Recent studies have applied machine learning (ML)-based limited sampling strategies (LSS) to predict drug exposure (AUC), achieving low prediction error and performance comparable to or better than multiple linear regression and population pharmacokinetics LSS. This study aimed to develop and validate a machine learning-based limited sampling strategy capable of predicting raltegravir (RAL) exposure. Four machine learning algorithms (XGBoost, Random Forest, GLMNet, and SVM) were trained using pharmacokinetic profiles generated via Monte Carlo simulation from a population pharmacokinetic (POPPK) model. Data were divided into training (75
Accurate size estimation of large (≥ 20 mm) colorectal laterally spreading tumors (LSTs) is essential for procedural planning, risk stratification, and predicting technical difficulty. Yet, the reliability of visual LST size assessment among endoscopists has not been systematically evaluated. 46 LSTs were recorded during colonoscopy. Twenty-four international expert endoscopists independently reviewed de-identified videos and provided visual estimates for (1) maximal diameter, (2) oral–anal axis, (3) left–right axis, and (4) percentage of colonic circumference involved. Each lesion was assessed twice in randomized order. Fleiss’s kappa, Krippendorff’s alpha, and intraclass correlation coefficients (ICC) were used to evaluate inter- and intra-rater agreement. A total of 1104 measurements were collected. Inter-endoscopist kappa agreement for maximal diameter was poor (κ = 0.16), with similarly poor agreement for the oral–anal (κ = 0.15) and left–right axes (κ = 0.14). The percentage of circumferential involvement demonstrated moderate reproducibility (ICC 0.74 and 0.70 across rounds). Subgroup analyses showed consistently poor agreement for larger lesions for diameter-based methods, whereas circumferential percentage estimation ranged from poor to good depending on LST size and morphology. Intra-endoscopist agreement for diameter- and axis-based approaches showed wide variability (κ range 0.01–0.67), while circumferential estimates achieved good to excellent agreement for most endoscopists. Visual estimation of large colorectal LST size is highly variable among expert endoscopists. Maximal diameter and axial lengths demonstrate poor inter- and intra-observer reliability. Circumferential extent is the most reproducible descriptor and may be the preferred approach for reporting LST size in clinical practice and research.
Per-oral endoscopic myotomy (POEM) achieves an 80–90
BACKGROUND:More high-quality evidence for stent-assisted coiling (SAC) of cerebral aneurysms is needed. Randomized controlled trials have shown lower recurrence with the HydroCoil Embolic System (HES) compared with bare platinum coils, and self-adjudicated LVIS Evo SAC studies have demonstrated favorable safety and technical success. The objective of the SEALANT study is to investigate the safety and effectiveness of this device combination. METHODS:SEALANT is an open-label, prospective, single-arm, multicenter study including patients with aneurysms up to a maximum size of 12 mm. The primary effectiveness endpoint was complete occlusion on digital subtraction angiography (DSA), and the primary safety endpoint was major ipsilateral stroke or neurological death. Adverse events and imaging were independently adjudicated. RESULTS:206 patients were enrolled; 193 had follow-up imaging for outcome assessment; and 164 satisfied all criteria, were treated with both study devices (including ≥90% HES), and had DSA follow-up. The participants' mean age was 56.8±11.6 years; 65% were female. Of 212 aneurysms, 86.3% were bifurcation and 13.2% sidewall; 25.0% were previously treated and 17.9% previously ruptured; 89.6% were anterior circulation, 34.0% at the anterior communicating artery, and 30.7% at the middle cerebral artery; and 84.9% were wide-necked (neck ≥4 mm or dome/neck ratio <2). The primary effectiveness endpoint of complete occlusion was achieved in 82.1% (95% CI 76.4% to 87.9%) of aneurysms. 2.4% (95% CI 0.3% to 4.5%) of patients met the primary safety endpoint of major ipsilateral stroke or neurological death. The primary safety endpoint was significantly associated with baseline modified Rankin Scale and history of previous stroke. CONCLUSION:1-year SEALANT results confirm the safety and effectiveness of SAC using LVIS Evo and HES. TRIAL REGISTRATION NUMBER:NCT04999423.
Background Epstein-Barr virus (EBV)-associated primary central nervous system lymphoma (PCNSL) is a rare form of extranodal non-Hodgkin's lymphoma closely linked to immunodeficiency. Imaging characteristics of EBV-associated are reported to differ from those of typical EBV-negative PCNSL. This study aims to describe the radiological and nuclear medicine imaging features in a large cohort of patients with EBV-associated PCNSL.Methods We conducted a multicenter retrospective descriptive study between 2008 and 2025 on patients with a diagnosis of EBV-associated PCNSL. MRI variables and FDG-PET/CT uptake were assessed.Results Fifty-eight cases of EBV-associated PCNSL were included. All but 1 patient were immunosuppressed. Multiple lesions were present in 71% of cases (41/58). Supratentorial involvement was observed in 90% of cases (52/58). Heterogeneous contrast enhancement was noted in 90% (52/58), with ring-like enhancement in 41% (24/58). Leptomeningeal enhancement occurred in 31% of cases (18/58), and within this group, 50% showed perivascular space enhancement. Lesions showed hypercellularity in 83% (48/58) and intralesional hemorrhage in 81% (47/58). An "eccentric target" sign was present in 26% of cases (15/58), while a "concentric target" sign in 14% (5/35). On FDG-PET, 25/30 patients had hypermetabolic lesions (25/30, 83%).Conclusion Diagnosing EBV-associated PCNSL is challenging due to its rarity and the broad differential diagnosis. Multiple necrotic and hemorrhagic lesions are the most suggestive MRI feature of EBV-associated PCNSL. "Eccentric" and "concentric" target signs, typically associated with CNS toxoplasmosis, can be observed. FDG-PET often reveals hypermetabolic lesions that support a neoplastic diagnosis. Histological confirmation remains essential for confidently treating this tumor entity.
