Design
The study was designed as a cross-sectional exploratory pilot study and included 10 participants. Analysis of stretch sensor data was performed blinded for the results of pMRI measurements.
Participants
Randomly recruited voluntary subjects from members of staff at the Radiology Department, Frederiksberg Hospital, Copenhagen and Institute of Sports Medicine Copenhagen, Bispebjerg Hospital gave informed consent to participate in the study, which was approved by the local ethics committee (protocol H-2-2012-151). Eligibility criteria were: age 20–50 years; no contraindications to MRI. Exclusion criteria were: self-reported foot pain or known foot disorder such as osteoarthritis, inflammatory arthritis or congenital foot deformity. The same cohort participated in a previous study in which reproducibility of pMRI measurements was assessed [16].
MRI procedure
MRI was conducted using a positional MRI-system (0,25 T G-scan, Esaote SpA, Genoa, Italy). The participants were scanned in both supine (SUP) and 90 degrees standing position (SP). The applied MRI protocol included gradient echo scout (slice thickness: 5 mm, field of view (FOV): 280x280mm, scan time 39 s) and Steady-State Free-Precession 3D (SHARC) (TE: 14 ms, TR: 28 ms, FA: 35, FOV: 230×230; Matrix: 256×256, Scan time: 6 m 54 s) sequences. As a precaution to counteract the symptomatic orthostatic hypotension or syncope during scanning a crural pneumatic pumping device was applied to stimulate the venous backflow [18]. The pMRI scanning procedure has previously been described in detail [16]. Briefly, during scanning in supine position (SUP) the talo-crural joint was positioned in a 90° angle to the long axis of the tibia. In the standing position (ST) subjects were positioned in a one-legged stance and instructed to stand with equally distributed pressure on the heel and anterior plantar sole. The non-loaded extremity rested on the housing of the scanner magnet. The foot was oriented parallel to the scanner patient table (Fig. 1). The distance from the table to the medial aspect of the foot was measured allowing the foot to be positioned similarly for stretch sensor measurements. Scanning in the standing position was repeated with addition of 10 % body weight (ST + W) carried by the participant in a backpack.
Image analyses
All image analyses were performed in a commercially available DICOM viewer (Osirix, Pixmeo SARL, Bernex, Switzerland). We have previously described the measurements of NVH and MNP in detail [16]. Briefly, owing to the 3D nature of the MRI sequences the imaging planes could be adjusted in the multiplanar reconstruction (MPR) module of the DICOM viewer in a standardised fashion before actual measurements of NVH and MNP (Fig. 2). Time consumption to perform all measurements of NVH and MNP was below 10 min per subject. One of the co-authors a 3rd year resident radiologist (PH) performed all radiological measurements reported in the present study. NVH and NMP were measured in SUP, ST and ST + W.
Stretch sensor measurements
The newly developed strain sensor is based on a dielectric electroactive polymer material produced by Danfoss PolyPower. The material acts as an elastic capacitive material that is strainable in one direction. The sensor measures length by means of change in electrical current. It is mechanically stable, reusable, portable and resistant to perspiration from the foot. Reliability has been tested and demonstrates ICC > 0.76 for barefoot measurements [17]. The stretch sensor was attached to the skin surface as described in previous studies assessing reproducibility of the device [17, 19]: One end was attached ≈ 20 mm behind the medial malleolus and secured by a Velcro strap. The other end was attached with adhesive tape ≈ 20 mm behind the prominence of the navicular bone (Fig. 3). Measurements sampled over 15 s were performed in supine, standing and standing with 10 % extra bodyweight. In supine position a cushion under the plantar sole was used to stabilize the foot in 90 degrees ankle dorsiflexion. It was ensured that the cushion exerted only very slight pressure on the foot sole. The baseline strain on the sensor upon attachment cannot be controlled for rigorously, therefore measurements are only valid by calculating the difference between standing and supine. In standing position the foot was placed in a position identical to the position during pMRI scanning (Fig. 3). Stretch sensor measurements of the dynamic movements of the medial plantar arch were performed during barefoot walking. Walking distance was the same for all participants (20 m). Stretch sensor measurements were sampled over two contiguous walking sessions. A previous study showed that navicular drop varies from step to step [19]. Therefore, stretch sensor data were collected across two walking trials to ensure an average of more than 20 steps. The normal walking speed of each participant was established and a metronome was used to ensure this pace was consistently demonstrated throughout the gait analysis.
Data analysis
Data were analysed using a custom-written Matlab script. Heel strike for each stance phase was manually determined using data from the accelerometer and gyroscope, which has excellent reliability and validity [19]. Afterwards, a custom written algorithm determined the maximal magnitude of navicular motion for each stance phase. The average of stretch sensor measurements over two walking sessions was used for statistical analyses. Stretch sensor measurements were performed immediately after the pMRI scanning.
Blinding procedure
All radiologic measurements of NVH and MNP were performed by one radiologist (PH). The stretch sensor recordings were performed by another co-author (SES). Stretch sensor measurements were assessed by a third co-author (MR) blinded for subject ID and pMRI results.
Calculation of navicular bone position changes
Change (Δ) in NVH and MNP was calculated between SUP and ST (ΔNVHST; ΔMNPST) and between ST and ST + W (ΔNVHST+W; ΔMNPST+W).
Additionally, “total” positional change of the navicular bone (ΔTPC) was calculated based on the assumption of a combined medial and caudal navicular displacement between SUP and ST. ΔTPC was expected to be better suited than NVH or MNP for direct comparison with stretch sensor measurements, as the stretch sensor measures the resultant of both vertical and medial displacement, which cannot be individually differentiated by the sensor [17]. ΔTPC was estimated by use of the equation of Pythagoras:
$$ {\mathrm{a}}^2 + {\mathrm{b}}^2 = {\mathrm{c}}^2 $$
Total positional change of the navicular bone would equal:
$$ \Delta \mathrm{TPC} = \surd \left(\Delta \mathrm{N}\mathrm{V}{\mathrm{H}}^2+\Delta \mathrm{M}\mathrm{N}{\mathrm{P}}^2\right) $$
Statistics
ΔNVH, ΔMNP and ΔTPC are presented in mm and reported as mean with 95 % confidence intervals (CI). Data were visually assessed for normal distribution by QQ-plots and further examined by a Kolmogorov-Smirnov test. Data on navicular position by pMRI were normally distributed. Changes between scanning positions were assessed using a two-tailed paired student’s t-test. An alpha level of 0.05 was used. Since ΔTPC in essence has a baseline value =0 no student’s t-test was performed for this parameter.
ΔTPC was compared both to the delta values from the stretch sensor measurements under static loading conditions between SUP and ST and to the dynamic stretch sensor measurements during walking. Since stretch sensor data during walking were non-normally distributed a two-tailed Spearman’s Rho correlation analysis was used.