Pre-operative anatomical data is essential for guidance in neurosurgery, but brain movement and tissue manipulation degrade its registration to the patient during surgery. Restoring the registration requires intra-operative updates of the surface geometry, which a calibrated stereo surgical microscope could triangulate from the rectified image pairs it delivers. However, low-texture surfaces, reflections, and blur make stereo matching in this domain difficult. Zero-shot stereo foundation models, deep networks trained on large synthetic datasets of natural images, generalize to unseen domains without fine-tuning and handle comparable conditions well on public benchmarks. Yet their performance for neurosurgery is unknown because no benchmark exists to evaluate them in this domain.

We close this gap by assessing whether zero-shot stereo foundation models are suitable for stereo reconstruction in neurosurgical microscopy by analyzing their accuracy, robustness, and failure predictability. To this end, the Microscopy Images for Neurosurgical Dense Stereo reconstruction (MINDS) benchmark is introduced, which provides realistic synthetic and ex-vivo porcine brain images with reference depth and segmentation masks. Four zero-shot models and two classical reference methods are evaluated on MINDS for reconstruction accuracy and robustness under reflections, blur, and weak texture. Additionally, a failure predictor is developed from inference-time confidence cues to investigate whether unreliable depth estimates could be caught in practice.

The best classical method places 23% and 76% of pixels within 0.1 mm and 1 mm of the reference, respectively. The best zero-shot model reaches 59% and 97% at the same thresholds. This accuracy gap persists across all texture levels. Under specular reflection, the best zero-shot model stays at 55% and 99%, while the reference methods drop to 4% and 42%. Defocus blur, by contrast, degrades every tested method to a similar extent and remains the dominant barrier to sub-millimeter accuracy. The failure predictor separates reliable from unreliable pixels at an average precision of 0.63 on synthetic data, nearly four times the chance level of 0.17, with recall rising from 0.75 for errors of 1 mm to 5 mm to 0.99 for errors above 10 mm. Together, these results establish the present capabilities and limitations of zero-shot foundation models for neurosurgical stereo reconstruction.
... Read more

While the adoption of programmable data planes is increasing, writing correct and efficient P4 programs remains a challenging and error-prone process. This hinders the rapid development of custom in-network functionality and limits the accessibility of data-plane programming to a small set of experts. In this paper, we present T4G, a system that automatically synthesizes P4 programs from network behavior expressed as input-output network traces. Unlike prior approaches that rely on high-level abstraction or domain-specific languages, T4G operates with example packet captures, eliminating the need for manual abstraction or rule specifications from network operators. T4G introduces a novel input-output grouping technique that identifies logical packet transformation behavior from packet traces, and performs a divide-and-conquer synthesis that scales to complex forwarding logic. Our experimental results show the ability of T4G to generate different types of programs and demonstrate the potential of T4G in generating complex P4 programs. ... Read more

The Greenland and Antarctic ice sheets are major drivers of global mean sea level rise and are predicted to continue to do so in the future. However, they also represent the largest source of uncertainty in projections of future sea level rise making robust estimates of observed ice sheet mass changes critical. Here, we compare and combine 42 independent estimates of ice sheet mass balance derived from satellite observations of temporal changes in ice sheet flow, volume, and gravitational attraction to determine the ice sheet mass balance from 1972 (Greenland) and 1979 (Antarctica) until 2023. We then use regional climate models to partition the total mass balance into contributions associated with surface mass balance and ice dynamical imbalance. The ice sheets lost 11,309 ± 565 billion tonnes of ice between 1979 and 2023, with glacier dynamical imbalance driving 84% of the ice loss and surface mass balance the remainder. This dataset can be used to track the contribution of the ice sheets to sea level rise and constrain projections of future sea level rise.

... Read more

A Collaborative Online International Learning program (COIL) is a virtual exchange model in which students from different universities use digital tools to get an international and intercultural experience. To provide such an experience to undergraduate civil engineering students from Ecuador (USFQ) and Chile (PUCV), a COIL module consisting of 13 shared sessions of lectures and three main activities (icebreaker, design project, final reflection) was designed. This paper reports on the curricular design of a COIL for bridge design teaching, the findings from the activities regarding code comparison and software usage, the joint midterm exam, the reflections of the students, the survey results, and the effect on the overall course evaluation. These insights can serve other bridge engineering instructors to develop similar virtual exchanges for their students, and add to the reflections on innovation in bridge engineering education, the role of digitalization, the development of intercultural competences, and teacher–student reflection. ... Read more

The GRACE and GRACE-FO missions have provided monthly gravity field models for more than 20 years. The upcoming GRACE-C mission will guarantee data continuation for the next decade following the concept of a pair of satellites flying in a polar orbit for global coverage. Along with an additional satellite pair flying in an inclined orbit, the two satellite pairs will form a Bender constellation resulting in the MAGIC mission. As a result, both temporal and spatial resolution of Earth’s gravity field will improve. The novel mission configuration, however, poses new challenges for mission design, such as optimal selection of the constellation orbital parameters. In this work, the lumped coefficients theory is employed to propagate the error spectra of inter-satellite range and GNSS observations to the Stokes coefficients representing the gravity field. Noise propagation from accelerometers and attitude noise to the observations is also considered. This methodology is computationally much more efficient than error analysis studies from full end-to-end simulation. Thus, the design space can be quickly explored to determine a global optimal inclination for the second pair, for example. However, the analytical methodology presents some limitations, mainly the stationarity of observation noise. Therefore, we do not consider background model errors and temporal aliasing cannot be accounted for. Nonetheless, it can be circumvented directly observing the high-frequency signals at daily and sub-daily timescales. Multiple pair constellations are a promising mission concept in this regard. This provides a framework for leveraging the benefits of the analytical methodology. We apply the lumped coefficients theory to multiple pair constellations and analyse their spatio-temporal resolution for constellations up to 100 satellite pairs. ... Read more