Assessing SPECT-CT Scanners for Lung Cancer Screening
Abstract
Purpose
Explore the possibility of using SPECT-CT scanners for Lung Cancer Screening (LCS).
Methods
Lung nodule specific image quality characteristics were assessed using the CTLX1 phantom on eight SPECT-CT scanners from two vendors. The CTLXI phantom is made up of three cylindrical modules placed at 0, 100, and 200 mm from the isocentre of the scanner and are used to assess the following RSNA-QIBA small lung nodule profiles at these three locations: edge enhancement (≤1.05), point spread function (≤1.5mm^3), 3D resolution aspect ratio (≤2.0), image noise standard deviation (≤50HU), CT number bias (≤35HU), and spatial warping root means squared error (≤0.33mm) to ensure quantitative measurements taken across vendors/models, and over time, are consistent. The scanners were also assessed against the Provincial LCS program requirements of ≥16 slice, 1-1.25 mm contiguous slices, and scan time <12s to ensure a viable patient breath hold. Typical patient scan length was determined from our screening population. Helical pitch and reconstruction kernel options were exhaustively explored.
Results
Of the eight scanners, two had insufficient slices, and one vendor was excluded entirely because there was no way to scan under 12s based on the typical patient scan length of 365mm. The remaining vendor passed all the Provincial requirements but was not able to pass all the RSNA-QIBA tests even after extensive protocol alterations.
Conclusion
None of the SPECT-CT scanners in this study were technically capable of LCS. One vendor failed to achieve the required scan time. The other vendor’s failure was a surprise, since 64 slice and greater versions of that machine easily passed the RNSA-QIBA requirements. However, the 16-slice version could not pass the 3D resolution or Spatial Warping requirements for the phantom target at 200mm from isocentre. Interestingly, RSNA-QIBA tests passed for axial scanning, but the scan time was too long.