Vision Transformer-based Decision Support for Neurosurgical Intervention in Acute Traumatic Brain Injury: Automated Surgical Intervention Support Tool.

Purpose To develop an automated triage tool to predict neurosurgical intervention for patients with traumatic brain injury (TBI). Materials and Methods A provincial trauma registry was reviewed to retrospectively identify patients with TBI from 2005 to 2022 treated at a specialized Canadian trauma center. Model training, validation, and testing were performed using head CT scans with binary reference standard patient-level labels corresponding to whether the patient received neurosurgical intervention. Performance and accuracy of the model, the Automated Surgical Intervention Support Tool for TBI (ASIST-TBI), were also assessed using a held-out consecutive test set of all patients with TBI presenting to the center between March 2021 and September 2022. Results Head CT scans from 2806 patients with TBI (mean age, 57 years ± 22 [SD]; 1955 [70%] men) were acquired between 2005 and 2021 and used for training, validation, and testing. Consecutive scans from an additional 612 patients (mean age, 61 years ± 22; 443 [72%] men) were used to assess the performance of ASIST-TBI. There was accurate prediction of neurosurgical intervention with an area under the receiver operating characteristic curve (AUC) of 0.92 (95% CI: 0.88, 0.94), accuracy of 87% (491 of 562), sensitivity of 87% (196 of 225), and specificity of 88% (295 of 337) on the test dataset. Performance on the held-out test dataset remained robust with an AUC of 0.89 (95% CI: 0.85, 0.91), accuracy of 84% (517 of 612), sensitivity of 85% (199 of 235), and specificity of 84% (318 of 377). Conclusion A novel deep learning model was developed that could accurately predict the requirement for neurosurgical intervention using acute TBI CT scans. Keywords: CT, Brain/Brain Stem, Surgery, Trauma, Prognosis, Classification, Application Domain, Traumatic Brain Injury, Triage, Machine Learning, Decision Support Supplemental material is available for this article. © RSNA, 2024 See also commentary by Haller in this issue.

Delayed Gadolinium Leakage in Ocular Structures on Brain MR Imaging: Prevalence and Associated Factors.

BACKGROUND AND PURPOSE: Gadolinium leakage in ocular structures (GLOS) is characterized by hyperintense signal in the chambers of the eye on FLAIR and has been reported in association with blood-ocular barrier breakdown in patients with ischemic strokes. The underlying mechanism of GLOS remains poorly understood; however, some studies suggest it may be part of a physiologic excretion pathway of gadolinium. This study aimed to determine the prevalence of GLOS in an unselected patient population. MATERIALS AND METHODS: A retrospective analysis was conducted on 439 patients who underwent brain MR imaging within 7 days of receiving a gadolinium-based contrast agent injection for a prior MR imaging study. Clinical, imaging, and laboratory data were collected. Descriptive and logistic regression analyses were performed. RESULTS: GLOS was observed in 26 of 439 patients (6%). The occurrence of GLOS varied with time, with 3 (12%), 14 (54%), 8 (31%), and 1 (4%) patient showing GLOS within 24, 25-72, 73-120, and >120 hours after gadolinium-based contrast agent injection, respectively. Patients with GLOS were older (median age: 72 versus 55 years, P = .001) and had higher median serum creatinine levels (73 versus 64 µmol/L, P = .005) and a lower median estimated glomerular filtration rate (84 versus 101 mL/min/1.73 m2, P < .001). A shorter median time interval between gadolinium-based contrast agent injection and the index brain MR imaging was observed in the group positive for GLOS (62 versus 91 hours, P = .003). Multivariable regression analysis identified the estimated glomerular filtration rate (OR = 0.970; 95% CI, 0.049-0.992; P = .008) and time interval since gadolinium-based contrast agent injection (OR = 0.987; 95% CI, 0.977-0.997; P = .012) as independent factors associated with GLOS. CONCLUSIONS: GLOS was observed in only a small percentage of patients receiving gadolinium-based contrast agent within 7 days before brain MR imaging. This phenomenon was noted in patients with normal findings on brain MR imaging and those with various CNS pathologies, and it was associated with lower estimated glomerular filtration rates and shorter time intervals after gadolinium-based contrast agent injection. While GLOS may be a physiologic gadolinium-based contrast agent excretion pathway, the presence of ocular disease was not formally evaluated in the included population. Awareness of GLOS is nonetheless useful for appropriate radiologic interpretation.

Correlations in distribution and concentration of calcium, copper and iron with zinc in isolated extracellular deposits associated with age-related macular degeneration.

Zinc (Zn) is abundantly enriched in sub-retinal pigment epithelial (RPE) deposits, the hallmarks of age-related macular degeneration (AMD), and is thought to play a role in the formation of these deposits. However, it is not known whether Zn is the only metal relevant for sub-RPE deposit formation. Because of their involvement in the pathogenesis of AMD, we determined the concentration and distribution of calcium (Ca), iron (Fe) and copper (Cu) and compared these with Zn in isolated and sectioned macular (MSD), equatorial (PHD) and far peripheral (FPD) sub-RPE deposits from an 86 year old donor eye with post mortem diagnosis of early AMD. The sections were mounted on Zn free microscopy slides and analyzed by microprobe synchrotron X-ray fluorescence (µSXRF). Metal concentrations were determined using spiked sectioned sheep brain matrix standards, prepared the same way as the samples. The heterogeneity of metal distributions was examined using pixel by pixel comparison. The orders of metal concentrations were Ca ⋙ Zn > Fe in all three types of deposits but Cu levels were not distinguishable from background values. Zinc and Ca were consistently present in all deposits but reached highest concentration in MSD. Iron was present in some but not all deposits and was especially enriched in FPD. Correlation analysis indicated considerable variation in metal distribution within and between sub-RPE deposits. The results suggest that Zn and Ca are the most likely contributors to deposit formation especially in MSD, the characteristic risk factor for the development of AMD in the human eye.

Evidence that the ZNT3 protein controls the total amount of elemental zinc in synaptic vesicles.

The ZNT3 protein decorates the presynaptic vesicles of central neurons harboring vesicular zinc, and deletion of this protein removes staining for zinc. However, it has been unclear whether only histochemically reactive zinc is lacking or if, indeed, total elemental zinc is missing from neurons lacking the Slc30a3 gene, which encodes the ZNT3 protein. The limitations of conventional histochemical procedures have contributed to this enigma. However, a novel technique, microprobe synchrotron X-ray fluorescence, reveals that the normal 2- to 3-fold elevation of zinc concentration normally present in the hippocampal mossy fibers is absent in Slc30a3 knockout (ZNT3) mice. Thus, the ZNT3 protein evidently controls not only the "stainability" but also the actual mass of zinc in mossy-fiber synaptic vesicles. This work thus confirms the metal-transporting role of the ZNT3 protein in the brain.