Instituto Mauá de Tecnologia

Intracranial carotid dissecting aneurysms (DA) are often treated emergently upon diagnosis. In contrast, extracranial DA are generally considered less likely to rupture and are commonly managed conservatively. Despite these clinical differences, limited data exists on the hemodynamic differences that might explain their divergent clinical behaviors. Retrospective analysis of intracranial and extracranial DA treated between 2011 to 2023. DA were reconstructed from computerized angiographic images using 3D Slicer software. Computational fluid dynamics (CFD) simulations were performed using ANSYS® Fluent package. Hemodynamic parameters calculated included time averaged wall shear stress (TAWSS), high shear areas (HSA), low shear areas (LSA), time averaged wall shear stress ratio (TAWSR), oscillatory shear index (OSI), and relative residence time (RRT). We compared these variables between extracranial and intracranial lesions using Mann–Whitney U and t-tests. Nineteen DA (10 extracranial, 9 intracranial) from 16 patients (age 48–82; 9 male) were analyzed. The average volume and area of the lesions evaluated were 187 mm3 and 158.6 mm2. Two of the intracranial DA were identified in the setting of subarachnoid hemorrhage. Extracranial DA showed significantly greater volume (234.6 vs. 83.0 mm3; p = 0.02) and area (210.9 vs. 59.4 mm2; p = 0.03). Intracranial DA demonstrated nonsignificant trends toward lower TAWSS at the aneurysm (1.04 vs. 1.53 Pa; p = 0.62), lower TAWSSR (0.49 vs. 0.7; p = 0.19), greater LSA (4.2

In MOS circuits, Total Ionizing Dose (TID) leads to charge trapping in oxide layers, which can modify key electrical parameters and impact circuit timing behavior. This work presents an experimental investigation of radiation-induced charge effects on the propagation delay of a commercial FPGA. The device was exposed to a cobalt-60 gamma radiation source for 23 consecutive days, reaching a total ionizing dose of 1.1 $\operatorname{Mrad}(\mathrm{Si})$ at an average dose rate of approximately 2.7 $\mathbf{k r a d}(\mathbf{S i}) / \mathbf{h}$. Throughout the irradiation, no functional failures were observed. However, the measured propagation delay initially decreased, then stabilized at its minimal value. Small temperature variations were monitored and showed no significant influence on the observed behavior. These results indicate that the tested FPGA maintains operational stability and exhibits tolerance to total ionizing dose effects on timing performance.

Accurate and robust localization is a critical challenge for autonomous vehicles, especially in urban environments where the Global Navigation Satellite System (GNSS) signal is often degraded or unavailable. As a low-cost alternative, this paper proposes a visual localization system based on a single camera using an omnidirectional catadioptric vision setup. The system estimates the vehicle’s 2D coordinates (X, Y) by framing the problem as a direct regression task. A systematic comparison was conducted between two pretrained convolutional neural network (CNN) architectures: a classical CNN (VGG-16) and a modern CNN (ConvNeXt V2). These networks were employed as feature extractors to feed a Multilayer Perceptron (MLP) regression network. To validate the proposed approach, a dataset of approximately 11,000 images was collected using an electric utility vehicle instrumented with an RTK GNSS receiver, providing high-precision ground truth. The results demonstrate that direct regression on raw (non-rectified) omnidirectional catadioptric images is feasible, with the VGG-16 model achieving a mean Euclidean distance error of 1.188 meters on the test set. This work validates the proposed approach as a low-cost alternative for vehicle localization.

Anaerobic co-digestion of sewage and glycerol under mesophilic conditions was evaluated for hydrogen and methane production in a soybean biorefinery plant. Seven simulated scenarios were assessed based on scalability and bioenergy production potential, including three single-stage scenarios (with applied organic loading rates of 4.5, 6.0, and 9.0 kg-COD.m⁻3.d⁻1) and four two-stage scenarios (with applied organic loading rates of 18 and 36 kg-COD.m⁻3.d⁻1 for acidogenic reactor and 3.9, 4.8, 6.0 and 11.6 kg-COD.m⁻3.d⁻1 for methanogenic reactor). The economic feasibility of these scenarios was examined by determining the net present value (NPV), internal rate of return (IRR), and payback period considering the cost savings from replacing diesel in the boiler with biogas (hydrogen - H2/ methane - CH4/ carbon dioxide - CO2) produced in the anaerobic co-digestion process, as well as the reduction in CO2 emissions due to this substitution. The single- (S1 scenario) and two-stage (S4 scenario) processes with the lowest glycerol: sewage ratio obtained the most favorable overall technical aspects of wastewater treatment for bioenergy production, including reactor volume (794 and 851 m3, respectively), organic matter removal efficiencies (75 and 83

This work investigates the surface integrity of AISI 316L produced by additive manufacturing (AM) and subsequently machined, with an emphasis on residual stresses. A full factorial design was applied to evaluate the influence of laser power from AM and milling parameters, including cutting speed, cutting depth, feed rate, and cutting fluid. Eight samples were selected for Vickers microhardness characterization. Machined surfaces were examined by scanning electron microscopy. Microhardness results showed higher values at top surfaces, reaching increases of up to 22.7%. Variations in machining parameters significantly affected residual stresses, with changes of up to 500 MPa; all measured values were tensile. Surface roughness Rt ranged from 1.59 to 4.34 µm. The best surface integrity was achieved using 160 W laser power, 170 m/min cutting speed, 0.1 mm/tooth feed rate, 0.35 mm cutting depth, and cutting fluid. ANOVA identified feed rate as the most influential factor, followed by laser power and cutting depth. Machine learning models were developed for residual stress prediction, with PCA-RBFN showing superior performance for parallel residual stress (R2 = 94,53%) and SVM for perpendicular residual stress (R2 = 89.85%). The hybrid approach enabled isolation of individual parameter effects while minimizing interactions between AM and machining.