Data from: On the partial volume effect in magnetic particle imaging
Data files
Feb 05, 2025 version files 1.66 GB
-
Good_et_al_PVE_Study_Data_-_2024.zip
1.66 GB
-
README.md
8.51 KB
Abstract
This dataset contains raw and processed data used to evaluate the partial volume effect (PVE) in magnetic particle imaging (MPI), an emerging tomographic imaging modality that visualizes superparamagnetic iron oxide nanoparticle (SPION) distributions with high sensitivity and quantitative accuracy. The dataset includes raw MPI images in NRRD and DICOM formats, optical images of scanned phantoms in TIFF format, and processed Excel files with quantitative analyses. Data were obtained using Feraheme and VivoTrax+ tracers scanned with various phantom geometries (prisms, cylinders, and capillaries) and characteristic sizes. Additional experiments included scans of VivoTrax+ phantoms across multiple concentrations and at two independent sites. MPI image processing was performed using a reported Wiener filter, with outputs such as signal recovery coefficients, signal-to-noise ratios, and voxel-based reconstructions.
The dataset is organized by model geometry and includes both raw data and processed results, enabling analysis of signal distribution, peak and plateau signal values, and image resolution effects. Researchers can use these data to study the PVE in MPI, evaluate the impact of finite resolution on signal quantification, and develop methods to sharpen MPI images or correct for PVEs. The files are compatible with standard data visualization and analysis tools such as 3DSlicer, Horos, and MATLAB, while Excel files are ready for direct interpretation.
This dataset has broader relevance for advancing quantitative imaging techniques in MPI and beyond by providing insights into the influence of object size and shape on imaging accuracy. The dataset is provided for non-commercial scientific purposes.
README: On the partial volume effect in magnetic particle imaging
Name: Carlos M. Rinaldi-Ramos
Institution: University of Florida
Email: carlos.rinaldi@ufl.edu
Name: Hayden J. Good
Institution: University of Florida
Email: hayden.good34@gmail.com
Dataset Overview
This dataset supports the publication "On the Partial Volume Effect in Magnetic Particle Imaging" and includes data on tracer concentration, image-object size relationships, and iron estimation accuracy in magnetic nanoparticle studies. The experiments examine the partial volume effect (PVE) in Magnetic Particle Imaging (MPI) and include raw imaging data, STL files of 3D printed models used and the sample bed designed to hold them, optical photos of filled models prior to MPI measurement, and processed meta-data used in the publication.
Dates of Data Collection
Data was collected from the summer of 2023 until the spring of 2024.
Funding
This work was supported by the National Institute for Biomedical Imaging and Bioengineering under award number R01EB031224, the National Cancer Institute under award number R21CA26353, and the National Institute of Neurological Disorders and Stroke under award number R21NS125089.
Data Sources
All data were derived from MPI imaging samples collected from 2023 - 2024 by the authors.
Recommended Citation
Good, Hayden J. et al. On the partial volume effect in magnetic particle imaging. Physics in Medicine and Biology. In review. 2024.
Description of the data and file structure
Multiple files in this repository are .csv, .dcm, and .nrrd file types. Users are not provided with the appropriate scripts to analyze the raw imaging date due to proprietal information; however, users are provided with the file outputs.
Some csv files have NaNs, which were created accidentally during xlsx to csv conversion and have no significance to the data. Their presence simply means that there was a gap in the data that was filled.
Acronyms and important naming
HS - High sensitivity (scan mode)
HR - High resolution (Scan mode)
Std - Standard (scan mode)
RC - Recovery Coefficient
FWHM - Full-width at half-maximum
Native - 'Native' and minimally processed scan data
Sharpened - 'Sharpened' processed scan data
Wiener Filter - Native data that was transformed through a Wiener filter
Parent folder structure
10mmX Prisms - Feraheme
These files are associated with rectangular prism models with fixed x dimension of 10mm and variable dimension of z dimension filled with tracer solution of Feraheme.
10mmX Prisms - Vivotrax+
These files are associated with rectangular prism models with fixed x dimension of 10mm and variable dimension of z dimension filled with tracer solution of Vivotrax+.
10mmX Prisms - MI
These files are associated with rectangular prism models with fixed x dimension of 10mm and variable dimension of z dimension filled with tracer solution of Vivotrax+, filled and imaged at Magnetic Insight campus.
5mmX Prisms
These files are associated with rectangular prism models with fixed x dimension of 5mm and variable dimension of z dimension filled with tracer solution of Vivotrax+.
15mmX Prisms
These files are associated with rectangular prism models with fixed x dimension of 15mm and variable dimension of z dimension filled with tracer solution of Vivotrax+.
Cylinders
These files are associated with cylindrical models with varying x and z (radial) dimensions filled with tracer solution of Vivotrax+.
Small Cylinders
These files are associated with capillary tubing with varying x and z (radial) dimensions filled with tracer solution of Vivotrax+.
Empty Bed scans
These files are associated with empty bore scans of the MOMENTUM scanner.
[Fe] Cylinders
These files are associated with a set of 14mm Diameter cylinder models filled with carying concentrations of Vivotrax+.
[Fe] Prisms
These files are associated with a set of 14x10mm rectangular prism models filled with carying concentrations of Vivotrax+.
RELAX Scans
These files are associated with the RELAX relaxometry output of the UF MOMENTUM Scanner for both Vivotrax+ and Feraheme batches used for scans.
STL files for models
These files are STL files for the models used in the study.
Model specific folder structure
Most model folders listed above have the following format:
DICOM Native
These are the 'native' imaging data in DICOM format from the MOMENTUM scanner associated with the model specific folder.
Nrrd Native
These are the 'native' imaging data in NRRD format from the MOMENTUM scanner associated with the model specific folder.
DICOM Sharpened
These are the 'sharpened' imaging data in DICOM format from the MOMENTUM scanner associated with the model specific folder.
Nrrd Sharpened
These are the 'sharpened' imaging data in NRRD format from the MOMENTUM scanner associated with the model specific folder.
Excel files
These are the excel files of processed meta data used in the manuscript. Sheets are
CSV files
These are the same excel files and sheets of processed meta data used in the manuscript
Photos
These are the optical photos taken from the MOMENTUM MPI scanner prior to the model being scanned.
Description of file names
MPI output names
Each DICOM and NRRD file is formatted where the first three numbers are the scan number of the day, followed by scan mode (HS - High sensitivity, HR - High resolution, Std - Standard), followed by user interface (AUI - advanced user interface (this means the data had extra processing. this data was not presented in the manuscript), SUI - Standard user interface (this data was published)), and finally the model variable identifier.
For example, 004_HS_SUI_30mm.dcm refers to the 4th scan of the day, performed in High sensitivity scan mode using the standard user interface, and the 30mm model of the parent model (in this case the author was using the 10mmX Prisms - Feraheme model folder as an example).
Excel file format
The file begins with the original MPI folder format (may not match published folder name) associated with the data, then a descriptor of what is in the excel file. Descriptors include Full scan data (digitized full data set in excel sheet), Linescans (extracted line scans of interest from the full scan data), and RC 2x2mm (recovery coefficient data calculations where the center of the model was found and statistics regarding the center was extracted). If the file name includes "HS" it was scanned under high sensitivity mode; if not, it was scanned under Std mode.
CSV file format
The CSV files are simply extractions of the Excel files listed above in a format that does not require Microsoft software to load.
Description of Excel file naming
Full Scan data
Each sheet refers to the model name (Std_SUI_30mm is the 30mm model scanned under Std SUI settings)
Linescan Results
Z-scans Auto
This sheet refers to the linescans (automatically found) in the z direction. The first column is the imaging spacing dimensions, and the following columns are specific to each model.
Z-scans Manual
This sheet refers to the linescans (manually selected) in the z direction. The first column is the imaging spacing dimensions, and the following columns are specific to each model.
X-scans Auto
This sheet refers to the linescans (automatically found) in the x direction. The first column is the imaging spacing dimensions, and the following columns are specific to each model.
X-scans Manual
This sheet refers to the linescans (manually selected) in the x direction. The first column is the imaging spacing dimensions, and the following columns are specific to each model.
FWHMzs or FWHMxs
This sheet refers to the calculated Full-width at half-maximums (FWHMs) of the linescans from the above models, and the FWHM of several linescans adjacent to the central FHWM.
RC 2x2mm
This data depicts the center of each model. the extracted center is shown in the table, then the mean, stdev, and maximum of the matrix of data are provided. the Sheets are associated with the scan mode and model.
Others
There are some unique excel files that are self-explanatory. If any questions arise, please contact either primary author.
Methods
Data were collected using magnetic particle imaging (MPI) scanners to evaluate the partial volume effect (PVE) in imaging superparamagnetic iron oxide nanoparticle (SPION) distributions. Two tracers, Feraheme and VivoTrax+, were imaged in 3D-printed phantoms of varying geometries and sizes, including prisms, cylinders, and capillaries. MPI scans were conducted under controlled conditions at two independent sites (University of Florida and Magnetic Insight, Inc.). Additional experiments included imaging phantoms with varying tracer concentrations. A Wiener filter was applied to process select images of VivoTrax+ phantoms to enhance resolution. Quantitative analyses were performed to estimate signal recovery coefficients, signal-to-noise ratios, image size, and voxel-based signal distributions.