From e3ab749b285759573140c48f06c28462b8b8e802 Mon Sep 17 00:00:00 2001 From: Kris Thielemans Date: Tue, 22 Oct 2024 08:21:49 +0000 Subject: [PATCH] add README for Mediso_NEMA_IQ_lowcounts --- .../Mediso_NEMA_IQ_lowcounts/README.md | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) create mode 100644 SIRF_data_preparation/Mediso_NEMA_IQ_lowcounts/README.md diff --git a/SIRF_data_preparation/Mediso_NEMA_IQ_lowcounts/README.md b/SIRF_data_preparation/Mediso_NEMA_IQ_lowcounts/README.md new file mode 100644 index 0000000..b702c6f --- /dev/null +++ b/SIRF_data_preparation/Mediso_NEMA_IQ_lowcounts/README.md @@ -0,0 +1,17 @@ +# Mediso_NEMA_IQ_lowcounts +Aside from usual steps + +## +Copy VOIs from high count data, but with adjustement of size. KT +did this via STIR utilities (but can be done via SIRF as well) +``` +for f in VOI*.hv; do + zoom_image --template data/Mediso_NEMA_IQ_lowcounts/OSEM_image.hv data/Mediso_NEMA_IQ_lowcounts/$f $f; +done +``` +## BSREM +Given large oscillations with subsets, I reduced the number of subsets from 12 to 6 (in `dataset_settings.py`) but increased the initial step size. +``` +python -m SIRF_data_preparation.get_penalisation_factor --dataset=Mediso_NEMA_IQ_lowcounts --ref_dataset=Mediso_NEMA_IQ -w +python -m SIRF_data_preparation.run_BSREM Mediso_NEMA_IQ_lowcounts --initial_step_size=.5 --relaxation_eta=.0105 +``` \ No newline at end of file