NiftyPET: High-throughput image reconstruction and analysis¶
NiftyPET is a software platform and a Python namespace package encompassing sub-packages for high-throughput PET image reconstruction, manipulation, processing and analysis with high quantitative accuracy and precision. One of its key applications is brain imaging in dementia with the use of amyloid tracers. See below for the description of the above amyloid PET image reconstructed using NiftyPET, superimposed on the MR T1 weighted image *.
Advanced quantitative evaluation of PET systems using the ACR phantom and NiftyPET software
NiftyPET includes two packages:
The core routines are written in CUDA C and embedded in Python C extensions to enable user-friendly and high-throughput executions on NVIDIA graphics processing units (GPU). For the scientific aspects of this software platform see section Scientific Output.
Although, NiftyPET is dedicated to high-throughput image reconstruction and analysis of brain images, it can equally well be used for whole body imaging. Strong emphasis is put on the data, which are acquired using positron emission tomography (PET) and magnetic resonance (MR), especially using the hybrid and simultaneous PET/MR scanners.
This software platform covers the entire processing pipeline, from the raw list-mode (LM) PET data through to the final image statistic of interest (e.g., regional SUV), including LM bootstrapping and multiple independent reconstructions to facilitate voxel-wise estimation of uncertainties.
The above dynamic transaxial and coronal images show the activity of 18F-florbetapir during the one-hour dynamic acquisition. Note that the signal in the brain white matter dominates over the signal in the grey matter towards the end of the acquisition, which is a typical presentation of a negative amyloid beta (Abeta) scan.
- Scientific Output
- NiftyPET Example
- Accessing and querying GPU devices
- DICOM anonymisation
- List-mode processing and motion detection
- Basic PET image reconstruction
- Dynamic image reconstruction
- Corrections for quantitative PET
- Phantom image reconstruction and registration
- Imaging initialisation
- List-mode histogramming
- NAC PET image reconstruction
- Generate \(\mu\)-map and NAC PET templates
- Stage I image registration
- Quantitative reconstruction for stage II registration
- Generate the resolution rod template
- Stage II image registration
- Compose the final \(\mu\)-map
- Reuse of registraion for VOI sampling
- Advanced analysis