Dataset Open Access

Data from Study: Respiratory Motion Correction of PET using MR-Constrained PET-PET Registration

Balfour, Daniel

This dataset contains the data used to arrive at the conclusions in the research article Respiratory Motion Correction of PET using MR-Constrained PET-PET Registration, by Balfour et al [BioMedical Engineering OnLine 2015, 14:85].

This study was based upon motion-affected PET images simulated from real dynamic MR image volumes, simulated and reconstructed using the Software for Tomographic Image Reconstruction ("STIR", see This dataset includes data from MR scans of 4 healthy volunteers (male, aged 22-33).

Three types of data are provided, which should be sufficient for repeating the findings of the study:

  • Reconstructed PET image volumes, split into 6 respiratory bins ("gates") for each simulation
  • The dynamic 3D MR volumes used to derive the respiratory motion of each volunteer
  • Text files outline which dynamics have NOT been used for PET simulation - these are the ones used to make the motion model in the study

These MR volumes were registered and combined with the head-foot position of the right hemidiaphragm to form a respiratory motion model, which was subsequently used to constrain PET to PET image registration, attempting to correct for the motion in the PET images.

For more detailed information regarding the method, please refer to the article.

The PET data is split into several sub-categories:

  • Volunteer ID (4 possibilities, anonymised)
  • Lesion position (9 possibilities - see article for locations)
  • Lesion diameter, in millimetres (10 or 14 mm)
  • Respiratory gate number, ranging from 1 (most inhaled) to 6 (most exhaled)

Note that there are two types of each simulation: with motion, and without motion. These are included in the respective zip files for each volunteer ID.


The file names reflect which image the file contains according to a number of substrings. As outlined in the description, there are several tags associated with each PET simulation. Specifically these are: - volnN, where N refers to the volunteer identification number (1-4) - motionless simulations contain the tag "stat", short for "static" - LesPosN, where N refers to the lesion position (1-9) - LesDiamN, where N refers to the lesion diameter in mm (10 or 14) - gateN, where N is the respiratory bin Note that all PET image intensities are multiplied by 100 to avoid problems with numerical precision. All PET filenames contain the substring "x100" to remind the user of this. Each set of PET images are zipped according to volunteer ID and motion status: whether motion is included (substring "wMotion") or not included ("noMotion"). Note that the latter was used for validation purposes. All MR dynamic images are contained within "", which follow the tagging system mentioned above - only volunteer ID and the dynamic number are included. Note that there are a total of 35 dynamic volumes for each volunteer, except for volunteer 4, for which there are 70 volumes available. These have been included for completeness.
Files (1.7 GB)
Name Size
622.4 MB Download
170 Bytes Download
136.8 MB Download
136.7 MB Download
160 Bytes Download
132.3 MB Download
131.9 MB Download
160 Bytes Download
123.8 MB Download
123.9 MB Download
160 Bytes Download
125.1 MB Download
125.1 MB Download
All versions This version
Views 318318
Downloads 1,3951,395
Data volume 400.5 GB400.5 GB
Unique views 299299
Unique downloads 211211


Cite as