Author: Christopher P. Favazza, Andrea Ferrero, Taylor Froelich, Forrest Linch, Scott Thompson 👨🔬
Affiliation: Mayo Clinic 🌍
Purpose: Photon counting detector (PCD) CT is a promising tool to enhance pre-procedural planning and intra-procedural guidance for prostate artery embolization due to its unsurpassed ultra-high spatial resolution (UHR). This study investigated the impact of different reconstruction approaches on the accuracy of automatic vessel tracing when using UHR-PCD CT with a commercial embolization guidance software.
Methods: Three patients were scanned using a commercial PCD CT (NAEOTOM Alpha.Peak, Siemens Healthineers), following standard clinical protocol for pelvis CT angiography (120 kV, 165 IQ level, 120x0.2mm collimation). The nominal reconstruction utilized Tlow data (20-120 keV), voxel size of 0.3x0.3x0.2 mm3, Bv56 reconstruction kernel. Three alternative reconstructions were separately explored: sharper kernel (Bv64), enhanced iodine contrast (50 keV monoenergetic data) and material specific data (Iodine map). Three vessels in each patient were identified and manually traced on the nominal series to provide a ground truth (GT) reference. Subsequently, an automatic vessel trace was generated for each of the 4 reconstructions, using the same start and end points as the ground truth. Trace data were exported and minimum distances between each automatic trace-point and the ground truth were measured. From these data, root mean squared error (RMSE) was calculated for each auto trace relative to the ground truth.
Results: The embolization guidance software was able to load and utilize the PCD CT reconstructions to automatically trace vessels. Compared to the ground truth vessel, Iodine maps consistently outperformed the other reconstruction approaches, yielding 0.9mm median RMSE and 1.9mm inner quartile range. The other three reconstruction approaches yielded similar results 5.2, 3.9, 7.3mm median RMSE and 20.1, 15.8, 14.4mm inner quartile range for Bv56, Bv64, and Mono50keV, respectively.
Conclusion: This investigation showed how iodine-specific reconstructions from pre-procedural PCD CTA could be leveraged to improve the performance of automatic vessel tracing algorithms for prostate artery embolizations.