results Final analysis of 60 participants (mean age 48.0 years; 49/60, 82% female) demonstrated that pre‐procedural pain levels were not significantly different between baseline and the active REN or sham ( p > 0.999 and p = 0.485, respectively). However, during and after BoNT‐A administration, the active REN group reported significantly lower pain scores compared to both the sham and baseline conditions. At intra‐procedure, the REN group experienced a mean pain reduction of 15.0 points ( p < 0.001), and at post‐procedure experienced a 19.1‐point reduction ( p < 0.001). Sham treatment did not result in significant pain reduction compared to baseline ( p > 0.999 for both intra‐procedure and post‐procedure). Additionally, REN lowered the incidence of headache as an adverse event, with only 15% (8/52) of participants experiencing post‐procedural headache compared to 55% (29/53) in the sham group and 39% (23/59) at baseline (odds ratio = 0.28, 95% confidence interval: 0.10 to 0.69, p = 0.008). No additional adverse events were reported. Conclusions REN significantly reduces procedural pain and post‐procedural headache associated with BoNT‐A injections for chronic migraine and may serve as a noninvasive, easily implemented pain management strategy for acute procedural pain. REN represents a promising approach to improving patient comfort during routine migraine treatment as well as reducing post‐procedural headache.

Background: We aimed to evaluate the outcomes of patients with confirmed and possible cardiac sarcoidosis who had an ICD implant and identify predictors of ventricular arrhythmias and mortality in this patient population. Methods: We performed a retrospective study patients with a diagnosis of cardiac sarcoidosis (CS) at Mayo Clinic and had an ICD implanted for primary or secondary prevention indication. Adult patients (≥18 years) with a diagnosis of possible or confirmed (biopsy-proven) CS with an ICD implanted between 1993 and 2022 were included. Primary outcomes included sustained ventricular arrhythmia (VT) resulting in a documented appropriate ICD shock and death from any cause. Secondary outcomes included ICD-related complications. Results: We identified 148 patients, in which 23 patients (15.5%) experienced VT events with a total of 160 treated episodes. 4 patients (2.7%) experienced inappropriate ICD shocks (11 episodes). During a median follow up of 3.1 years, 26 patients deceased, and the overall survival was 98% and 77% at 1 and 5 year. Older age, CAD, CKD, paced rhythm and QRS duration were associated with worse survival (Figure 1 – 4) . Device related complications were rare, except for the higher percentage of CHF in patients with VT after implant, however this was not statistically significant. Patients with ICD therapies were younger (53 vs 58 years, p=0.029), had higher LGE burden (26.9 vs 16.3%, p=0.01), and higher percentage of LV mass (26.9 vs 16.3, p=0.011). Multivariate logistic regression model which showed that patients with biopsy proven sarcoidosis had significantly higher odds of VT events (OR: 3.04, 95% CI: 1.02 to 9.08, p = 0.046). In patients with increased LV mass (every 10 g increase), the odds of developing VT increased by 1.49 (95% CI: 1.02 to 2.19, p = 0.041). LV mass and LGE percentage showed fair ability of predicting the outcome by ROC analysis. Youden’s index showed 26 for LV mass and 16% for LGE percent as the optimal cut-off points for predicting VT outcomes (Figure 5) . The sensitivity, specificity, positive predictive value, negative predictive value were 81.8%, 74.1%, 37.5%, 95.6% and 75.4% respectively for an LV mass of 26 grams and 81.8%, 63.8%, 30.0%, 94.9% and 66.7% respectively for an LGE percentage of 16%. Conclusion: LGE burden continues to be a major predictor for VT events in patients with CS. Myocardial tissue burden and electrical abnormalities from CS predict mortality than myocardial function loss.

Normothermic machine perfusion (NMP) enables preservation and functional assessment of liver allografts, improving outcomes, particularly in marginal organs. However, the TransMedics Organ Care System (OCS) permits only single hepatic artery (HA) cannulation, limiting its use in grafts with variant arterial anatomy. Arterial reconstruction before NMP may overcome this limitation and expand the pool of grafts eligible for perfusion. We conducted a retrospective single-center cohort study of adult liver transplants utilizing OCS from January 2022 to September 2024. Allografts with variant hepatic artery anatomy requiring arterial reconstruction before NMP initiation were identified. Postoperative outcomes, including vascular and biliary complications, were compared between reconstructed and non-reconstructed grafts. Among 541 OCS-preserved grafts, 74 (13.7%) underwent arterial reconstruction. Compared with non-reconstructed grafts, reconstructed grafts had longer cross-clamp to cannulation time (2.8 vs. 2.4 h, p <0.001), but similar rates of early allograft dysfunction (23.0% vs. 20.8%, p =0.65), ischemic cholangiopathy (4.1% vs. 4.1%, p =1.00), and primary nonfunction (0.0% vs. 0.4%, p =1.00). Early (2.7% vs. 0.2%, p =0.05) and late hepatic artery thrombosis (HAT) (4.1% vs. 2.1%, p =0.40) were infrequent and not significantly different between non-reconstructed and reconstructed groups, respectively. Hepatic artery stenosis occurred more often in reconstructed grafts (12.2% vs. 4.1%, p =0.008), though most cases were managed with interventional radiology or surgical revision. Only one reconstructed graft required retransplantation due to vascular complications. Arterial reconstruction before NMP is safe and feasible, enabling successful perfusion and transplantation of liver allografts with variant HA anatomy. These findings support the broader application of NMP in anatomically complex donor livers.

Comparative analysis of fibrin sealant and recombinant thrombin hydrogels reveals distinct microarchitectures and cellular responses, emphasizing recombinant thrombin’s potential to enhance cellular activity and enable vesicle delivery.

Objectives/Goals: Aspiration causes or aggravates lung diseases. While bedside swallow evaluations are not sensitive/specific, gold standard tests for aspiration are invasive, uncomfortable, expose patients to radiation, and are resource intensive. We propose the development and validation of an AI model that analyzes voice to noninvasively predict aspiration. Methods/Study Population: Retrospectively recorded [i] phonations from 163 unique ENT patients were analyzed for acoustic features including jitter, shimmer, harmonic to noise ratio (HNR), etc. Patients were classified into three groups: aspirators (Penetration-Aspiration Scale, PAS 6–8), probable (PAS 3–5), and non-aspirators (PAS 1–2) based on video fluoroscopic swallow (VFSS) findings. Multivariate analysis evaluated patient demographics, history of head and neck surgery, radiation, neurological illness, obstructive sleep apnea, esophageal disease, body mass index, and vocal cord dysfunction. Supervised machine learning using five folds cross-validated neural additive network modelling (NAM) was performed on the phonations of aspirator versus non-aspirators. The model was then validated using an independent, external database. Results/Anticipated Results: Aspirators were found to have quantifiably worse quality of sound with higher jitter and shimmer but lower harmonics noise ratio. NAM modeling classified aspirators and non-aspirators as distinct groups (aspirator NAM risk score 0.528+0.2478 (mean + std) vs. non-aspirator (control) risk score of 0.252+0.241 (mean + std); p Discussion/Significance of Impact: We report the use of voice as a novel, noninvasive biomarker to detect aspiration risk using machine learning techniques. This tool has the potential to be used for the safe and early detection of aspiration in a variety of clinical settings including intensive care units, wards, outpatient clinics, and remote monitoring.

Background Neuropsychiatric symptoms (NPS) are frequent in Alzheimer's disease (AD) dementia, but a higher NPS burden is found in dementia with Lewy bodies (DLB). Lewy body (LB) pathology frequently co-occurs with AD pathology and may not meet neuropathological criteria for DLB (ADLB). NPS trajectories over disease course in these subgroups is not well understood. Objective We investigated changes in NPS severity over time, at two time points, comparing clinicopathologically defined cohorts of AD (without LB), ADLB, DLB, and controls. Methods Cases with two available Neuropsychiatric Inventory-Questionnaire (NPIQ), at the time of enrollment and within 2.5 years of death, were selected from the Arizona Study of Aging and Neurodegenerative Disorders. Differences and rate of change in NPIQ scores were compared between AD (n = 75), ADLB (n = 48) DLB (n = 65), and controls (n = 32) with covariates for age, sex, and cognition. Results First NPIQ scores were highest in ADLB when compared to AD (p = 0.04) and controls (p = 0.01) but not different from DLB. A significant increase in NPS severity was observed in DLB and AD (p < 0.001) over a mean follow up time of 4.9 ± 3.0 years, and the rate of change was significantly greater in DLB when compared to other groups. Final NPIQ scores were highest in DLB when compared to AD (p = 0.03) but not ADLB, and in DLB, ADLB, and AD than controls ( all p < 0.001). Conclusions Early NPS burden as well as NPS severity progression rate, independently of cognitive status, might be useful clinical metrics and may help predict underlying pathological diagnoses.