BACKGROUND
- Dynamic sequences of Contrast-Enhanced (DCE) imaging techniques (CT, MRI…) associate 2 main issues: i) a poor signal-to-noise ratio, ii) movements of patients.
- Image segmentation and registration are the solution but available products are rigid and/or elastic registration
- Specific issue encountered in medical image registration is the presence of many classes of pixels having different properties of intensity change
To circumvent the above issues and facilitate clinicians interpretations, there is a need for automation of advanced image registration to allow quantitative analysis and decrease variability
HOW IT WORKS
Dynamic Imaging is a classifying registration technique for the estimation of contrast enhancement curves (temporal intensity evolution of pixels) of DCE sequences.
Dynamic Imaging improves both classification and registration simultaneously in an iterative way.
Abdominal DET-CT scan sequence:
Before registration After registration
KEY BENEFITS vs. STATE OF THE ART
- Computes enhancement curves on pixel classes instead of single pixels: faster, more robust to noise than in pixel-based techniques; reduces the computation time
- Improves 3D classification of anatomical structures in functional compartments
- Robustness: computes 80-450 images as opposed to a couple with available techniques
- Accurately distinguish dynamic properties of tissues
DEVELOPMENT STATUS
Validated using ground truth data (sequences of DCE-CT/IRM from abdomen and thorax)
APPLICATIONS
New CAD software in Medical Imaging Equipment (MRI, Scanner…) for dynamic contrast-enhanced medical imaging
For diagnostic and new drugs development