Clinical Trial Details
— Status: Completed
Administrative data
NCT number |
NCT05673473 |
Other study ID # |
17025334 |
Secondary ID |
|
Status |
Completed |
Phase |
|
First received |
|
Last updated |
|
Start date |
January 1, 2016 |
Est. completion date |
January 1, 2020 |
Study information
Verified date |
January 2023 |
Source |
Hvidovre University Hospital |
Contact |
n/a |
Is FDA regulated |
No |
Health authority |
|
Study type |
Observational
|
Clinical Trial Summary
Radiographical images in Adolescent idiopathic scoliosis (AIS) can have a potential
radiation-induced oncogenic effect. In this study, the investigators aim to compare a
fluoroscopic imaging technique (LFT) with traditional radiographs for scoliosis (ORT), to see
if LFT is adequate for clinical evaluation of AIS and having a lower radiation dose.
Method Image quality will evaluated for LTF and ORT of phantom images and images from 3D
printed models of AIS. The investigators will measure primary physical characteristics of
noise, contrast, spatial resolution, SNR, and CNR. Three independent raters will evaluate the
images by observer-based methods of ICS and VGAS. Radiation doses will be evaluated by DAP
measurements. Two raters will perform measurements of 6 radiographic parameters for the LFT
images of AIS
Description:
The present study aims to evaluate whether a LFT is adequate as a clinical radiographic for
initial evaluation and/or the subsequent follow-up monitoring of AIS by evaluating image
quality, reliability, and agreement and measuring radiation dose.
Materials and methods Study design This study was divided into two parts: an experimental
part and a clinical part. The aim of the experimental part was to evaluate the physical
characteristics of image quality and observer-based evaluation of the LFT and ORT using a
pediatric trunk phantom and to measure the radiation exposure. Image quality was also
evaluated for radiographs of two "3D-printed" models of scoliotic spines (3DPSs), where the
agreement and reliability for LFT and ORT were determined. The aim of the clinical part was
to evaluate the inter-rater reliability of the actual clinical examinations for LFT
radiographs of AIS from our outpatient clinic. Ideally, "agreement" should be determined on
the clinical radiograph using both the LFT and ORT techniques. However, this would require
"simultaneous" double examinations, thus exposing the participants to excess radiation. This
was deemed ethically unviable; instead, we performed double examinations on the 3DPS.
Reliability and agreement were evaluated as proposed by Langensiepen et al.2 This study was
evaluated and approved by the local Research Ethics Committee (Journal number: 17025334).
Imaging system setups of the LFT and the ORT The examinations were performed in either the
posteroanterior or anteroposterior projection for the ORT and LFT, respectively. The patient
was standing facing the image intensifier/detector or radiation tube with extended hips and
knees and with their feet 10 cm apart. Two lead aprons were placed in the interscapular
(mammary) and sacral (genitals) regions. The distances to the radiation tube (source to
detector) were 100 cm for LFT and 230 cm for ORT. For the LFT, two images of the thoracic and
lumbar spine were necessary to cover the thoracolumbar spine. For ORT, only one exposure was
needed. The LFT was performed on a DelftDI D2RS system with fluoroscopic exposure of 4 pulses
per second/frame per second, and the ORT was performed using a digital Carestream
DRX-Evolution system with automatic exposure control. The following acquisition parameters
were assessed: tube potential (85 kVp, density set at low for LFT and 71 kVp, density set at
0 for ORT), both with grid, using manual exposures (LFT) and automated mA selection exposure
control (ORT) and no additional filtration for LFT and additional filtration (1 mm Al + 0.1
mm Cu) for ORT. The pixel size of 0.139 mm for ORT gives a maximum resolution of 3.6 lp/mm.
The pixel size of 0.159 mm for LFT gives a maximum resolution of 3.1 lp/mm.
Evaluators In the experimental part, observer-based assessments were performed once by 3
evaluators. Evaluator 1 was an experienced (22-year practice) pediatric orthopedic surgeon,
evaluator 2 was an experienced (16-year practice) pediatric orthopedic surgeon, and evaluator
3 was a reporting radiographer with 10 years of practice. Evaluator 1 conducted all analyses
for the primary physical characteristics. For the 3DPS, measurements of CA and classification
according to Nash and Moe (NM) were conducted three times as single sessions at least 1 week
apart. In the clinical part, the clinical radiographs were analyzed in a single session by
evaluators 1 and 2. Evaluators were blinded to patient identity and clinical information. All
image evaluations were performed on a PACS system (Impax 6.4.0, Agfa ® HealthCare, Mortsel,
Belgium) on a 3-megapixel viewing station by the 3 evaluators separately.
Part 1. Evaluation of image quality of the LFT and ORT imaging systems The pediatric trunk
phantom The Pediatric Whole Body Phantom "PBU-70" was examined radiographically for image
quality. We chose only the phantom torso of a 4-year-old child with a height of 105 cm with
life-size, full-body anthropomorphic measurements with embedded soft tissue substitutes of a
synthetic skeleton, lungs, liver, mediastinum, and kidneys (phantom).16 We performed 25
radiographic examinations with both techniques, where the phantom was placed and repositioned
for every recording.
The two 3D-printed AIS models Radiographs of the two 3DPSs with small and severe lumbar
scoliosis were recorded using both techniques. We performed 1 set of radiographs with both
techniques. The three evaluators performed 3 measurements of CA and NM at least 1 week apart.
Inter- and intra-rater reliability for the three evaluators were assessed using analysis of
interclass correlation as well as the mean absolute difference (MAD), standard error of
measurement (SEM), and Bland-Altman plots for the mean differences with additional analysis
for systematic differences. We defined accuracy from direct measurement with a protractor for
medical purposes. This was seen as a surrogate measure of overall accuracy. The phantom and
3DPSs are shown in Figure 1, and the radiographs are shown in Figure 2.
Figure 1. The pediatric whole-body phantom "PBU-70" and one of the two 3D-printed scoliosis
models.
Figure 2. Radiographs using the LFT (darker images) and the ORT (brighter images) of the
3D-printed scoliosis models (top four left) and the phantom (top two right). Radiographs
(ORT) of scoliosis with regions of interest (below left) and clinical LFT radiographs (below
right).
Primary physical characteristics of image quality evaluation: noise, contrast, and SNR We
examined the imaging characteristics of the LFT and ORT radiographs of the phantom and the
two 3DPSs. Twenty-five consecutive radiographs of the phantom using both techniques were
evaluated. The objective primary physical characteristics were the contrast, random noise,
signal-to-noise ratio, and contrast-to-noise ratio. The contrast was defined as the relative
signal difference between the two predefined locations of the bony vertebral spine and the
surrounding adjacent tissue of the spine, namely, a square region of interest of the
vertebral spine, where the upper and lower pedicles and endplates were included, along with a
similar region of interest of the adjacent soft tissue, where no bony soft tissue was
included (see Figure 2). Random noise was defined as the fluctuations of the signal over the
image when uniformly exposed, as expressed by the standard deviation.14,17 SNR was defined as
the ratio of the signal (defined as the signal of the object) divided by the standard
deviation (of the background).17 CNR was defined as the ratio of the signal (defined as the
difference of the signal of the object and the background) divided by the standard deviation
(of the background). Figure 2 shows these regions of interest and radiographs of the phantom,
the 3DPSs, and the clinical radiographs using the LFT.
Observer-based methods of image quality evaluation using the visibility of anatomical
structures For ethical reasons, we only performed observer-based evaluations on the images
from the phantom and the 3DPSs since we did not want to expose patients to double
examinations of the ORT and LFT. We utilized two observer-based methods for comparison of
image quality, namely, scores of the image criteria (ICS) and visual grading analysis
(VGAS).14,18 We compared a reference LFT image to all 25 ORT images as well as an ORT image
to all 25 LFS images. The reference images were selected randomly.14 Using the ICS, the
absolute level of image quality of specific structures was determined by evaluating a
specific structure compared to the same structure in the reference image by the observer;
thus, the observer's decision threshold was constant. The task of the observer was to decide
whether the specific structure was superiorly visually reproduced1 or inferiorly visually
reproduced when compared to the reference (0) (the criterion). We calculated the ICS for the
lumbar (L3) and thoracic (Th6) vertebrae as ICS=∑Ii=1∑Cc=1 ∑Oo=1 Fi,c,o IxCxO where Fi,c,o =
fulfillment of criterion c for image i and observer o. Fi,c,o = 1 if criterion c is
fulfilled; otherwise, Fi,c,o = 0, I = number of images, C = number of criteria, and O =
number of observers. The number of images assessed for both the LFT and ORT was 25 (I = 25),
the number of evaluators was 3 (O = 3), and the numbers of criteria were 6 for the thoracic
spine and 7 for the lumbar spine (C = 6 for the thoracic spine and C = 7 for the lumbar
spine) in accordance with the chosen CEC criteria.
We also graded the appearance using the VGA method by grading the visibility of specific
anatomical structures between images of the two radiographic techniques according to
Månsson.18 The visibility was graded according to a five-level scale: clearly inferior
visibility of a specific structure in the image compared to the same structure in the
reference image (-2), slightly inferior to (-1), equal to (0), slightly better than (+1), or
clearly better than (+2). We calculated the VGAS for the lumbar (L3) and thoracic (Th6)
vertebrae as VGAS=∑Ii=1∑Ss=1 ∑Oo=1 Gi,s,o IxSxO where Gi,s,o = grading (-2, -1, 0, +1, or +2)
for image i, structure s, and observer o; I = number of images; S = number of structures; and
O = number of observers. The number of images assessed for both the LFT and ORT was 25 (I =
25), the number of evaluators was 3 (O = 3), and the numbers of criteria were 6 for the
thoracic spine and 7 for the lumbar spine (C = 6 for thoracic spine and C = 7 for lumbar
spine) in accordance with the chosen CEC criteria. The specific evaluated (vertebral)
structures were in accordance with the European guidelines on Quality Criteria for Diagnostic
Radiographic Images in 1990 and 1996.19-21 The CEC lumbar spine criteria, 1990, were chosen
for the thoracic evaluation of the images since they focused on osseous vertebral spine
structures as well as adjacent soft tissues. For the lumbar evaluation, the CEC lumbar spine
criteria, 1996, were chosen since they also included an evaluation of the sacroiliac joints.
Radiation dose The recording systems of LFT and ORT measured the dose area product (DAP) by
an integrated DAP meter, and these were stored with the images. Monte Carlo calculations
using X-ray dosimetry software (PCXMC, version 2.0; Stuk, Helsinki, Finland) were utilized to
determine the effective radiation doses. This was based on the recorded DAP and the principal
patient size of the phantom as well as technical and geometric exposure parameters.
Part 2: Evaluation of AIS radiographs using the LFT for clinical reliability Clinical routine
radiographs using LFT Low-dose fluoroscopic technique was our regular routine clinical
radiographic method for AIS for 3 years. One hundred thirty-six adolescent patients with AIS
were included as participants. The sex ratio F:M was approximately 2:1. The average age was
13.4 years (range 6-17). We retrieved 342 LFT images of 136 patients with AIS. Two
independent evaluators performed measurements of six radiographic parameters once and
separately, where they were blinded to the clinical data and previous evaluations. The
parameters were the CA, the level of the upper and lower vertebrae used for determining CA,
the NM, and the Metha angles on the left and right sides at the apex vertebrae. Table 1 shows
the radiographic characteristics of the participants. Figure 2 illustrates the standard
clinical LFT radiographs.
Table 1. Distribution of participants according to the classification of King and Moe. Type
1: an "S"-shaped deformity, in which both curves are structural and cross the CSVL (midline),
with the lumbar curve being larger than the thoracic curve. Type 2: an "S"-shaped deformity,
in which both curves are structural and cross the CSVL, with the thoracic curve being larger
than or equal to the lumbar curve. Type 3: major thoracic curve in which only the thoracic
curve is structural and crosses the CSVL. Type 4: long "C"-shaped thoracic curve in which the
fifth lumbar vertebra is centered over the sacrum and the fourth lumbar vertebra is tilted
into the thoracic curve. Type 5: double thoracic curve.
Dextro convex Sinistro convex Type 1 7 29 Type 2 10 0 Type 3 24 5 Type 4 22 28 Type 5 2 1
Unclassifiable 8 Statistical analyses Reliability was assessed using the interclass
correlation coefficient (ICC). In part 1 of the two 3DPS models, inter- and intra-reliability
for CA and NM were assessed using a 2-way mixed model for consistency for the 3 evaluators
for their 3 separate measurements. Agreement was defined as the MAD, SEM, and Bland-Altman
plots for the mean differences with additional analyses of one-way t-test and logistic
regression for significant and systematic differences, respectively. In part 2 of the
clinical evaluation, the single measurements of CA and the other 5 radiographic parameters of
the two evaluators were assessed using a 2-way mixed model of ICC for absolute agreement for
inter-rater reliability.
Inter- and intra-rater reliability assessed by the ICC was considered with the following
limits of agreement: poor; 0.0-0.20 slight; 0.21-0.40 fair; 0.41-0.60 moderate; 0.61-0.80
substantial; and 0.81-1 almost perfect. Independent sample t-tests or Mann-Whitney
nonparametric tests were conducted for differences in SNR and CNR, differences in image
quality, and radiation dose. This depended on if normal distribution was present. This was
tested by Kolmogorov-Smirnov and Shapiro-Wilk tests and evaluation of QQ plots. We considered
a p-value of <0.05 as a significant result. ICC and other statistical evaluations were
performed using IBM SPSS Statistics, version 22 (IBM©, Chicago, Il, USA).