

DOS Kongressen 2017 ·
205
Single- and dual energy QCT around acetabular cups in
total hip arthroplasty using 3-dimensional segmentation
Bo Mussmann, Poul Erik Andersen, Trine Torfing, Søren Over-
gaard
Dept. of radiology, Odense University Hospital; Dept. of Orthopaedic surgery
and traumatology, Odense University Hospital
Background:
Bone density measurements around hip implants are challenged
by artifacts and the complex anatomy of the acetabulum. We developed 3D
segmentation software and used dual energy CT to reduce artifacts.
Purpose / Aim of Study:
To test the between-scan agreement and reliability
of segmentation software and to compare bone mineral density (BMD) mea-
surements in single- and dual energy CT (SECT and DECT)
Materials and Methods:
24 male patients with total hip arthroplasty (12 ce-
mented and 12 uncemented) were scanned and rescanned using SECT and vir-
tual monochromatic DECT images. 3D- ROIs were defined slice-by-slice and
BMD was calculated adjacent to the acetabular cup.
Findings / Results:
Mean BMD for SECT was 411mg/ccmwith a between scan
difference of 21 mg/ccm, p=<0.0001 in the uncemented cup. DECT showed a
mean BMD of 153 mg/ccm with a difference of 10 mg/ccm, p<0.0001. Around
the cemented cup the mean BMD for SECT was 523 mg/ccm with a between-
scan difference of 14 mg/ccm, p=0.25 and 186 mg/ccm in DECT with a differ-
ence of 6 mg/ccm, p=0.15. ICC was >0.95 with more narrow limits of agree-
ment in DECT compared with SECT. Computed tomography dose index (CTDI)
was 25% higher with DECT and subjective image quality was better in SECT.
Conclusions:
Equally reliable BMD measurements adjacent to acetabular ce-
mented and uncemented cups can be performed using the segmentation
software. The absolute between-scan agreement was better in DECT. Virtual
monochromatic DECT with fast kVp switching may be beneficial in estimating
BMD adjacent to metal implants, but radiation dose and image quality should be
taken into consideration. BMD cannot be measured interchangeably with SECT
and DECT.
No conflicts of interest reported
157.