Analysis: Evaluate performance of rigid brain motion correction on three tracer datasets
Keyn34 opened this issue · 5 comments
Now that the ZUBAL case is sorted, evaluation of FALCON's capabilities concerning rigid brain motion correction will be assessed based on three tracer datasets (AMT, FDG, FET) with dice score evaluation.
- Collect ~10 datasets for each tracer
- extract full head mask for evaluation purposes
- tweak/prepare code to let FALCON run on all datasets
- gather dice scores before & after registration and visualize the metric
@Keyn34: Please let me know the tracers (along with their scanners), before you start the analysis!
Update: adjusted deadline due to delays (thanks, Ubuntu...)
Currently scanning through FET data.
25 datasets (5 AMT, 10 FET and 10 FDG [5 Siemens, 5 Philips]) were collected and delineated for evaluation purposes.
We will perform the evaluation soon.
Redoing the segmentations for FDG now:
Siemens starting frame: 25
Philips starting frame: 8
Additional ToDos:
- get framing for all datasets
- rerun evaluation pipeline on FDG set after segmentation individually
Tasks will be probably finished by Monday next week (11.07.2022).
Analysis was done so far, waiting for Philips framing.
Philips segmentations were done with anisotropic filtering to enhance the segmentation quality.