Matthias Walle Postdoctoral associate in bone imaging, AI, and scientific software

Projects

These projects reflect the way I approach scientific computing: build tools that are methodologically serious, realistic about messy imaging data, and practical enough that collaborators can run them, inspect them, and extend them.

software imaging hr-pqct ct slicer
Timelapsed HR-pQCT
Timelapsed HR-pQCT Longitudinal remodelling analysis pipeline

A highlighted project and the main current codebase for automated analysis of time-lapsed HR-pQCT scans, including multistack support, registration, transform handling, fused outputs, and remodelling metrics.

hr-pqct longitudinal analysis
spineFE-benchmark
spineFE-benchmark Vertebral strength benchmark from CT

Reproducible code and supporting resources for benchmarking linear and nonlinear finite element models of vertebral strength across the thoracolumbar spine using density-calibrated CT data.

ct finite-element benchmark

More Work

Multistack Registration
wallematthias/MultistackRegistration  ·   · 
A refactored package for longitudinal HR-pQCT processing with explicit transform semantics, single-resampling workflow design, and YAML-driven configuration.
registration hr-pqct python
MotionScoreCNN
wallematthias/MotionScoreCNN  ·   · 
A command-line tool for automated and semi-manual motion grading of HR-pQCT scans using an ensemble of deep convolutional neural networks and visual diagnostics.
hr-pqct deep-learning quality-control
BoneQuantMRI
wallematthias/BoneQuantMRI  ·   · 
A MATLAB tool for analyzing bone MRI images with ROI-based segmentation, phantom calibration, visualization, and export workflows for quantitative image analysis.
mri matlab quantitative-imaging
ImageGuidedStrain
wallematthias/ImageGuidedStrain  ·   · 
A project for multi-resolution Demons registration on 3D images and von Mises strain estimation, following the strain-analysis approach described by Zwahlen et al.
registration strain image-analysis

Data

Some project outputs also ship with public datasets and benchmark resources. The current public dataset section lives on the Datasets page and is set up to expand over time.