Zeroth-order (ZO) optimization enables dimension-free communication in federated learning (FL), making it attractive for fine-tuning of large language models (LLMs) due to significant communication savings. However, existing ZO-FL methods largely overlook curvature information, despite its well-established benefits for convergence acceleration. To address this, we propose HiSo, a Hessian-informed ZO federated optimization method that accelerates convergence by leveraging global diagonal Hessian approximations, while strictly preserving scalar-only communication without transmitting any second-order information. Theoretically, for non-convex functions, we show that HiSo can achieve an accelerated convergence rate that is independent of the Lipschitz constant L and model dimension d under some Hessian approximation assumptions, offering a plausible explanation for the observed phenomenon of ZO convergence being much faster than its worst-case 𝒪 ( d ) -bound. Empirically, across diverse LLM fine-tuning benchmarks, HiSo delivers a 1~5× speedup in communication rounds over existing state-of-the-art ZO-FL baselines. This superior convergence not only cuts communication costs but also provides strong empirical evidence that Hessian information acts as an effective accelerator in federated ZO optimization settings.
Does the time elapsed between planned oocyte cryopreservation cycles affect oocyte yield at the second retrieval? This retrospective cohort study included 2412 patients undergoing at least two oocyte retrievals within 180 days for elective oocyte cryopreservation between 2010 and 2022. The primary outcome was mature oocyte yield in the second cycle. Secondary outcomes included changes in total oocyte yield, stimulation days, and gonadotrophin dose (IU) required in the second cycle, and frequency of cancellation of the second cycle. Patients were categorized by time elapsed between stimulation cycles: immediate second cycle (≤28 days, n = 399), standard timing (29-90 days, n = 1448), or delayed second cycle (91-180 days, n = 565). After adjusting for age, mature oocyte yield at first stimulation, protocol, and trigger medication, mature oocyte yield at the second retrieval was 9% lower for the immediate group (rate ratio 0.91, 95% CI 0.86-0.96) and 7% lower for the delayed group (rate ratio 0.93, 95% CI 0.89-0.98) compared with standard timing. These differences did not meet the prespecified criteria for inferiority, defined as a 20% decrease in oocyte yield. The number of days of stimulation did not differ by interval group. Total gonadotrophin dose was significantly higher for patients who underwent their second cycle immediately compared with those who underwent their second cycle with standard timing (mean difference 187 IU, 95% CI 35-339). Compared with waiting 1-3 months before the second stimulation for planned oocyte cryopreservation, shorter or longer intervals (up to 180 days) are non-inferior, although associated with a small decrease in oocyte yield. The clinical relevance of these findings may vary depending on patient scenario.
A relatively small number of rapidly spreading wildfire events burn disproportionately large areas. But beyond area burned, do fast fires produce distinct postfire ecological outcomes? Here, we measure daily linear fire growth rates and assess relationships between fire spread, burn severity, and metrics of postfire landscape resilience from 3499 fires occurring between 2012 and 2023 in conifer-dominated forests of Canada and the western US. We find that as linear spread rate increases, so too does fire severity, proportion of area burned at high severity, and distance to surviving live tree seed sources. Thus, not only do the fastest fires burn outsized areas, but these areas have less remaining tree cover and reduced capacity for recovery, setting the stage for long-term vegetation shifts. Given anticipated increases in extreme fire spread as fire seasons lengthen and weather becomes more fire-conducive, our findings highlight growing needs to effectively reduce undesired impacts, manage postfire landscapes to sustain forest ecosystem functions, and foster social-ecological adaptation.
BackgroundAmyloid-negative tau-related neurodegeneration represents a non-Alzheimer biomarker-defined profile; however, its clinical heterogeneity and prognostic relevance remain unclear within Alzheimer's disease-related frameworks.ObjectiveTo characterize the clinical spectrum and identify predictors of mortality in patients with this profile.MethodsIn this retrospective cohort study, 1280 patients evaluated at a tertiary neurology center were screened, and 130 with an amyloid-negative cerebrospinal fluid (CSF) pattern [amyloid-β (Aβ)42 normal, phosphorylated tau (pTau), and total tau (tTau) elevated] were included. Survival was assessed using Kaplan-Meier analysis, and predictors of mortality were evaluated using Cox models.ResultsThe cohort (mean age 68.8 ± 10.1 years; 45.4% female) showed heterogeneous diagnoses, mainly mild cognitive impairment and frontotemporal dementia (each 30%). Survival differed across diagnostic groups (log-rank p = 0.037), with more favorable outcomes in mild cognitive impairment. Male sex was more frequent among non-survivors (88.9% versus 45.6%, p < 0.001). Higher CSF tTau levels were associated with mortality in the joint Cox model (HR 1.003, 95% CI 1.001-1.006, p = 0.006) and faster clinical progression (r = 0.22, p = 0.011). When modeled separately, neither tTau nor pTau was independently associated with mortality; however, both became significant in opposite directions when included jointly.ConclusionsThe amyloid-negative A-T + N + profile represents a clinically heterogeneous subgroup with prognostic relevance. CSF tTau was associated with mortality and faster clinical progression, but the joint-model findings involving tTau and pTau should be interpreted cautiously as hypothesis-generating. Further studies are needed to clarify the prognostic value of tau-related biomarkers in this population.
Diabetic ketoacidosis (DKA) at type 1 diabetes diagnosis is common in low-resource settings and associated with prolonged hospitalisation. We aimed to identify determinants of time to discharge among children with new-onset DKA in Zambia. We conducted a retrospective cohort study of children aged 0-16 years with newly diagnosed T1DM and DKA at the University Teaching Hospital, Lusaka (2018-2023). The primary outcome was time to discharge (days). We used Kaplan-Meier with log-rank tests and Cox regression. To address proportional hazards (PH) violation, we fitted a stratified Cox model (stratified by HbA1c, vomiting and DKA severity) for PH validation and a granulated Cox model with DKA severity (mild, moderate, severe) to estimate independent effects. PH was verified using Schoenfeld residuals. Among 353 children (mean age 8.3 years, SD 4.6), median stay was 5 days (IQR 4-8). Log-rank tests showed significant differences in discharge probability by DKA severity (χ2 = 70.32, p < 0.001) and vomiting (χ2 = 30.35, p < 0.001). In the granulated model, DKA severity was the strongest predictor of delayed discharge (moderate: HR = 0.01, p < 0.001; severe: HR = 0.01, p < 0.001), while infection was associated with faster discharge (HR = 1.51, p < 0.001). Ketonuria lost significance after DKA adjustment (HR = 0.90, p = 0.770), indicating a marker not an independent predictor. The stratified model confirmed PH satisfaction (global test χ2 = 5.62, df = 7, p = 0.584). Age, sex, polyuria, family history and residence were not significant. DKA severity is the strongest independent predictor of prolonged hospital stay in children with new-onset DKA. Ketonuria is a useful bedside marker but lacks independent effect beyond severity. Infection independently predicts faster discharge. These findings could inform risk stratification and resource allocation in similar settings.
Conformational dynamics in toxin inhibitors are an important contributor to ion channel affinity, yet toxin multidimensional energy landscapes remain largely unexplored. In the current work, we combine parallel-bias metadynamics-metainference (PBMetaD) simulations with relaxation dispersion NMR to define, at atomistic resolution, the thermodynamics and kinetics of Hui1, a de novo three disulfide toxin derived from the SAK-I family that targets K+-channels. Using the three χ3 disulfide dihedrals as collective variables, an extensive 48-replica well-tempered PBmetaD simulation (16.2 μs cumulative sampling) resulted in a fully converged three-dimensional (3D) free-energy surface comprising eight Hui1 conformers. These basins account for ∼96% of the bias-weighted ensemble and partition into four low- and four high-energy states separated by 7.5 kJ/mol associated with the (-) and (+)Cys12-Cys28 χ3 states, respectively. Transition-state theory identifies rota-isomerization of Cys3-Cys35 as the slowest, and therefore rate-determining, coordinate, while the analogous motions around Cys12-Cys28 and Cys17-Cys32 are ∼5-fold faster. 15N R1ρ relaxation dispersion NMR measurements confirmed these kinetics, identifying two structurally distinct residue clusters exhibiting intermediate and faster exchange processes. The Key Interaction Finder (KIF) approach reveals that Cys17-Cys32 conformation modifies connectivities within a dense interaction network between Cys17 and residues Gln14, Tyr23, Arg24, and Lys29, and correlates with the accessibility of residues Tyr23 and Arg24 of the helix-kink-helix region for interaction with the channel vestibule. Our work establishes PBMetaD as a powerful framework for mapping coupled disulfide and backbone dynamics in toxins and reveals specific conformers and interactions likely to control K+ channel recognition.
A key challenge in the development of materials for the next generation of solar cells, sensors and transistors is linking macroscopic device performance to underlying microscopic properties. For years, fabrication of devices has been faster than our ability to characterize them. This has led to a random walk of material development, with new materials being proposed faster than our understanding. We present two neural network-based methods for extracting key material parameters, including charge carrier mobility and trap state density, in optoelectronic devices such as solar cells. Our methods require only a single measured light current-voltage curve and modest computational resources, making our approach applicable in even minimally equipped laboratories. Unlike traditional machine learning models, our methods place the final material values in a non-Gaussian likelihood distribution, allowing confidence assessment of each predicted parameter.We demonstrate these techniques on freshly fabricated PM6:Y12 and PM6:BTP-eC9 organic solar cells, and then track a single PM6:BTP-eC9 device as it degrades in air, recovering the evolution of carrier lifetime, mobility and shunt resistance. This approach enables rapid, low-cost extraction of key material parameters from simple JV measurements alone, providing a practical route to accelerate optimisation of next-generation solar-energy materials and devices.
Kv3.1 channels are high-voltage-activated potassium channels that facilitate high-frequency firing in auditory brainstem neurons. Fusiform neurons in the dorsal cochlear nucleus (DCN) can fire rapid trains of action potentials and express Kv3.1 channels. Their ability to fire high-frequency action potential trains increases after hearing onset at postnatal day 14. Simultaneously, the action potentials become shorter and faster. We tested whether Kv3.1 channels drive the developmental maturation of action potentials in DCN fusiform neurons. We used Swiss female mice from postnatal day 8 to 25 and divided them into pre-hearing (< P14) and post-hearing (> P14). We measured Kv currents and action potentials in whole-cell patch-clamp. Tetraethylammonium (TEA) was applied at a concentration of 1-5 mM to block Kv3 channels, and we detected the expression of Kv3.1b channel subunits by immunocytochemistry. We found increased Kv3.1b subunit expression in the DCN after hearing onset, suggesting that this subunit could be responsible for the electrophysiological changes in post-hearing fusiform neurons. TEA-sensitive high-threshold currents were a significant component of voltage-dependent potassium currents in both pre- and post-hearing fusiform neurons, with similar magnitudes in both groups. However, blocking Kv3 currents with 1-5 mM TEA broadens action potential waveforms in neurons from both groups while maintaining the developmental differences between groups. On the other hand, TEA was more effective at reducing the firing of post-hearing neurons. We conclude that the rise in Kv3.1b channel expression in DCN fusiform neurons after hearing does not account for the shorter, faster action potentials observed at that time, but could contribute to their ability to fire action potentials at high frequencies.
Tuberculosis (TB) is a global health challenge and is curable, but tools for rapid treatment monitoring are limited. Current methods, like smear microscopy, lack sensitivity, while culture-based approaches require weeks for results and specialized biosafety facilities. This paper presents a comprehensive protocol for the PATHFAST TB LAM Ag assay, a rapid chemiluminescent enzyme immunoassay that quantifies lipoarabinomannan (LAM) mainly in sputum and likely other biological samples within one hour. This protocol details sample collection, heat inactivation for biosafety, and the automated measurement process. Heat inactivation at 100 °C for 20 minutes renders samples non-infectious while preserving LAM for detection. The assay demonstrates excellent analytical performance with a detection limit of 6.67 pg/mL and a linear measurement range of 10.0-50,000 pg/mL. Published clinical evaluation demonstrated 88.8% (95% CI: 80.0-94.0%) sensitivity and 100% (95% CI: 83.9-100%) specificity compared to culture, with strong correlation to bacterial load. The PATHFAST TB LAM Ag assay successfully tracks bacterial decline during treatment, with LAM concentration reductions mirroring culture results but available significantly faster. Unlike culture methods that suffer from contamination (4.2-9.6% invalid results), the PATHFAST TB LAM Ag assay consistently provides valid results. This protocol enables implementation of a rapid, quantitative TB treatment monitoring tool that enhances biosafety, reduces turnaround time, and maintains performance comparable to conventional methods, addressing critical needs in both clinical care and drug development settings.
Tetrazine-trans-cyclooctene (Tz-TCO) click chemistry has been widely adopted for various applications due to its faster reaction kinetics and higher biocompatibility compared to conventional copper-catalyzed click chemistry. However, most click chemistry groups do not appear to be fully bioorthogonal, as several of them-including tetrazine-have been found to directly modify proteins. Here, we report the unexpected discovery of non-specific protein modifications by TCO during the development of a tetrazine-based proximity labeling method. TCO probes readily conjugate to cysteine residues on proteins, and pre-treatment with iodoacetamide significantly reduces TCO-mediated background labeling. Chemoproteomic analyses reveal that proteins involved in translation are heavily modified by TCO, pointing to potential off-target effects when TCO probes are used in live cells. Our findings underscore the need for further improving the biocompatibility of click chemistry to enable more precise biological applications.
To investigate the clinical efficacy of laparoscopic total colectomy, ileal J-pouch, and ileorectal anastomosis (LTC + IRA) in the treatment of slow transit constipation (STC). The clinical data of six patients with STC who underwent LTC + IRA surgery from January to December 2023 were retrospectively analyzed. Before the operation, colonic transit experiment, fiber colonoscopy, defecography examinations, colography, and psychological evaluations were performed. The operation time, application of antidiarrheal drugs, and the occurrence of complications were collected. At the 3rd, 6th, and 12th months after surgery, the patients were followed up with the number of daily bowel movements, Wexner constipation score, anal incontinence score, gastrointestinal quality of life index score (GIQLI), abdominal pain frequency score, and abdominal distension frequency score. All patients successfully completed laparoscopic surgery, and they had no complications and were followed up for 1 year. With the extension of postoperative follow-up time, all patients' defecation and quality of life significantly improved compared with before surgery. The number of daily bowel movements at 3, 6, and 12 months after surgery was significantly higher than before surgery. At 12 months postoperatively, the bowel movement frequency was basically close to normal. Wexner's constipation score, anal incontinence score, GIQLI score, abdominal pain frequency score, and abdominal distension frequency score were significantly improved compared with before surgery. The quality of life of all patients had significantly improved, and they were satisfied. On the premise of strictly adhering to the indications for surgery, LTC + IRA were safe, effective, and feasible for the treatment of STC, which is less invasive and has faster recovery.
Management of shigellosis is increasingly complicated by antimicrobial resistance, yet susceptibility testing is rarely performed, and the role of broad-spectrum empiric therapy in hospitalized adults remains unclear. We conducted a multi-center retrospective cohort study of adults hospitalized with stool-confirmed Shigella spp. from October 2023 to August 2024 across five hospitals in the Portland metropolitan area. The primary outcome was time from antibiotic initiation to symptom resolution (less than 3 stools in 24 h) and the secondary outcome was length of hospital stay. Of 45 patients included in the outcome analyses, 29 (64%) received carbapenems and 16 (36%) received non-carbapenem antibiotics. Median time from antibiotic initiation to symptom resolution was 18 hours (IQR 9-37) in the carbapenem group and 29 hours (IQR 15-36) in the non-carbapenem group, with no significant difference between groups (P = .89). The median length of hospital stay was significantly longer in the carbapenem group, at 5.4 days (IQR 3.6-8.3), compared with 3.9 days (IQR 2.6-4.4) in the non-carbapenem group (P = .02). Only 26% of patients had susceptibility testing ordered. Carbapenem use was not associated with faster symptom resolution and was associated with significantly longer hospitalization. Susceptibility testing was underutilized and may be essential to guide local empiric and targeted therapy decisions.
Delayed wound healing remains a common clinical challenge and is associated with increased infection risk, prolonged hospitalization, and poor scar outcomes. Whether a multimodal strategy comprising early wound closure supported by continuous negative-pressure irrigation and drainage can improve healing dynamics compared with conventional open management remains insufficiently explored. This comparative clinical study included 62 patients with wounds who were admitted to Yanqing Hospital of Beijing Chinese Medicine Hospital between January and December 2023. Thirty-two patients received conventional open wound care (control group), and 30 patients underwent early primary closure with post-operative continuous negative-pressure irrigation and drainage (intervention group).Baseline characteristics, clinical outcomes, and wound healing trajectories were analyzed. Time to wound healing was evaluated using Kaplan-Meier analysis and Cox proportional hazards models. Multivariable logistic regression analyses were performed to assess wound healing within 14 and 28 days after adjustment for potential confounders. Patients in the intervention group exhibited a significantly shorter median wound healing time compared with controls (17.5 vs. 24.5 days, P = 0.003). The cumulative probability of wound healing was significantly higher in the intervention group(log-rank P = 0.003), and this association remained significant after multivariable adjustment (adjusted HR = 3.17, 95% CI: 1.66-6.05). Although no significant difference in healing within 14 days was observed after adjustment, the intervention was associated with a higher likelihood of complete wound healing within 28 days (adjusted OR = 11.63, 95% CI: 1.06-127.72). In addition, the intervention was associated with faster wound area reduction and better early wound cosmetic scores. A multimodal strategy consisting of early primary closure supported by continuous negative-pressure irrigation and drainage was associated with accelerated wound healing and improved early wound appearance. E-value analysis suggested that unmeasured confounding is unlikely to fully explain the observed association, though residual confounding from selection bias cannot be excluded.No increase in adverse events was observed. Due to the retrospective design and relatively small sample size, these findings are hypothesis-generating. Prospective randomized trials are required to establish definitive clinical recommendations.
Health care consolidation in the United States has yielded mixed outcomes for patients; however, little is known about the impact of integration between insurers and specialty pharmacies. Through improved data and clinical integration, integrated specialty pharmacy care models may improve health care access and reduce costs for beneficiaries. However, this type of vertical integration may allow increased market power, which could lead to higher costs for beneficiaries. Using Humana prescription claims, our cross-sectional study used adjusted regression analyses evaluating the association between the use of an integrated specialty pharmacy care model with pharmaceutical access and costs for beneficiaries managing complex specialty conditions in 2024. After adjusting for demographic and clinical characteristics, the integrated specialty pharmacy care model was associated with fewer days from prescription to fill (absolute difference: -5.7 days; 95% CI: -6.7 to -4.7 days) and, reduced yearly beneficiary pharmacy costs (-$106; 95% CI: -$163 to -$49) and yearly beneficiary costs for the specialty drug (-$77; 95% CI: -$129 to -$27) compared with nonintegrated specialty pharmacies. We found additional benefits for beneficiaries who also used integrated primary care services. Integrated specialty pharmacy care models may deliver faster and less costly pharmaceutical care for beneficiaries.
To develop and validate a hierarchical deep learning model for differentiating acute and chronic lumbar osteoporotic vertebral compression fractures (OVCFs) using X-ray images. We retrospectively reviewed approximately 2600 lateral lumbar radiographs obtained from patients clinically suspected of having OVCFs between 2007 and 2022. After excluding poor-quality images and surgically instrumented vertebrae, 1299 radiographs (6495 vertebral patches, L1-L5) were included. Labeling was performed by neurosurgeons and radiologists using X-ray images, with CT and/or MRI findings serving as the reference standard. A two-step hierarchical classification was implemented: first classifying vertebrae into Normal-Chronic, Acute, and Indeterminate (cement-augmented vertebrae without instrumentation) groups, followed by subdivision of the Normal-Chronic group into Normal and Chronic categories. A total of 1299 radiographs were evaluated. The hierarchical model achieved an accuracy of 91% in the initial three-class step. For the detection of acute fractures in the final classification step, the model demonstrated a sensitivity of 91.0% (95% CI 84.8-95.0%), a specificity of 82.1% (95% CI 79.8-84.5%), and a high negative predictive value (NPV) of 98.8% (95% CI 97.9-99.3%). The Normal-aligned hierarchical approach outperformed the Acute-aligned and end-to-end models, particularly for acute and chronic cases. The proposed hierarchical approach enhances the diagnostic utility of standard X-ray images by enabling more accurate classification of lumbar fracture types. This study is limited by its single-institution retrospective design. This model may reduce reliance on advanced imaging and support faster and more informed clinical decision-making.
Hybrid aortic repair (HAR) has emerged as a less invasive alternative to conventional open arch repair for selected patients with aortic arch disease, but its comparative safety and mid-term effectiveness remain uncertain. This study aimed to compare the clinical outcomes of HAR and the frozen elephant trunk (FET) procedure for the treatment of aortic arch disease. We conducted a retrospective analysis of 785 patients with aortic arch disease treated at Beijing Anzhen Hospital between January 2015 and February 2023. Among these patients, 406 underwent HAR, which consisted of supra-aortic debranching combined with thoracic endovascular aortic repair (TEVAR), and 379 received the FET procedure. Inverse probability of treatment weighting (IPTW) was used to eliminate baseline differences between the two groups, ensuring comparability of outcomes. The primary endpoints of the study were postoperative mortality, while the secondary endpoints included reintervention, perioperative stroke, paraplegia, and dialysis. Before IPTW adjustment, there was a significant difference in age between the two groups. Specifically, the patients in the HAR group were significantly older than those in the FET group [57.0 (49.0, 66.0) vs. 49.0 (40.0, 56.0) years, P<0.001]. The rates of diabetes (9.6% vs. 5.5%, P=0.045) and hyperlipidemia (21.7% vs. 9.0%, P<0.001) were also significantly higher in the HAR group than in the FET group. However, after adjusting for baseline differences using IPTW, none remained statistically significant. The pre-IPTW baseline data showed that, compared with the FET group, the HAR group had a significantly shorter operative time [5.5 (4.5, 6.5) vs. 6.5 (6.0, 7.0) minutes, P=0.02], lower blood loss [400.0 (150.0, 800.0) vs. 800.0 (600.0, 1,000.0) mL, P<0.001], and faster postoperative recovery [165.0 (75.0, 268.7) vs. 225.0 (145.0, 300.0) hours, P<0.001]. After IPTW adjustment, the HAR group continued to demonstrate significantly lower blood loss [500.0 (200.0, 900.0) vs. 800.0 (600.0, 1,000.0) mL, P=0.03]. No significant difference was observed between the two groups in terms of in-hospital mortality. The mid-term survival analysis revealed a higher mortality rate in the HAR group (8.6% vs. 3.9%, P=0.04); however, after IPTW adjustment, this difference was no longer statistically significant (P=0.18). This study found that, compared with the FET group, the HAR group demonstrated comparable postoperative mortality and reintervention rates, along with a shorter operative time, reduced blood loss, and a lower incidence of specific pulmonary complications, supporting its use in high-risk patients. However, HAR may increase the risk of neurological complications. Therefore, strict patient selection, careful anatomical assessment, and technical optimization are required to ensure long-term safety and efficacy.
Metabolic and alcohol-associated liver disease (MetALD) is an emerging phenotype within the steatotic liver disease spectrum, characterized by cardiometabolic risk factors coexisting with alcohol exposure, resulting in synergistic liver injury and fibrosis progression. Therapeutic development remains limited because most steatotic liver disease trials exclude patients with ongoing alcohol use. In contrast, alcohol-associated liver disease (ALD) trials have focused primarily on severe alcohol-associated hepatitis. Current management of patients with MetALD relies on an integrated approach that simultaneously controls alcohol use and cardiometabolic risk. In clinical practice, for patients with MetALD and ongoing alcohol use, therapies aimed at controlling alcohol use remain most critical, given the faster, more progressive disease course related to alcohol as compared to metabolic liver injury. Given the dynamic nature of alcohol intake and metabolic risk factors, longitudinal monitoring of disease stage with noninvasive fibrosis tests is essential. Liver-directed therapies with efficacy in metabolic dysfunction-associated steatotic liver disease (MASLD), including incretin-based agents, fibroblast growth factor 21 analogs, peroxisome proliferator-activated receptor agonists, and thyroid hormone receptor beta agonists, may benefit selected MetALD patients. Several agents may modulate both metabolic pathways and alcohol consumption through central reward mechanisms. Specific pharmacotherapies targeting alcohol use (acamprosate, naltrexone) combined with structured psychosocial interventions are effective in controlling alcohol use. Given the lack of dedicated clinical trials in MetALD patients, we synthesized data from clinical trials in MASLD and ALD. We propose adapting these data to inform the design of future clinical trials in patients with MetALD.
Chorioretinal atrophy is a leading cause of vision loss in highly myopic eyes. Myopic macular neovascularization (mMNV) may accelerate atrophy, but progression determinants remain unclear. In this retrospective study, atrophy area and perimeter were quantified at three time points: pre-MNV, at mMNV diagnosis, and final follow-up using ImageJ. Variables analyzed included macular staphyloma, membrane-atrophy contact, intravitreal injections, and best-corrected visual acuity (BCVA). Accelerated progression rate was defined as an annual increase exceeding twice the pre-mMNV atrophy. Univariate and multivariate analyses identified predictors of final atrophy. Of 225 eyes with high myopia (axial length ≥ 26 mm) and confirmed mMNV, 168 met inclusion criteria; mean follow-up was 4.96 ± 2.5 years. Atrophy area increased from 0.77 ± 1.92 mm² pre-MNV to 3.95 ± 7.41 mm² at diagnosis and 6.95 ± 9.76 mm² at final visit (p < 0.001). Area-based atrophy progression increased significantly after mMNV onset (0.25 ± 0.62 vs. 0.92 ± 1.96 mm²/year; p = 0.023), indicating greater absolute lesion enlargement over time. However, perimeter-adjusted progression rates, which normalize lesion growth for lesion size, did not differ significantly (0.50 ± 0.44 vs. 0.82 ± 0.43 mm²/mm/year; p = 0.40), suggesting that the increase in area mainly reflects larger lesion size rather than faster boundary expansion. Membrane-atrophy contact was associated with larger atrophy in univariate analysis (p < 0.001) but was not independent. Independent determinants were pre-MNV atrophy (β = 0.92; p < 0.001) and macular posterior staphyloma (β = 0.15; p = 0.009). Finally, final BCVA showed a moderate positive correlation with final atrophy area (r = 0.41; p < 0.001) and with foveal involvement (r = 0.326; p < 0.001). Atrophy progresses continuously in eyes with mMNV and accelerates after neovascular development, independent of anti-VEGF therapy. Pre-existing atrophy and macular posterior staphyloma determine final lesion size, while membrane-atrophy contact contributes to rapid progression.
X-ray micro-computed tomography (μCT) is widely used in biomedical research for non-destructive, high-resolution imaging. Synchrotron Radiation Phase-Contrast μCT (SR-PCI-μCT) further enhances image quality with higher Signal to Noise Ratio (SNR), improved contrast, and faster acquisition. However, segmenting SR-PCI-μCT images remains challenging due to their heterogeneous image property and limited training data. Additionally, the increasing throughput of synchrotron facilities demands robust, efficient segmentation methods. This paper reviews a list of recent segmentation approaches in biomedical μCT. Traditional methods remain simple and effective for simple, high-contrast structures but require extensive tuning and generalize poorly to complex, low-contrast soft tissues. Data-driven models provide higher accuracy and robustness yet rely heavily on large expert-annotated datasets, limiting reproducibility and cross-dataset adaptability. Recent advance on vision transformers have shifted the paradigm from task-specified to more domain-specified segmentation, though these techniques are still evolving and require adaptation for SR-PCI-μCT studies. This survey provides the first bi-modality review covering both laboratory and synchrotron biomedical μCT segmentation. It consolidates recent segmentation methodologies, identifies major trends in deep-learning techniques, and highlights current limitations across SR-PCI-μCT. Additionally, it outlines open challenges to guide future research and practical advancements in biomedical μCT segmentation.
The primary limitation for the application of alchemical free energy methods to a wider variety of complex molecular systems is achieving reasonable sampling. Flexible binding complexes often have high free energy barriers, which require prohibitively long simulations or carefully tuned enhanced sampling methods in order to gather sufficient uncorrelated samples to obtain reliable free energy estimates. An example of such a flexible system is the complex formed between FabB, an elongating β-ketoacyl-acyl carrier protein (ACP) synthase (KS) from Escherichia coli, and ACP, which carries acyl chains of varying lengths. Previous experimental evidence suggests that growing acyl chains can bind to at least two pockets in FabB. With the multiple topology replica exchange of expanded ensemble (MT-REXEE) enhanced sampling approach, we can obtain highly efficient sampling of both pockets by adaptively growing and shrinking the chains in the simulation ensemble, allowing each simulation to visit chain lengths that can transition between pockets. This approach enables unbiased sampling of alternate configurational states for large complex systems without prior pocket definitions, as collective-variable based enhanced sampling methods would require. Using the new swapping approach gives significantly enhanced sampling even for comparatively simple small molecule binding , as demonstrated by faster convergence of free energy estimates of relative binding affinity between mouse major urinary protein 1 ligands. This case study demonstrates the utility of MT-REXEE and its open-source implementation particularly for systems that feature high free energy barriers for a subset of ligands of interest, demonstrating a valuable addition to the existing stable of enhanced sampling methods.