fidoptCT-vs-PET-Ventilation-Imaging, attenuation CT to Galligas ventilation PET
The low-dose attenuation CT of one free-breathing session of a patient with lung cancer, resampled onto the PET grid, and as target the Galligas ventilation PET divided by its own 99th percentile inside the body.
10 test caseslungsource CT, target PET
Read before comparing (6)
- Images. Only the PET's attenuation CT is used; the breath-hold CT and the respiratory phases are not. The CT is resampled onto the PET grid through the scanner frame by a cubic B-spline and rounded to whole Hounsfield units.
- Target. The PET in Bq/mL divided by its 99th percentile inside the mask, so that 1 is the case's high ventilation: not Bq/mL or SUV. That percentile varies about 300-fold between patients (37 to 17,400 Bq/mL) with the activity inhaled, which no CT predicts, so the absolute scale is not scored. The supplied SUV conversion is not valid for an inhaled tracer.
- Range. The dynamic range of 1 sets the scale of PSNR and SSIM. Values above 1, one voxel in a hundred of the mask by construction and up to 29.5 at an airway deposit of CT-PET-VI-11, are kept, not clipped.
- Mask. The body on the resampled CT, PET zeros included. It is not a lung segmentation, so whole-mask errors include tissue outside the lungs.
- Alignment. The CT is a snapshot and the PET averages free breathing. CT-PET-VI-06's CT and PET have different frame-of-reference UIDs; their scanner coordinates are kept after geometry checks, with no landmark registration to confirm voxelwise alignment. The provider warns that CT-PET-VI-15's PET truncates the inferior lung.
- Overlap. Lung250M's two TCIA-Ventilation pairs (its cases 094 and 097) are breath-hold CTs of two patients of this collection. Neither dataset hides what the other shows: Lung250M hides its landmarks, this one the ventilation PET, which Lung250M does not hold. Both are kept as they are.
No method has a result on every case of this dataset yet.
Case by case
Click a score to look at that result: the target image, and over it the method's synthetic image. Underlined scores have their image; for the others you see the case without the result.
About the data
The first test caseClick to look at every case.
- Test cases
- Ten test patients drawn with seed 2026 and nine development patients: all nineteen patients with the attenuation CT, one development case fewer than usual. CT-PET-VI-07 lacks that CT and is not a case.
- Truth
- target image as released the same-session Galligas ventilation PET, divided by its own 99th percentile inside the body mask: a relative ventilation map on its native grid
- Source
- www.cancerimagingarchive.net/collection/ct-vs-pet-ventilation-imaging
doi.org/10.7937/3ppx-7s22 - Licence
- CC BY 4.0
9 cases for tuning, never tested
CT-PET-VI-19, CT-PET-VI-02, CT-PET-VI-15, CT-PET-VI-20, CT-PET-VI-17, CT-PET-VI-11, CT-PET-VI-09, CT-PET-VI-10, CT-PET-VI-08