nifti_dynamic 0.2.0

A package for dynamic NIFTI analysis

nifti_dynamic 🧠⚡

Efficient TAC extraction and Patlak analysis of HUGE PET arrays in NIfTI format. Works with gzipped 4D PET images (saving 90% disk space vs DICOM) 📦. All algorithms load partial chunks to prevent memory crashes when handling massive arrays (440×440×645×69×4 bytes = 32GB).

Implementation based on Andersen TL, et al. Diagnostics 2024;14(15):1590. doi:10.3390/diagnostics14151590

Installation 💾

pip install nifti_dynamic

Important: Always ensure indexed_gzip is installed for significantly faster reading of gzipped NIfTI arrays (.nii.gz) ⚡

Performance ⏱️

• Automatic extraction of 4 VOIs: 30 seconds
• TAC for single organs: 10 seconds
• TACs for all 114 TotalSegmentator organs: 10 minutes
• Full voxel Patlak for 440×440×645×69 array: 30 seconds

CLI Usage 🚀

Extract Input Function from Aorta

extract_input_function \
  --pet dpet.nii.gz \
  --totalseg totalseg.nii.gz \
  --output ./results

Extract TACs for All Organs

extract_tacs \
  --pet dpet.nii.gz \
  --segmentation totalseg.nii.gz \
  --output ./tacs

Voxel-wise Patlak Analysis

voxel_patlak \
  --pet dpet.nii.gz \
  --input-function input_fun.csv \
  --output ./patlak

Downsample Dynamic PET (2x)

resample_pet \
  --pet dpet.nii.gz \
  --output dpet_2x.nii.gz

Python API Usage

For programmatic access, see example.py for detailed Python API usage.