arXiv subjects

Christian Heiliger

Publications and source records attributed to Christian Heiliger.

At least 19 recordsLinked to original sources

arXiv2026

A 4D Representation for Training-Free Agentic Reasoning from Monocular Laparoscopic Video

Spatiotemporal reasoning is a fundamental capability for artificial intelligence (AI) in soft tissue surgery, paving the way for intelligent assistive systems and autonomous robotics. While 2D vision-language models show increasing promise at understanding surgical video, the spatial complexity of surgical scenes suggests that reasoning systems may benefit from explicit 4D representations. Here, we propose a framework for equipping surgical agents with spatiotemporal tools based on an explicit 4D representation, enabling AI systems to ground their natural language reasoning in both time and 3D space. Leveraging models for point tracking, depth, and segmentation, we develop a coherent 4D model with spatiotemporally consistent tool and tissue semantics. A Multimodal Large Language Model (MLLM) then acts as an agent on tools derived from the explicit 4D representation (e.g., trajectories) without any fine-tuning. We evaluate our method on a new dataset of 134 clinically relevant questions and find that the combination of a general purpose reasoning backbone and our 4D representation significantly improves spatiotemporal understanding and allows for 4D grounding. We demonstrate that spatiotemporal intelligence can be "assembled" from 2D MLLMs and 3D computer vision models without additional training. Code, data, and examples are available at https://tum-ai.github.io/surg4d/

cs.CV↗

arXiv2025

BridgeSplat: Bidirectionally Coupled CT and Non-Rigid Gaussian Splatting for Deformable Intraoperative Surgical Navigation

We introduce BridgeSplat, a novel approach for deformable surgical navigation that couples intraoperative 3D reconstruction with preoperative CT data to bridge the gap between surgical video and volumetric patient data. Our method rigs 3D Gaussians to a CT mesh, enabling joint optimization of Gaussian parameters and mesh deformation through photometric supervision. By parametrizing each Gaussian relative to its parent mesh triangle, we enforce alignment between Gaussians and mesh and obtain deformations that can be propagated back to update the CT. We demonstrate BridgeSplat's effectiveness on visceral pig surgeries and synthetic data of a human liver under simulation, showing sensible deformations of the preoperative CT on monocular RGB data. Code, data, and additional resources can be found at https://maxfehrentz.github.io/ct-informed-splatting/ .

cs.CV↗

arXiv2025

Realistic modelling of transport properties at finite temperature in magnetic materials by local quantization of a Heisenberg model

The quantitative description of the electrical resistivity of a magnetic material remains challenging to this day. Qualitatively, it is well understood that the temperature-induced lattice and spin disorder determines the temperature dependence of the resistivity. While prior publications reached good agreement with experiment in the so-called supercell or direct approach for non-magnetic materials where the spin-disorder contribution to the resistivity is negligible, an accurate, purely theoretical description of magnetic materials remains elusive. This shortcoming can be attributed to the missing accuracy in the description of the temperature-dependent spin-disorder itself. In this work, we employ a joint approach from \textit{ab-initio} transport calculations and atomistic modeling of the temperature-dependent spin-disorder. Using the example of $\alpha$-Fe, we demonstrate that the inclusion of quantum mechanical effects using a semiclassical local quantization of the Heisenberg model significantly improves the description of the spin-disorder component to the electrical resistivity. Compared to previous approaches, this model includes the description of magnetic short-range order effects, enabling us to study temperature effects around and above the Curie temperature, where prior mean-field theory-based approaches inevitably predicted a constant contribution.

cond-mat.mes-hall↗

arXiv2025

TrackOR: Towards Personalized Intelligent Operating Rooms Through Robust Tracking

Providing intelligent support to surgical teams is a key frontier in automated surgical scene understanding, with the long-term goal of improving patient outcomes. Developing personalized intelligence for all staff members requires maintaining a consistent state of who is located where for long surgical procedures, which still poses numerous computational challenges. We propose TrackOR, a framework for tackling long-term multi-person tracking and re-identification in the operating room. TrackOR uses 3D geometric signatures to achieve state-of-the-art online tracking performance (+11% Association Accuracy over the strongest baseline), while also enabling an effective offline recovery process to create analysis-ready trajectories. Our work shows that by leveraging 3D geometric information, persistent identity tracking becomes attainable, enabling a critical shift towards the more granular, staff-centric analyses required for personalized intelligent systems in the operating room. This new capability opens up various applications, including our proposed temporal pathway imprints that translate raw tracking data into actionable insights for improving team efficiency and safety and ultimately providing personalized support.

cs.CV↗

arXiv2025

Recognizing Surgical Phases Anywhere: Few-Shot Test-time Adaptation and Task-graph Guided Refinement

The complexity and diversity of surgical workflows, driven by heterogeneous operating room settings, institutional protocols, and anatomical variability, present a significant challenge in developing generalizable models for cross-institutional and cross-procedural surgical understanding. While recent surgical foundation models pretrained on large-scale vision-language data offer promising transferability, their zero-shot performance remains constrained by domain shifts, limiting their utility in unseen surgical environments. To address this, we introduce Surgical Phase Anywhere (SPA), a lightweight framework for versatile surgical workflow understanding that adapts foundation models to institutional settings with minimal annotation. SPA leverages few-shot spatial adaptation to align multi-modal embeddings with institution-specific surgical scenes and phases. It also ensures temporal consistency through diffusion modeling, which encodes task-graph priors derived from institutional procedure protocols. Finally, SPA employs dynamic test-time adaptation, exploiting the mutual agreement between multi-modal phase prediction streams to adapt the model to a given test video in a self-supervised manner, enhancing the reliability under test-time distribution shifts. SPA is a lightweight adaptation framework, allowing hospitals to rapidly customize phase recognition models by defining phases in natural language text, annotating a few images with the phase labels, and providing a task graph defining phase transitions. The experimental results show that the SPA framework achieves state-of-the-art performance in few-shot surgical phase recognition across multiple institutions and procedures, even outperforming full-shot models with 32-shot labeled data. Code is available at https://github.com/CAMMA-public/SPA

cs.CV↗

arXiv2025

Beyond Role-Based Surgical Domain Modeling: Generalizable Re-Identification in the Operating Room

Surgical domain models improve workflow optimization through automated predictions of each staff member's surgical role. However, mounting evidence indicates that team familiarity and individuality impact surgical outcomes. We present a novel staff-centric modeling approach that characterizes individual team members through their distinctive movement patterns and physical characteristics, enabling long-term tracking and analysis of surgical personnel across multiple procedures. To address the challenge of inter-clinic variability, we develop a generalizable re-identification framework that encodes sequences of 3D point clouds to capture shape and articulated motion patterns unique to each individual. Our method achieves 86.19% accuracy on realistic clinical data while maintaining 75.27% accuracy when transferring between different environments - a 12% improvement over existing methods. When used to augment markerless personnel tracking, our approach improves accuracy by over 50%. Through extensive validation across three datasets and the introduction of a novel workflow visualization technique, we demonstrate how our framework can reveal novel insights into surgical team dynamics and space utilization patterns, advancing methods to analyze surgical workflows and team coordination.

cs.CV↗

arXiv2022

Know your sensORs -- A Modality Study For Surgical Action Classification

The surgical operating room (OR) presents many opportunities for automation and optimization. Videos from various sources in the OR are becoming increasingly available. The medical community seeks to leverage this wealth of data to develop automated methods to advance interventional care, lower costs, and improve overall patient outcomes. Existing datasets from OR room cameras are thus far limited in size or modalities acquired, leaving it unclear which sensor modalities are best suited for tasks such as recognizing surgical action from videos. This study demonstrates that surgical action recognition performance can vary depending on the image modalities used. We perform a methodical analysis on several commonly available sensor modalities, presenting two fusion approaches that improve classification performance. The analyses are carried out on a set of multi-view RGB-D video recordings of 18 laparoscopic procedures.

cs.CV↗

arXiv2021

Spin accumulation from non-equilibrium first principles methods

For the technologically relevant spin Hall effect most theoretical approaches rely on the evaluation of the spin-conductivity tensor. In contrast, for most experimental configurations the generation of spin accumulation at interfaces and surfaces is the relevant quantity. Here, we directly calculate the accumulation of spins due to the spin Hall effect at the surface of a thin metallic layer, making quantitative predictions for different materials. Two distinct limits are considered, both relying on a fully relativistic Korringa-Kohn-Rostoker density functional theory method. In the semiclassical approach, we use the Boltzmann transport formalism and compare it directly to a fully quantum mechanical non-equilibrium Keldysh formalism. Restricting the calculations to the spin Hall induced, odd in spatial inversion, contribution in the limit of the relaxation time approximation we find good agreement between both methods, where deviations can be attributed to the complexity of Fermi surfaces. Finally, we compare our results to experimental values of the spin accumulation at surfaces as well as the Hall angle and find good agreement for the trend across the considered elements.

cond-mat.mtrl-sci↗

arXiv2020

Surface Diffusion Control Enables Tailored Aspect Ratio Nanostructures in Area-Selective Atomic Layer Deposition

Area-selective atomic layer deposition is a key technology for modern microelectronics as it eliminates alignment errors inherent to conventional approaches by enabling material deposition only in specific areas. Typically, the selectivity originates from surface modifications of the substrate that allow or block precursor adsorption. The control of the deposition process currently remains a major challenge as the selectivity of the no-growth areas is lost quickly. Here, we show that surface modifications of the substrate strongly manipulate the surface diffusion. The selective deposition of TiO$_2$ on poly (methyl methacrylate) and SiO$_2$ yields localized nanostructures with tailored aspect ratios. Controlling the surface diffusion allows to tune such nanostructures as it boosts the growth rate at the interface of the growth and no-growth areas. Kinetic Monte-Carlo calculations reveal that species move from high to low diffusion areas. Further, we identify the catalytic activity of TiCl$_4$ during the formation of carboxylic acid on poly (methyl methacrylate) as the reaction mechanism responsible for the loss of selectivity, and show that process optimization leads to higher selectivity. Our work enables the precise control of area-selective atomic layer deposition on the nanoscale, and offers new strategies in area-selective deposition processes by exploiting surface diffusion effects.

cond-mat.mtrl-sci↗

arXiv2019

Skyrmion meets magnetic tunnel junction: an efficient way for electrical skyrmion detection investigated by ab initio theory

In our proof-of-principle study we examine the influence of skyrmions on magnetoresistive transport. In particular, we show that magnetic tunnel junctions are a technologically appealing and promising way for electrical detection of non-collinear magnetic structures. The calculated effect is shown to originate from scattering between different k-states and cannot be identified through densities of states alone. Our results suggest that the detection efficiency strongly depends on the utilized materials.

cond-mat.mtrl-sci↗

arXiv2018

Tunnel magneto-Seebeck effect

The interplay of charge, spin and heat transport is investigated in the fascinating research field of spin caloritronics, the marriage of spintronics and thermoelectrics. Here, many new spin-dependent thermal transport phenomena in magnetic nanostructures have been explored in the recent years. One of them is the tunnel magneto-Seebeck (TMS) effect in magnetic tunnel junctions (MTJs) that has large potential for future nanoelectronic devices, such as nanostructured sensors for three-dimentional thermal gradients, or scanning tunneling microscopes driven by temperature differences. The TMS describes the dependence of the MTJ's thermopower on its magnetic cofiguration when a thermal gradient is applied. In this review, we highlight the successful way from first observation of the TMS in 2011 to current ongoing developments in this research area. We emphasize on different heating techniques, material designs, applications, and additional physical aspects such as the role of the thermal conductivity of the barrier material. We further demonstrate the efficient interplay between ab initio calculations and experiments within this field, as this has led, e.g., to the detection of large TMS ratios in MTJs with half-metallic Heusler electrodes.

cond-mat.mes-hall↗

arXiv2017

Large magneto-Seebeck effect in magnetic tunnel junctions with half-metallic Heusler electrodes

Spin caloritronics studies the interplay between charge-, heat- and spin-currents, which are initiated by temperature gradients in magnetic nanostructures. A plethora of new phenomena has been discovered that promises, e.g., to make wasted heat in electronic devices useable or to provide new read-out mechanisms for information. However, only few materials have been studied so far with Seebeck voltages of only some μV, which hampers applications. Here, we demonstrate that half-metallic Heusler compounds are hot candidates for enhancing spin-dependent thermoelectric effects. This becomes evident when considering the asymmetry of the spin-split density of electronic states around the Fermi level that determines the spin-dependent thermoelectric transport in magnetic tunnel junctions. We identify Co$_2$FeAl and Co$_2$FeSi Heusler compounds as ideal due to their energy gaps in the minority density of states, and demonstrate devices with substantially larger Seebeck voltages and tunnel magneto-Seebeck effect ratios than the commonly used Co-Fe-B based junctions.

cond-mat.mtrl-sci↗

arXiv2017

Enhancement of thermovoltage and tunnel magneto-Seebeck effect in CoFeB based magnetic tunnel junctions by variation of the MgAl$_2$O$_4$ and MgO barrier thickness

We investigate the influence of the barrier thickness of Co$_{40}$Fe$_{40}$B$_{20}$ based magnetic tunnel junctions on the laser-induced tunnel magneto-Seebeck effect. Varying the barrier thickness from 1nm to 3nm, we find a distinct maximum in the tunnel magneto-Seebeck effect for 2.6nm barrier thickness. This maximum is independently measured for two barrier materials, namely MgAl$_2$O$_4$ and MgO. Additionally, samples with an MgAl$_2$O$_4$ barrier exhibit a high thermovoltage of more than 350$μ$V in comparison to 90$μ$V for the MTJs with MgO barrier when heated with the maximum laser power of 150mW. Our results allow for the fabrication of improved stacks when dealing with temperature differences across magnetic tunnel junctions for future applications in spin caloritronics, the emerging research field that combines spintronics and themoelectrics.

cond-mat.mtrl-sci↗

arXiv2015

Thermal transport and non-equilibrium temperature drop across a magnetic nanostructured interface

In a number of current experiments in the field of spin-caloritronics a temperature gradient across a nanostructured interface is applied and spin-dependent transport phenomena are observed. However, a lack in the interpretation and knowledge let it unclear how the temperature drop across a magnetic nanostructured interface looks like where both phonons and electrons may contribute to thermal transport. We answer this question for the case of a magnetic tunnel junction (MTJ) where the tunneling magneto Seebeck effect occurs. Nevertheless, our results can be extended to other nanostructured interfaces as well. Using an \textit{ab initio} method we explicitly calculate phonon and electron thermal conductance across the Fe/MgO/Fe-MTJs by using Green's function method. Further, by estimating the electron-phonon interaction in the Fe leads we are able to calculate the electron and phonon temperature profile across the Fe/MgO/Fe-MTJ. Our results show that there is an electron-phonon temperature imbalance at the Fe-MgO interfaces. In consequence, a revision of the interpretation of current experimental measurements might be necessary.

cond-mat.mes-hall↗

arXiv2014

On/off switching of bit readout in bias-enhanced tunnel magneto-Seebeck effect

Thermoelectric effects in magnetic tunnel junctions are currently an attractive research topic. Here, we demonstrate that the tunnel magneto-Seebeck effect (TMS) in CoFeB/MgO/CoFeB tunnel junctions can be switched on to a logic 1 state and off to 0 by simply changing the magnetic state of the CoFeB electrodes. We enable this new functionality of magnetic tunnel junctions by combining a thermal gradient and an electric field. This new technique unveils the bias-enhanced tunnel magneto-Seebeck effect, which can serve as the basis for logic devices or memories in a green information technology with a pure thermal write and read process. Furthermore, the thermally generated voltages that are referred to as the Seebeck effect are well known to sensitively depend on the electronic structure and therefore have been valued in solid-state physics for nearly one hundred years. Here, we lift Seebeck's historic discovery from 1821 to a new level of current spintronics. Our results show that the signal crosses zero and can be adjusted by tuning a bias voltage that is applied between the electrodes of the junction; hence, the name of the effect is bias-enhanced tunnel magneto-Seebeck effect (bTMS). Via the spin- and energy-dependent transmission of electrons in the junction, the bTMS effect can be configured using the bias voltage with much higher control than the tunnel magnetoresistance (TMR) and even completely suppressed for only one magnetic configuration, which is either parallel (P) or anti-parallel (AP). This option allows a readout contrast for the magnetic information of -3000% at room temperature while maintaining a large signal for one magnetic orientation. This contrast is much larger than the value that can be obtained using the TMR effect. Moreover, our measurements are a step towards the experimental realization of high TMS ratios, which are predicted for specific Co-Fe compositions.

cond-mat.mes-hall↗

arXiv2013

Unusual angular dependence of tunneling magneto-Seebeck effect

We find an unusual angular dependence of the tunneling magneto-Seebeck effect (TMS). The conductance shows normally a cosine-dependence with the angle between the magnetizations of the two ferromagnetic leads. In contrast, the angular dependence of the TMS depends strongly on the tunneling magneto resistance (TMR) ratio. For small TMR ratios we obtain also a cosine-dependence whereas for very large TMR ratios the angular dependence approaches a step-like function.

cond-mat.mes-hall↗

arXiv2013

Thermal spin-transfer torque in magnetic tunnel junctions

The thermal spin-transfer torque (TSTT) is an effect to switch the magnetic free layer in a magnetic tunnel junction by a temperature gradient only. We present ab initio calculations of the TSTT. In particular, we discuss the influence of magnetic layer composition by considering $\text{Fe}_\text{x}\text{Co}_{\text{1-x}}$ alloys. Further, we compare the TSTT to the bias voltage driven STT and discuss the requirements for a possible thermal switching. For example, only for very thin barriers of 3 monolayers MgO a thermal switching is imaginable. However, even for such a thin barrier the TSTT is still too small for switching at the moment and further optimization is needed. In particular, the TSTT strongly depends on the composition of the ferromagentic layer. In our current study it turns out that at the chosen thickness of the ferromagnetic layer pure Fe gives the highest thermal spin-transfer torque.

cond-mat.mes-hall↗

arXiv2013

Band Structure and Effective Masses of ZnMgO

We analyze the influence of the Mg concentration on several important properties of the band structure of ZnMgO alloys in wurtzite structure using ab initio calculations. For this purpose, the band structure for finite concentrations is defined in terms of the Bloch spectral density, which can be calculated within the coherent potential approximation. We investigate the concentration dependence of the band gap and the crystal-field splitting of the valence bands. The effective electron and hole masses are determined by extending the effective mass model to finite concentrations. We compare our results with experimental results and other calculations.

cond-mat.mtrl-sci↗