Rodrigo

Current Research Interests

My research centers on developing mathematically grounded computational tools for analyzing medical images from various imaging modalities. These tools are not only intended to enhance image interpretation but also to optimize the acquisition process itself.

A core principle of my work is efficiency, aiming to build methods that can be directly used in clinical practice. I am highly motivated to bridge the gap between fundamental research and real-world clinical tools, which necessitates strong collaborations with industry.

I am particularly focused on:

  • Applying advanced artificial intelligence techniques to solve complex image analysis tasks
  • Optimizing image acquisition and processing for advanced MRI imaging modalities

Current Projects

1. Magnetic Resonance Elastography (MRE)

We aim to characterize the mechanical properties of the brain using MRE in various cohorts:

  • Parkinson’s disease
  • Alzheimer’s disease
  • Brain tumors
  • Multiple sclerosis
  • Healthy children and adults

Goals:

  • Characterize brain mechanics across age
  • Improve early diagnosis and treatment monitoring
  • Develop new techniques to extract more nuanced mechanical properties

2. Analysis of Brain Connectivity

We work with diffusion MRI (dMRI) to analyze white matter connectivity using tractography. Despite high sensitivity, existing methods suffer from low specificity.

Our focus:

  • Developing AI-based methods to enhance tractography pipelines
  • Improving accuracy and reliability of brain connectivity analysis

3. Synthetic MRI Aging

Aging effects confound image analysis in Alzheimer’s diagnosis. This project develops AI models to:

  • Simulate age progression or regression from a single structural MRI
  • Use diffeomorphic registration to preserve anatomical fidelity

Applications:

  • Prospective prediction of clinical progression
  • Retrospective inference of pre-morbid states

4. Image Analysis in Cancer

We use advanced AI techniques to analyze 3D radiological images across several imaging modalities:

  • Digital breast tomosynthesis
  • 3D ultrasound
  • PET
  • CT
  • MRI

Objectives:

  • Detection, segmentation, prediction, and synthetic generation of cancer-related imaging data

Previous Projects

1. Deep Segment

Funded by Eurostars, this project built a full pipeline for segmentation in medical/biological imaging:

  • Sample preparation
  • Scanning
  • Manual & automatic segmentation
  • Visualization and analysis
  • Support for large-scale studies and deep learning

2. Biomechanics at Interactive Speed

Biomechanical simulations are typically too slow for clinical use. We developed AI-based methods to estimate biomechanical parameters in near real-time, removing the need for computationally expensive simulations.

3. Analysis of Trabecular Bone Microstructure

Osteoporosis detection through high-resolution imaging (CBCT, HR-pQCT). Developed methods for:

  • Trabecular thickness analysis
  • Fabric tensor estimation
  • Trabecular classification

Validated against micro-CT data and supported by clinical studies to assess treatment effects. Funded by Eurostars.

4. Analysis of Blood Vessels

Aimed at non-invasive quantification of coronary artery stenosis and plaque characterization using:

  • CTA, MRA, and advanced MRI (Dixon, PC-MRI)
  • Robust AI methods for low-resolution, noisy data
  • Histopathological validation for clinical translation

5. Medical Image Analysis through Tensor Voting

We extended tensor voting for use in medical imaging modalities such as CT and MRI. Applications include:

  • Detection of vessel bifurcations
  • Identification of vortices in blood flow
  • Analysis of trabecular bone structure

Funded by The Swedish Research Council (VR).

Source Code

Available at: https://github.com/rodrigomorenokth

Notable Repositories:

  • Generalized Mean Intercept Length Tensor
  • Vesselness using the Ring Pattern Detector
  • Efficient Tensor Voting
  • Classification of Trabeculae
  • Distance Between Sets of Points

Other codebases are available upon request or will be published soon on GitHub.