News | Volker J Schmid

Freie Mitarbeiterstelle an der Arbeitsgruppe Bioimaging

Wed, 25 Jul 2018 10:18:57 +0000

Aktualisierung: Die Stelle ist vergeben!

An der Ludwig-Maximilians-Universität München (LMU) ist vorbehaltlich der endgültigen Finanzierung am Institut für Statistik, Arbeitsgruppe Bioimaging und räumliche Statistik ab sofort für die Dauer von zunächst 3 Jahren die folgende Stelle zu besetzen:

Wissenschaftliche/r Mitarbeiter/in
(E13 TV-L, Vollbeschäftigung, befristet, Qualifizierungsstelle)
im Bereich Bayesianischer Bildverarbeitung
im Rahmen eines beantragten Kompetenzzentrum für Machine Learning

Die Ausschreibung richtet sich vorrangig an Doktorandinnen und Doktoranden. Sie kann aber auch mit einer Postdoktorandin oder einem Postdoktoranden besetzt werden.

Einstellungsvoraussetzungen

‐ abgeschlossenes wissenschaftliches Hochschulstudium (Staatsexamen, Master, Diplom oder vergleichbarer Abschluss) in Statistik oder einem eng verwandten Gebiet mit mindestens gutem Ergebnis

‐ Kenntnisse in Bayesianischer und/oder räumlicher Statistik

– Programmierkenntnisse, idealerweise in R und C++

‐ sichere Kenntnisse der deutschen und englischen Sprache

‐ Bereitschaft, ernsthaft und engagiert an einem Vorhaben der eigenen wissenschaftlichen Qualifizierung (Promotion) bzw. Weiterqualifizierung zu arbeiten

Bewerbungsadresse

Bewerbungen mit den üblichen Unterlagen (Lebenslauf, Zeugnisse, …) senden Sie bitte ausschliesslich in elektronischer Form als ein PDF-File an Prof. Volker Schmid, Mail: volker.schmid@lmu.de

Für weitere Informationen wenden Sie sich formlos an Volker Schmid.

Hinweis

Ihr Arbeitsplatz befindet sich in zentraler Lage in München und ist sehr gut mit öffentlichen Verkehrsmitteln zu erreichen. Wir bieten Ihnen eine interessante und verantwortungsvolle Tätigkeit mit guten Weiterbildungs‐ und Entwicklungsmöglichkeiten. Schwerbehinderte Bewerber/Bewerberinnen werden bei ansonsten im Wesentlichen gleicher Eignung bevorzugt. Die Bewerbung von Frauen wird begrüßt.

Fitting large-scale structured additive regression models using Krylov subspace methods

Sat, 23 Jul 2016 19:11:16 +0000

Just accepted for publication at Computational Statistics & Data Analysis: Fitting large-scale structured additive regression models using Krylov subspace methods by Paul Schmidt, Mark Mühlau and Volker Schmid.

Abstract:

Fitting regression models can be challenging when regression coefficients are high-dimensional. Especially when large spatial or temporal effects need to be taken into account the limits of computational capacities of normal working stations are reached quickly. The analysis of images with several million pixels, where each pixel value can be seen as an observation on a new spatial location, represent such a situation. A Markov chain Monte Carlo (MCMC) framework for the applied statistician is presented that allows to fit models with millions of parameters with only low to moderate computational requirements. The method combines a modified sampling scheme with novel accomplishments in iterative methods for sparse linear systems. This way a solution is given that eliminates potential computational burdens such as calculating the log-determinant of massive precision matrices and sampling from high-dimensional Gaussian distributions. In an extensive simulation study with models of moderate size it is shown that this approach gives results that are in perfect agreement with state-of-the-art methods for fitting structured additive regression models. Furthermore, the method is applied to two real world examples from the field of medical imaging.

Bayesian mixed-effects model for the analysis of a series of FRAP images

Thu, 26 Mar 2015 16:26:34 +0000

The binding behavior of molecules in nuclei of living cells can be studied through the analysis of images from fluorescence recovery after photobleaching experiments. However, there is still a lack of methodology for the statistical evaluation of FRAP data, especially for the joint analysis of multiple dynamic images. We propose a hierarchical Bayesian nonlinear model with mixed-effect priors based on local compartment models in order to obtain joint parameter estimates for all nuclei as well as to account for the heterogeneity of the nuclei population. We apply our method to a series of FRAP experiments of DNA methyltransferase 1 tagged to green fluorescent protein expressed in a somatic mouse cell line and compare the results to the application of three different fixed-effects models to the same series of FRAP experiments. With the proposed model, we get estimates of the off-rates of the interactions of the molecules under study together with credible intervals, and additionally gain information about the variability between nuclei. The proposed model is superior to and more robust than the tested fixed-effects models. Therefore, it can be used for the joint analysis of data from FRAP experiments on various similar nuclei.

Published in Statistical Applications in Genetics and Molecular Biology

Pattern Recognition and Signal Analysis in Medical Imaging

Mon, 05 May 2014 08:21:45 +0000

Medical imaging is one of the heaviest funded biomedical engineering research areas. The second edition of Pattern Recognition and Signal Analysis in Medical Imaging brings sharp focus to the development of integrated systems for use in the clinical sector, enabling both imaging and the automatic assessment of the resultant data.

Since the first edition, there has been tremendous development of new, powerful technologies for detecting, storing, transmitting, analyzing, and displaying medical images. Computer-aided analytical techniques, coupled with a continuing need to derive more information from medical images, has led to a growing application of digital processing techniques in cancer detection as well as elsewhere in medicine.

This book is an essential tool for students and professionals, compiling and explaining proven and cutting-edge methods in pattern recognition for medical imaging.

Link

Table of Content:

Introduction
Feature Selection and Extraction
Subband Coding and Wavelet Transform
The Wavelet Transform in Medical Imaging
Genetic Algorithms
Statistical and Syntactic Pattern Recognition
Foundations of Neural Networks
Transformation and Signal-Separation Neural Networks
Neuro-Fuzzy Classification
Specialized Neural Networks Relevant to Bioimaging
Spatio-Temporal Models in Functional and Perfusion Imaging
Analysis of Dynamic Susceptibility Contrast MRI Time-Series Based on Unsupervised Clustering Methods
Computer-Aided Diagnosis for Diagnostically Challenging Breast Lesions in DCE-MRI

Neue Veröffentlichung: Spatially regularized estimation for the analysis of DCE-MRI data

Sat, 15 Mar 2014 00:00:00 +0000

Abstract

Competing compartment models of different complexities have been used for the quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging data. We present a spatial elastic net approach that allows to estimate the number of compartments for each voxel such that the model complexity is not fixed a priori. A multi-compartment approach is considered, which is translated into a restricted least square model selection problem. This is done by using a set of basis functions for a given set of candidate rate constants. The form of the basis functions is derived from a kinetic model and thus describes the contribution of a specific compartment. Using a spatial elastic net estimator, we chose a sparse set of basis functions per voxel, and hence, rate constants of compartments. The spatial penalty takes into account the voxel structure of an image and performs better than a penalty treating voxels independently. The proposed estimation method is evaluated for simulated images and applied to an in vivo dataset.

More-compartment models for DCE-MRI

Tue, 11 Mar 2014 09:26:03 +0000

Standard compartment models for the analysis of Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) are based on one or two compartments, parts of the tissue, which exchange the contrast agent. However, these models are often to simplistic, especially when analysing DCE-MR images of cancerous tissue. On the other hand, models with more than two compartments suffer from redundancy issues, that is, the model parameters cannot be identified. Julia Sommer has worked on ways to make more-compartment-models identifiable by using spatial regularization methods. In two recent publications, she also shows how the number of compartments can be found.

Spatial two-tissue compartment model for dynamic contrast-enhanced magnetic resonance imaging. By Julia Sommer, and Volker J Schmid. Journal of the Royal Statistical Society: Series C. (LINK)

In the quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging compartment models allow the uptake of contrast medium to be described with biologically meaningful kinetic parameters. As simple models often fail to describe adequately the observed uptake behaviour, more complex compartment models have been proposed. However, the non-linear regression problem arising from more complex compartment models often suffers from parameter redundancy. We incorporate spatial smoothness on the kinetic parameters of a two-tissue compartment model by imposing Gaussian Markov random-field priors on them. We analyse to what extent this spatial regularization helps to avoid parameter redundancy and to obtain stable parameter point estimates per voxel. Choosing a full Bayesian approach, we obtain posteriors and point estimates by running Markov chain Monte Carlo simulations. The approach proposed is evaluated for simulated concentration time curves as well as for in vivo data from a breast cancer study.

Spatially regularized estimation for the analysis of dynamic contrast-enhanced magnetic resonance imaging data. By Julia Sommer, Jan Gertheiss, and Volker J Schmid. Statistics in Medicine. (LINK)

Competing compartment models of different complexities have been used for the quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging data. We present a spatial elastic net approach that allows to estimate the number of compartments for each voxel such that the model complexity is not fixed a priori. A multi-compartment approach is considered, which is translated into a restricted least square model selection problem. This is done by using a set of basis functions for a given set of candidate rate constants. The form of the basis functions is derived from a kinetic model and thus describes the contribution of a specific compartment. Using a spatial elastic net estimator, we chose a sparse set of basis functions per voxel, and hence, rate constants of compartments. The spatial penalty takes into account the voxel structure of an image and performs better than a penalty treating voxels independently. The proposed estimation method is evaluated for simulated images and applied to an in vivo dataset.