Evaluation of prosthetic complications and tumor recurrence with the single energy metal artifact reduction algorithm in patients with hip tumor prostheses

Background: To evaluate the effectiveness of the single energy metal artifact reduction (SEMAR) algorithm with a second-generation 320-row multi-detector computed tomography (MDCT) on complications and tumor recurrence detection in patients with hip tumor prostheses. Methods: From February 2016 to June 2019, 31consecutive patients with tumor prostheses of the hip joint underwent CT scans. Lesions were conrmed by histology or clinical and imaging follow-up. Images were reconstructed using 2 methods: iterative (IR) algorithm alone and SEMAR algorithm (IR+ SEMAR). Two radiologists graded the image quality visually by a 6-point (from 0 to 5) ordinal scale. Standard deviations (SD) of CT attenuation value dened as the artifact index (AI) were compared between the two reconstructed methods. Paired sample t-test was adopted to compare the AI values on IR and SEMAR images. Wilcoxon matched-pairs signed rank test was performed to compare the visual scores on IR and SEMAR images. A p- value less than 0.05 was considered statistically signicant. Results: The artifacts of the SEMAR images were reduced compared to the Non-SEMAR images (113.94 ±128.54 vs 35.98 ± 53.75HU (cid:0) t=2.867, P < 0.05). 20 and 16 more lesions were detected by observer 1 and observer 2 with SEMAR algorithm respectively. The mean scores of lesions without SEMAR were 1.39 ± 1.45 (observer 1) and 1.55± 1.34 (observer 2); with SEMAR, the scores were signicantly higher, 4.42±0.56 (z=-4.752, p < 0.001) and 4.54± 0.72 (z=-4.837, p < 0.001) respectively. Conclusion: The SEMAR algorithm can effectively reduce metal artifacts in patients with hip tumor prostheses and increase the diagnostic accuracy of prosthetic complications and tumor recurrence. form of dark and bright streaks or bands due to scattering, x-ray beam hardening and photon starvation, partial volume effect and patient motion, signicantly degrading image quality and diagnostic accuracy [1–3] . Metal artifact reduction (MAR) reconstruction algorithms that can ideally reduce or even eliminate image artifacts are urgently needed [4, 5] . In the past decades, various MAR algorithms from different vendors have been reported for clinical use to reduce metal artifacts. The single-energy metal artifact reduction (SEMAR) algorithm was clinically used on a second-generation 320-row multi-detector computed tomography (MDCT) recently. Previous studies have demonstrated the effectiveness of SEMAR in patients with dental lling, hip prosthesis and metal coil [6–8] . Although several authors have demonstrated the satisfactory effect of SEMAR on the visualization of bone-prosthesis interface, periarticular soft structures and pelvic organs in total hip prosthesis patients, our study mainly focused on the value of SEMAR on the follow-up of tumor recurrence and prosthetic complications in patients with hip tumor prosthesis.


Background
With the number of hip tumor patients undergoing joint replacement gradually increased in recent years, imaging follow-up is important in determining whether tumors recur and potentially complications happen. Metallic hardware can cause severe artifacts on computed tomography (CT) images in the form of dark and bright streaks or bands due to scattering, x-ray beam hardening and photon starvation, partial volume effect and patient motion, signi cantly degrading image quality and diagnostic accuracy [1][2][3] . Metal artifact reduction (MAR) reconstruction algorithms that can ideally reduce or even eliminate image artifacts are urgently needed [4,5] . In the past decades, various MAR algorithms from different vendors have been reported for clinical use to reduce metal artifacts. The single-energy metal artifact reduction (SEMAR) algorithm was clinically used on a second-generation 320-row multi-detector computed tomography (MDCT) recently. Previous studies have demonstrated the effectiveness of SEMAR in patients with dental lling, hip prosthesis and metal coil [6][7][8] . Although several authors have demonstrated the satisfactory effect of SEMAR on the visualization of bone-prosthesis interface, periarticular soft structures and pelvic organs in total hip prosthesis patients, our study mainly focused on the value of SEMAR on the follow-up of tumor recurrence and prosthetic complications in patients with hip tumor prosthesis.
The right side was involved 14 times, and the left side was involved 17 times. Detailed information of patients was showed in Table1. Tumor recurrence and prosthetic complications were con rmed by histology or clinical and imaging follow-up. This study was approved by the hospital ethics committee, and the need for signed informed consent was waived.
Image acquisition and reconstruction CT images were acquired by a 320-row MDCT scanner (Aquilion ONE, Canon Medical Systems, Otawara, Japan) in helical mode. The axial scan parameters were as follows: tube voltage, 135 kV; tube current, automatic exposure control (SURE exposure 3D, Canon Medical Systems, Otawara, Japan); detector collimation, 320 × 0.5mm; gantry rotation time, 1.0 s; matrix 512×512. Contrast agent with an iodine concentration of 300 mg/ml (Ultravist 300, Bayer AG, Berlin, Germany) were used in all the patients after plain scan. The total volume of contrast material (ml) was determined by multiplying the body weight (kg) by two, with an upper limit of 100 ml. It was injected at a rate of 2.5 ml/s via a 22-gauge intravenous catheter placed in an antecubital vein. The enhanced CT scan began 70 s after the initiation of contrast injection. Adaptive IR algorithm (AIDR 3D, Canon Medical Systems) and IR + SEMAR algorithm (Canon Medical Systems) were applied to the raw data for each patient. Standard soft tissue kernel (FC08) and bone kernel (FC30) were used for the reconstructions with IR alone and IR+SEMAR respectively. The SEMAR images were reconstructed according to the published methods [8,9] .

Image analysis
Subjective image assessment Images with IR/SEMAR were evaluated by two board-certi ed radiologists with 15 and 5 years of experience blindly and independently. Axial enhanced images with a 40/400 Houns eld unit (HU) window width/level setting for soft tissue evaluation and 400/2200 HU window width/level setting for bone evaluation were showed randomly and on a high-resolution 20-inch monitor (M21, Jusha Nanjing, China). The visual scores of lesions was graded as follows: 0= completely obscured; 1=marked artifacts with questionable recognition; 2=faint anatomic recognition; 3=recognition with low con dence; 4=recognition with medium con dence; 5=recognition with high con dence [8,10] .

Objective image assessment: comparison of artifact index (AI)
For each patient, circular regions of interest (ROI) were drawn within the lesions on enhanced images with different reconstructive methods and kept the same size at the same axial level, approximately 50-80mm 2 . To quantify the image noise, the standard deviation (SD) of CT attenuation value de ned as AI was measured [11,12] .

Statistical analysis
Statistical analysis was performed by SPSS 20.0 software (SPSS Inc. Chicago, IL, USA). Paired sample t-test was used to compare the AI values of IR and SEMAR images. Wilcoxon matched-pairs signed rank test was adopted to compare the scores between the two reconstruction methods. Intra-class correlation coe cient (ICC) was calculated to evaluate inter-observers , variability. A p-value less than 0.05 was considered statistically signi cant.

Results
Subjective image assessment On SEMAR images, 6 patients with periprosthetic soft tissue infection, 8 patients with uid collection, 2 patients with heterotopic ossi cation, 2 patients with periprosthetic osteolysis, and 13 patients with tumor recurrence were diagnosed. On IR images, 11 and 15 lesions were detected by observer 1 and observer 2 respectively, and the numbers increased to 31 and 31 with SEMAR algorithm. The SEMAR algorithm made more lesions identi ed (an increase of 64.5 % and 51.6 % by observer 1 and observer 2). The mean score of lesions with IR was 1.39 ± 1.45 and 1.55± 1.34 for observer 1 and observer 2, respectively; with SEMAR, the corresponding scores were signi cantly higher to 4.42±0.56 and 4.54± 0.72 (z=-4.752, p < 0.001 and z=-4.837, p < 0.001).
Inter-observers , agreement was considered to be excellent (ICC=0.967

Discussion
There are many potential complications of hip arthroplasty, such as leg length discrepancy, prosthetic loosening, subluxation or dislocation and abnormalities of bones or soft tissues(periprosthetic fracture and osteolysis,periprosthetic soft tissue uid collections, infection and heterotopic ossi cation) [13] . CT can be used to supplement the radiograph to assess the symptomatic hip prostheses especially for the peri-implant uid collection and soft-tissue infection. However, artifacts from the prosthetic hip joint make the follow-up remain challenging. SEMAR is an algorithm that works with the single-energy mode and is applicable to routine data retrospectively [6,14,15] . There have been no published articles regarding the usefulness of the SEMAR on evaluating post-hip tumor prosthesis complications and tumor recurrence. Although Gondim Teixeira et al. reported the effectiveness of the SEMAR in patients with hip prostheses, the focus of their study was the visualization of speci c periarticular normal muscles and pelvic organs rather than post-arthroplasty lesions and their study population was not tumor patients [8] .
Infection is the most serious complication, happening in 0.3 to 1.7% of hip arthroplasties and the frequency is increasing [16] . In our study, the most common prosthetic complication was periprosthetic uid collection (8/18) and the following is soft tissue infection (6/18), which was higher than that reported in previous studies. Inclusion criteria (patient with symptomatic hip tumor prosthesis) may be account for this phenomenon. Our results showed that the SEMAR algorithm can potentially increase the performance of CT in the follow-up of hip tumor prostheses. Compared with the IR algorithm, the SEMAR technique can substantially reduce artifacts of hip tumor prostheses and improve lesion detectability. Not only did both observers detect more anomalies, but also the diagnostic con dence also increased (visual scores were signi cantly improved). Inter-observer agreement was considered to be excellent between the experienced and young radiologists. Moreover, the SEMAR algorithm has been used mostly on volume scan clinically at present.
The 320-row MDCT system yields 16-cm-range volume data [7] , which limits the clinical applicability when the range is wider than 16 cm. We showed successful MAR effect with SEMAR by helical scan and the depiction of each anomaly was signi cantly improved.  [17] can lead to decrease of artifacts in ROIs both adjacent to the head of the prostheses and also between the prostheses. Jessie Y Huang also reported that MARS was the most effective method in reducing streak artifacts of the pelvic phantom with hip prosthesis among O-MAR, GE's monochromatic imaging of dual-energy CT and MARS [18] . Projection-based MAR algorithms, such as O-MAR, SEMAR, are primarily introduced to compensate for photon starvation phenomenon, whereas virtual monochromatic images obtained with dual-energy CT mainly reduce artifacts caused by beam hardening effect. Bright and dark band artifacts caused by photon starvation from large metallic hardware and from metal with high atomic numbers, for example hip prosthesis and endovascular coil, are too strong for the dual-energy CT alone to be removed [3,19] . Hence, the projection-based MAR algorithms become more effective than the monochromatic imaging of dual-energy CT in this situation. In this study, although the overall image quality was greatly improved by SEMAR, there was still a small remnant of artifacts, suggesting the SEMAR technique with Canon Medical Systems could not eliminate metal artifacts. Moreover, newly developed artifacts can be seen when the SEMAR algorithm was used for several reasons according to previous studies [17,20] . These new artifacts were especially evident when the SEMAR was combined with IR. New high-density artifacts were also observed on the SEMAR images in our study. However, we were not able to make comparison among the SEMAR and other algorithms/techniques in patients with hip tumor prostheses. This is work that should be performed in future.
Our study has several limitations. First, only a small number of patients were included. Second, considering the retrospective nature of this study, further prospective studies are needed to verify the effectiveness of SEMAR algorithm.

Conclusion
Based on this study, we can conclude that SEMAR algorithm can effectively reduce the metal artifacts of hip tumor prostheses and is ideal in the evaluation of prosthetic complications and tumor recurrence. The current study was approved by the Institutional Ethics Committee of the First A liated Hospital of Sun Yat-Sen University, and the need for signed informed consent was waived.

Consent for publication
Not applicable.

Availability of data and materials
The dataset supporting the conclusions of this article is included within the article. Data and materials during the current study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable.