arXiv subjects

Steffen Löck

Publications and source records attributed to Steffen Löck.

16 recordsLinked to original sources

arXiv2026

Combined photon-proton modeling of radiation-induced brain imaging changes supports variability in proton relative biological effectiveness and increased periventricular radiosensitivity

Purpose: Recent proton-only investigations of radiation-induced contrast enhancements (RICE) in brain tumor patients indicated variability in proton relative biological effectiveness (RBE) and increased radiosensitivity of the periventricular region (PVR). Because RBE is defined relative to a reference radiation, these studies required assumptions on the photon dose-response relationship. This study aimed to validate proton RBE variability and PVR radiosensitivity using predictive modeling of RICE in a combined photon-proton cohort. Methods and Materials: Predictive models for RICE detected on follow-up magnetic resonance imaging were developed in 152 intracranial tumor patients treated with photons or protons. Logistic regression was applied at the voxel level to model spatial occurrence and at the patient level to model incidence. A clinical RBE model was derived from voxel-wise comparisons of estimated risk between photon and proton irradiation. Results: In total, 128 RICE of various grades occurred in 64 patients. Voxel-level modeling validated absorbed dose (D), D multiplied by dose-averaged linear energy transfer (LETd) for proton therapy, and PVR as independent predictors of RICE. The model implied a variable proton RBE described by RBE=1+m$\cdot$LETd, with m=0.10 $μ$m/keV. At the patient level, the PVR D2ml based on this RBE achieved the highest predictive performance. Conclusions: RICE was spatially associated with dose and PVR proximity across photon and proton therapy, with an additional LET-dependent component after proton therapy. The cross-modality framework validates proton RBE variability against an observed photon reference rather than predefined reference assumptions, and supports PVR radiosensitivity as modality-independent. Accounting for variable proton RBE and the PVR as an organ at risk may improve risk assessment and mitigation of radiation-induced side effects.

physics.med-ph↗

arXiv2024

Unraveling Radiomics Complexity: Strategies for Optimal Simplicity in Predictive Modeling

Background: The high dimensionality of radiomic feature sets, the variability in radiomic feature types and potentially high computational requirements all underscore the need for an effective method to identify the smallest set of predictive features for a given clinical problem. Purpose: Develop a methodology and tools to identify and explain the smallest set of predictive radiomic features. Materials and Methods: 89,714 radiomic features were extracted from five cancer datasets: low-grade glioma, meningioma, non-small cell lung cancer (NSCLC), and two renal cell carcinoma cohorts (n=2104). Features were categorized by computational complexity into morphological, intensity, texture, linear filters, and nonlinear filters. Models were trained and evaluated on each complexity level using the area under the curve (AUC). The most informative features were identified, and their importance was explained. The optimal complexity level and associated most informative features were identified using systematic statistical significance analyses and a false discovery avoidance procedure, respectively. Their predictive importance was explained using a novel tree-based method. Results: MEDimage, a new open-source tool, was developed to facilitate radiomic studies. Morphological features were optimal for MRI-based meningioma (AUC: 0.65) and low-grade glioma (AUC: 0.68). Intensity features were optimal for CECT-based renal cell carcinoma (AUC: 0.82) and CT-based NSCLC (AUC: 0.76). Texture features were optimal for MRI-based renal cell carcinoma (AUC: 0.72). Tuning the Hounsfield unit range improved results for CECT-based renal cell carcinoma (AUC: 0.86). Conclusion: Our proposed methodology and software can estimate the optimal radiomics complexity level for specific medical outcomes, potentially simplifying the use of radiomics in predictive modeling across various contexts.

eess.IV↗

arXiv2023

Graph data modelling for outcome prediction in oropharyngeal cancer patients

Graph neural networks (GNNs) are becoming increasingly popular in the medical domain for the tasks of disease classification and outcome prediction. Since patient data is not readily available as a graph, most existing methods either manually define a patient graph, or learn a latent graph based on pairwise similarities between the patients. There are also hypergraph neural network (HGNN)-based methods that were introduced recently to exploit potential higher order associations between the patients by representing them as a hypergraph. In this work, we propose a patient hypergraph network (PHGN), which has been investigated in an inductive learning setup for binary outcome prediction in oropharyngeal cancer (OPC) patients using computed tomography (CT)-based radiomic features for the first time. Additionally, the proposed model was extended to perform time-to-event analyses, and compared with GNN and baseline linear models.

cs.CV↗

arXiv2019

Image biomarker standardisation initiative

The image biomarker standardisation initiative (IBSI) is an independent international collaboration which works towards standardising the extraction of image biomarkers from acquired imaging for the purpose of high-throughput quantitative image analysis (radiomics). Lack of reproducibility and validation of high-throughput quantitative image analysis studies is considered to be a major challenge for the field. Part of this challenge lies in the scantiness of consensus-based guidelines and definitions for the process of translating acquired imaging into high-throughput image biomarkers. The IBSI therefore seeks to provide image biomarker nomenclature and definitions, benchmark data sets, and benchmark values to verify image processing and image biomarker calculations, as well as reporting guidelines, for high-throughput image analysis.

cs.CV↗

arXiv2018

Assessing robustness of radiomic features by image perturbation

Image features need to be robust against differences in positioning, acquisition and segmentation to ensure reproducibility. Radiomic models that only include robust features can be used to analyse new images, whereas models with non-robust features may fail to predict the outcome of interest accurately. Test-retest imaging is recommended to assess robustness, but may not be available for the phenotype of interest. We therefore investigated 18 methods to determine feature robustness based on image perturbations. Test-retest and perturbation robustness were compared for 4032 features that were computed from the gross tumour volume in two cohorts with computed tomography imaging: I) 31 non-small-cell lung cancer (NSCLC) patients; II): 19 head-and-neck squamous cell carcinoma (HNSCC) patients. Robustness was measured using the intraclass correlation coefficient (1,1) (ICC). Features with ICC$\geq0.90$ were considered robust. The NSCLC cohort contained more robust features for test-retest imaging than the HNSCC cohort ($73.5\%$ vs. $34.0\%$). A perturbation chain consisting of noise addition, affine translation, volume growth/shrinkage and supervoxel-based contour randomisation identified the fewest false positive robust features (NSCLC: $3.3\%$; HNSCC: $10.0\%$). Thus, this perturbation chain may be used to assess feature robustness.

cs.CV↗

arXiv2015

Experimental Observation of Resonance-Assisted Tunneling

We present the first experimental observation of resonance-assisted tunneling, a wave phenomenon, where regular-to-chaotic tunneling is strongly enhanced by the presence of a classical nonlinear resonance chain. For this we use a microwave cavity made of oxygen free copper with the shape of a desymmetrized cosine billiard designed with a large nonlinear resonance chain in the regular region. It is opened in a region, where only chaotic dynamics takes place, such that the tunneling rate of a regular mode to the chaotic region increases the line width of the mode. Resonance-assisted tunneling is demonstrated by (i) a parametric variation and (ii) the characteristic plateau and peak structure towards the semiclassical limit.

cond-mat.mes-hall↗

arXiv2013

Integrable Approximation of Regular Islands: The Iterative Canonical Transformation Method

Generic Hamiltonian systems have a mixed phase space, where classically disjoint regions of regular and chaotic motion coexist. We present an iterative method to construct an integrable approximation, which resembles the regular dynamics of a given mixed system and extends it into the chaotic region. The method is based on the construction of an integrable approximation in action representation which is then improved in phase space by iterative applications of canonical transformations. This method works for strongly perturbed systems and arbitrary degrees of freedom. We apply it to the standard map and the cosine billiard.

nlin.CD↗

arXiv2012

Complex paths for regular-to-chaotic tunneling rates

In generic Hamiltonian systems tori of regular motion are dynamically separated from regions of chaotic motion in phase space. Quantum mechanically these phase-space regions are coupled by dynamical tunneling. We introduce a semiclassical approach based on complex paths for the prediction of dynamical tunneling rates from regular tori to the chaotic region. This approach is demonstrated for the standard map giving excellent agreement with numerically determined tunneling rates.

nlin.CD↗

arXiv2012

Consequences of Flooding on Spectral Statistics

We study spectral statistics in systems with a mixed phase space, in which regions of regular and chaotic motion coexist. Increasing their density of states, we observe a transition of the level-spacing distribution P(s) from Berry-Robnik to Wigner statistics, although the underlying classical phase-space structure and the effective Planck constant remain unchanged. This transition is induced by flooding, i.e., the disappearance of regular states due to increasing regular-to-chaotic couplings. We account for this effect by a flooding-improved Berry-Robnik distribution, in which an effectively reduced size of the regular island enters. To additionally describe power-law level repulsion at small spacings, we extend this prediction by explicitly considering the tunneling couplings between regular and chaotic states. This results in a flooding- and tunneling-improved Berry-Robnik distribution which is in excellent agreement with numerical data.

nlin.CD↗

arXiv2012

Coupling of bouncing-ball modes to the chaotic sea and their counting function

We study the coupling of bouncing-ball modes to chaotic modes in two-dimensional billiards with two parallel boundary segments. Analytically, we predict the corresponding decay rates using the fictitious integrable system approach. Agreement with numerically determined rates is found for the stadium and the cosine billiard. We use this result to predict the asymptotic behavior of the counting function N_bb(E) ~ E^δ. For the stadium billiard we find agreement with the previous result δ= 3/4. For the cosine billiard we derive δ= 5/8, which is confirmed numerically and is well below the previously predicted upper bound δ=9/10.

nlin.CD↗

arXiv2011

Partial Weyl Law for Billiards

For two-dimensional quantum billiards we derive the partial Weyl law, i.e. the average density of states, for a subset of eigenstates concentrating on an invariant region $Γ$ of phase space. The leading term is proportional to the area of the billiard times the phase-space fraction of $Γ$. The boundary term is proportional to the fraction of the boundary where parallel trajectories belong to $Γ$. Our result is numerically confirmed for the mushroom billiard and the generic cosine billiard, where we count the number of chaotic and regular states, and for the elliptical billiard, where we consider rotating and oscillating states.

quant-ph↗

arXiv2011

Direct Regular-to-Chaotic Tunneling Rates Using the Fictitious Integrable System Approach

In systems with a mixed phase space, where regular and chaotic motion coexists, regular states are coupled to the chaotic region by dynamical tunneling. We give an overview on the determination of direct regular-to-chaotic tunneling rates using the fictitious integrable system approach. This approach is applied to different kicked systems, including the standard map, and successfully compared with numerical data. This text corresponds to Chapter 6 of the book: Dynamical Tunneling - Theory and Experiment, edited by S. Keshavamurthy and P. Schlagheck [Taylor and Francis CRC (2011)]. For a more extensive exposition see [Phys. Rev. E 82, 056208 (2010); arXiv:1009.0418v2].

nlin.CD↗

arXiv2011

Fractional-Power-Law Level-Statistics due to Dynamical Tunneling

For systems with a mixed phase space we demonstrate that dynamical tunneling universally leads to a fractional power law of the level-spacing distribution P(s) over a wide range of small spacings s. Going beyond Berry-Robnik statistics, we take into account that dynamical tunneling rates between the regular and the chaotic region vary over many orders of magnitude. This results in a prediction of P(s) which excellently describes the spectral data of the standard map. Moreover, we show that the power-law exponent is proportional to the effective Planck constant h.

nlin.CD↗

arXiv2010

Direct regular-to-chaotic tunneling rates using the fictitious integrable system approach

We review the fictitious integrable system approach which predicts dynamical tunneling rates from regular states to the chaotic region in systems with a mixed phase space. It is based on the introduction of a fictitious integrable system that resembles the regular dynamics within the regular island. We focus on the direct regular-to-chaotic tunneling process which dominates, if nonlinear resonances within the regular island are not relevant. For quantum maps, billiard systems, and optical microcavities we find excellent agreement with numerical rates for all regular states.

nlin.CD↗

arXiv2010

Regular-to-Chaotic Tunneling Rates: From the Quantum to the Semiclassical Regime

We derive a prediction of dynamical tunneling rates from regular to chaotic phase-space regions combining the direct regular-to-chaotic tunneling mechanism in the quantum regime with an improved resonance-assisted tunneling theory in the semiclassical regime. We give a qualitative recipe for identifying the relevance of nonlinear resonances in a given $\hbar$-regime. For systems with one or multiple dominant resonances we find excellent agreement to numerics.

nlin.CD↗

arXiv2009

Quality factors and dynamical tunneling in annular microcavities

The key characteristic of an optical mode in a microcavity is its quality factor describing the optical losses. The numerical computation of this quantity can be very demanding for present-day devices. Here we show for a certain class of whispering-gallery cavities that the quality factor is related to dynamical tunneling, a phenomenon studied in the field of quantum chaos. We extend a recently developed approach for determining dynamical tunneling rates to open cavities. This allows us to derive an analytical formula for the quality factor which is in very good agreement with full solutions of Maxwell's equations.

physics.optics↗