Improved detection of endovascular stent fractures with photon counting computed tomography: A phantom study.

OBJECTIVE: To evaluate the ability of photon-counting detector computed tomography (PCD-CT) to detect visceral endovascular stent fractures compared to EID-CT, and to assess its impact on subjective image quality and quantitative image characteristics. METHODS: Five identical balloon-expandable covered stents were placed in an abdominal phantom and scanned using a PCD-CT (Naeotom Alpha) and an EID-CT scanner (Somatom Force, Siemens Healthineers). PCD-CT images were reconstructed using two slice thickness (0.2 mm: ultra-high resolution [UHR]; 0.4 mm: standard resolution [SR]) and both medium sharp (Br56) and sharp kernel (Br72). 1024 matrix were applied in UHR reconstructions, while both 512 and 1024 matrix were used in SR reconstructions. EID-CT images were reconstructed using medium sharp kernel (Br59) with 512 matrix. Stents were scanned intact and then sequentially fractured from Type 1 (single strut) to Type 4 (complete transverse fracture with displacement). Two independent readers assessed diagnostic accuracy and image quality, while an automatic algorithm evaluated strut sharpness, blooming, and noise levels. RESULTS: PCD-CT significantly improved stent fracture detection, with the highest overall diagnostic accuracy achieved by PCD-CTUHR-Br72 (96 %) and PCD-CTSR-Br72 with 1024 matrix (92 %), compared to PCD-CTSR with 512 matrix and EID-CT (50 %-56 %). PCD-CTBr72 identified more than 80 % of Type 1 and Type 2 fractures, whereas neither PCD-CTSR with 512 matrix nor EID-CT detected any single-strut fractures. PCD-CTBr72 with 1024 matrix also provided superior image quality, enhanced sharpness, and reduced blooming artifacts (all p < 0.001). CONCLUSIONS: PCD-CTBr72 with 1024 matrix significantly enhances the detection of visceral endovascular stent fractures, particularly single-strut fractures that remain undetectable on EID-CT.