Foot Scanner Reliability
The Reliability of the Associate Platinum Digital Foot Scanner in Measuring Previously Developed Footprint Characteristics: A Technical Note
M. Owen Papuga, MS, and Jeanmarie R. Burke, PhD
An ink pad and paper, pressure-sensitive platforms, and photography have previously been used to collect footprint data used in clinical assessment. Digital scanners have been widely used more recently to collect such data. The purpose of this study was to evaluate the intra- and interrater reliability of a flatbed digital image scanning technology to capture footprint data.
This study used a repeated-measures design on 32 (16 male 16 female) healthy subjects. The following measured indices of footprint were recorded from 2-dimensional images of the plantar surface of the foot recorded with an Associate Platinum (Foot Levelers Inc, Roanoke, VA) digital foot scanner: Staheli index, Chippaux-Smirak index, arch angle, and arch index. Intraclass correlation coefficient (ICC) values were calculated to evaluate intrarater, interday, and interclinician reliability.
The ICC values for intrarater reliability were greater than or equal to .817, indicating an excellent level of reproducibility in assessing the collected images. Analyses of variance revealed that there were no significant differences between raters for each index (P N .05). The ICC values also indicated excellent reliability (.881-.971) between days and clinicians in all but one of the indices of footprint, arch angle (.689), with good reliability between clinicians. The full-factorial analysis of variance model did not reveal any interaction effects (P N .05), which indicated that indices of footprint were not changing across days and clinicians.
Scanning technology used in this study demonstrated good intra- and interrater reliability measurements of footprint indices, as demonstrated by high ICC values. (J Manipulative Physiol Ther 2011;34:114-118)
Key Indexing Terms:
Flatfoot; Foot Diseases; Foot Deformities
Clinical measures of foot structure have been developed with the objective of quantifying particular aspects of foot geometry for purposes of diagnosis. One such measurement is the medial longitudinal arch height. It has been used in identifying those individual who have fallen arches (aka flat feet). The arch height is defined as the distance from the plantar supporting surface to the navicular tuberosity. The condition of flat feet may be due to underlying bone or ligament pathology or merely to abnormal development.1 Although reduced arch height is not always associated with a specific pathology or pain, it may be associated with abnormal biomechanics, which over time result in knee and/or ankle pathologies.2 Establishing the reliability of footprint measurement techniques is an initial step toward further developing the clinical relationship between the structural and functional characteristics of the foot.
The most direct approach to investigating underlying foot pathology is via diagnostic imaging. Measurements from a lateral radiograph view allow for the evaluation of the postural and structural integrity of the boney structures.3 Anthropometric data of the foot have also been collected based on palpation of boney landmarks.3-5 Arch height measured with palpation and sliding ruler has been shown to be highly correlated with lateral view radiographs taken of the same foot.4,5 Although much can be learned about boney structures via radiograph, little can be determined about the surrounding soft tissue. Unnecessary radiographic exposure should be avoided whenever possible; and therefore, several alternative methods of assessing foot structure, beyond the anthropometric measurement of arch height, have been developed, including foot casting,6 3-dimensional (3-D) photographic scans,7,8 and assessment of the footprint.9,10
Historically, ink pad and paper have been the methodology used to collect and measure footprint characteristics.11,12 In an attempt to modernize and improve upon the ink pad and paper technology and to potentially facilitate the ease of use in clinical environments, pressure-sensitive platforms,6,13 photography,14 and digital video13,15 have been introduced to collect and measure footprint characteristics. A direct application of the assessments of footprint characteristics in clinical environments relates to the prescription of foot orthoses. With respect to the prescription of custom-molded orthotic interventions, flatbed image scanning technology such as the Associate Platinum digital foot scanner (Foot Levelers Inc, Roanoke, VA) is being used to measure foot geometry. Determination of the reliability of footprint images collected with such flatbed image scanning technology is necessary to provide quality assurance that using such methodology in a clinical environment is appropriate.
Fig 1. The Staheli index (CD/EF): the ratio of the minimum width of the midfoot arch region to the maximum width of the rearfoot region. The Chippaux-Smirak (CD/AB): the ratio of the minimum width of the midfoot arch region to the maximum width of the forefoot region. The arch angle (∠GHI): the angle between the line connecting the most medial points of the heel and metatarsal regions and the line connecting the most lateral point on the medial foot border to the most medial point of the metatarsal region. Arch index (K/[J + K + M]): the area of the middle third of the toeless footprint to the total toeless footprint area.
The purpose of the study was to determine the reliability of a flatbed image scanning technology to capture footprint data using the Associate Platinum digital foot scanner. The following indices of footprint were calculated: Staheli index, Chippaux-Smirak index, arch angle, and arch index (Fig 1). Previous research documented that our selected indices could be measured with high reliability when using photography, videography, and ink; in addition, there is evidence of a correlation between these indices and arch height.11-15 Intra- and interrater reliability of scanning hardware and software to measure indices of footprint were assessed using intraclass correlation coefficients (ICCs).
The completed investigation included 32 healthy (16 male and 16 female) subjects: (29.9 ± 7.62 years, 169.7 ± 10.89 cm, 77.2 ± 17.52 kg.) The subjects self-reported that they were free of musculoskeletal disorders and that they were not experiencing any pain or medical conditions affecting their feet, ankles, knees, hips, or lower back pain. This study and all study protocols were approved by the New York Chiropractic College institutional review board before data collection. All subjects provided written informed consent.
Reliability was assessed by calculating ICCs for measurements of indices of footprint within clinicians and between clinicians in a 3-week, repeated-measures design.
Data Collection Procedures
There were 3 data collection sessions using the scanning technology. During each data collection session, 3 separate scans, by 3 separate clinicians, were taken of each foot while the subject was standing with feet shoulder-width apart and with equal weight on each foot, which follows the clinical protocol prescribed by Foot Levelers Inc. Each trial lasted approximately 30 seconds. Subjects were asked to step off the scanner between scans of each foot. On 2 subsequent visits, each approximately 1 week apart, the scanning protocol was repeated. At completion of the study, there were 9 scans of each foot for each participant, with a grand total of 576 images.
Image Analysis: Measurements of Indices of Footprint. Data were collected by the Associate Platinum digital foot scanner and custom software provided by Foot Levelers Inc allowing for the storage of JPEG images. Images of the bottom of the foot were taken, and the reflective brightness of the image was quantified as prescribed by the color replacement algorithm developed by Foot Levelers Inc. These colorreplaced (colorized) images were used to calculate the footprint indices. These indices include the Staheli, Chippaux-Smirak, arch angle, and arch index. A brief description of the calculation of each of these indices is shown in Figure 1. All direct image measurements were made using Image J software (version 1.37; National Institutes of Health, Bethesda, MD) by 3 independent raters. Software was calibrated to image dimensions before use; measurements used included lengths, angles, and areas. Manual contours of the contact area of each foot allowed us to calculate directly each of the indices.
Statistics. Reliability was determined by calculations of ICCs for each index of footprint. We used 3 raters to conduct image analysis of scanned footprints and as such needed to calculate intrarater reliability. Each rater calculated each index for a set of 8 colorized images, repeating the measurements 3 times to assess intrarater reliability (n = 24). The ICCs were calculated from a general linear repeated-measures analysis of variance (ANOVA) model (image repeated across the single factor of trial) using SPSS Statistical Software (SPSS v17.0, Chicago, IL). We then examined the interrater reliability. Each rater calculated each index for the same 5 subjects'data taken by one clinician (30 images: 5 subjects × 2 feet × 3 days). The ICCs were calculated from the general linear repeated-measures ANOVA model (image repeated across factors of rater and trial). The reliability of each index was then assessed between days (interday) and between clinicians (interclinician) for the entire data set (576 images). The ICCs were calculated from 2 separate general linear repeated-measures ANOVA models (trial × day; trial × clinician). Different benchmarks exist for the interpretation of ICC; one often-referred-to demarcation is that an ICC value of less than 0.40 indicates poor reproducibility, ICC values in the range 0.40 to 0.75 indicate fair to good reproducibility, and an ICC value of greater than 0.75 shows excellent reproducibility.16 A full-factorial ANOVA model was performed to reveal the presence of any interaction effects among trial, day, and clinician (trial × day, trial × clinician, trial × day × clinician). The presence of any interaction effects (P b .05) would indicate that indices of footprint were changing across days and clinicians as a function of trial.
Table 1 summarizes the intrarater reliability coefficients. The lowest reliability was found in the replication of the arch angle measurement, with one rater having an ICC of .817. However, all indices of footprint indicated excellent intrarater reliability.
Column 1 of Table 2 summarizes the interrater reliability coefficients. In this assessment of reliability, all indices of footprint, excluding the arch angle, have an ICC value greater than .80, which would indicate excellent interrater reliability. In addition, the ANOVA analyses revealed that there were no significant differences between raters for each index (P N .05).
Also contained in Table 2 are the ICC values for interday and interclinician reliabilities. The ICC values indicate excellent reliability in all of the footprint indices between days. The ICC values also indicate a high degree of reliability in all but one of the indices of footprint between clinicians, the arch angle. The full-factorial ANOVA model did not reveal any interaction effects (P N .05). Therefore, we can conclude that the indices of footprint were not changing across days and/or clinicians as a function of trial.
The ability to reliably measure indices of footprint collected with the Associate Platinum digital foot scanner was evident in the good to excellent ICC values obtained. Measurements of Staheli index, Chippaux-Smirak index, and arch index were reliable across trials, raters, days, and clinicians. Previous research into characteristics of footprint using varied approaches of collection has shown similar reliability in this set of indices.12-15 The arch index has been shown to be the most repeatable across such studies and also has been shown to have a high degree of correlation with measured arch height values.14,15 In a descriptive analysis of the published studies,12-15 the range of all 95% confidence intervals reported for the arch index was .194 to .239. The present study had a 95% confidence interval of .197 to .230, showing high agreement with previously published data. The fair to good interrater (.605), and interclinician (.689) ICC values obtained for the arch angle are consistent with the inherent measurement errors in calculating arch angle from a footprint. Although there were trial-to-trial variations in the measurement of arch angle, the mean arch angles were similar across days and clinicians.
The most likely explanation for the measurement of arch angle having reduced reliability between clinicians is due to inherent trial-to-trial variations in identifying anatomical landmarks on the footprint to calculate arch angle. Three clinicians asked the subjects to stand in a neutral quiet standing position for both the right and left foot; 3 separate “raters” were trained to make the measurements of arch angle. The arch angle measurement is prone to error induced by soft tissue variations and operator preference.15 The bias of operator preference involves choosing points on the image that are not adequately defined along the medial edge of the midfoot. Delineating across trials where the forefoot ends and where the midfoot begins is problematic with no boney landmark as a guide. The low interrater reliability substantiates the inherent trial-to-trial variations in identifying soft tissue landmarks on the footprint to calculate arch angle by different operators; and as such, this inherent measurement problem also manifests as low reliability between clinicians positioning the subject on the scanner.
The high day-to-day reliability of arch angle measurements reflects a bias in the calculation of ICC. The calculation of ICC depends upon a ratio of between-subject variability and within-subject variability. The larger the between-subject variability and the smaller the withinsubject variability, the greater the ICC value. Interrater measurement variability contributes to the between-subject variability when calculating the interday ICC value for arch angle. Thus, the variability of arch angle between subjects, which included large interrater variability (ICC = .605; Table 2), was much larger than the day-to-day variations (all done by the same rater, within-subject variability), resulting in the excellent interday ICC value (.895; Table 2).
The high intrarater reliability for arch angle (Table 1, ICC ≥.80) substantiates the presence of the ICC calculation bias for day-to-day reliability. In addition, low interday reliability has been demonstrated in previous research measuring arch angle, as Queen et al14 found an interday ICC value of .6377. We therefore conclude that trial-to-trial variations in the measurement of arch angle are due to the inherent difficulties of calculating the index from a footprint and not likely attributed to the reliability of the scanning methodology. To address the inherent difficulties of identifying anatomical landmarks from a footprint that are used in the calculation of arch angle, the development of a more precise measurement algorithm is necessary to overcome soft tissue variations and operator preference.
The basis for demonstrating reliability is not related to only the inherent variability of hardware that is being tested, but also to the skill and practice of the clinician, the measurements chosen, and the skill and practice of the raters chosen to make those measurements. In this study, we assessed data obtained by 3 clinicians and 3 raters. Our findings suggest that there is adequate reliability across both the clinicians and the raters used for all indices except the arch angle. If the reliability of all indices were dependent upon the skill of either the clinician or the reader, one would expect that the interclinician and interrater ICC values would be much lower than seen in the present study. As the collection protocol is not based upon rigorous posture and positioning of the subjects and was nonetheless able to attain high ICC values, the merit of grading clinicians or additional training is not prudent. The authors designed this study to replicate the most common use of the Associate Platinum digital foot scanner, which involves the collection of footprint scans by field clinicians without the need for extensive training in using the hardware.
Limitations and Future Research
One limitation of this study is the lack of comparison of these indices of footprint to the 3-D geometry of the foot architecture. However, the establishment of a true “criterion standard” in 3-D foot architecture remains controversial.7,17 Although standing planar radiographs give 2-dimensional information about the location of the bones during standing, they cannot give information about the forces acting upon those structures. Pressure mats have been used to establish loading conditions during standing but again give no information about the structural orientation of the foot. A correlation of the footprint measurements made here, the 3-D geometry of the foot, and loading conditions during static and functional tasks has yet to be demonstrated. This study did not evaluate the clinical application of this measurement. Establishing the reliability of footprint measurements is an initial step toward further developing the clinical relationship between structural and functional characteristics of the foot.
Future work in this area needs to focus on the ability of the scanner to approximate plantar pressure distribution based on skin blanching. A direct comparison with repeated measures and inter- and intraclinician reliability needs to be completed with both the scanner and a pressure mat system. Radiological measurement could also be used as the criterion standard to evaluate the discrepancy between palpation of boney landmarks and the true location of the underlying boney architecture of the foot. In this manner, measurement errors due to soft tissue displacement during stance can be effectively eliminated.
This study demonstrated that intra- and interrater reliability of scanning hardware and software (Associate Platinum digital foot scanner) to measure indices of footprints using ICCs is a reliable means by which to capture footprint data.
• Digital collection of footprints by the Associate Platinum digital foot scanner was reliable between days and clinicians in this study.
• The use of the Associate Digital Scanner resulted in a valid assessment of footprints with respect to previously reported values.
Funding Sources and Potential Conflicts of Interest
Funding for this study was provided by Foot Levelers Inc to Dr Papuga for salary support; subject compensation; wages for student research assistance; and laboratory supplies. No other conflicts of interest were reabout:newtabported for this study by the authors.
- Hogan MT, Staheli LT. Arch height and lower limb pain: an adult civilian study. Foot Ankle Int 2002;23:43-7.
- Knapik JJ, Swedler DI, Grier TL, Hauret KG, Bullock SH, Williams KW, Darakjy SS, Lester ME, Tobler SK, Jones BH. Injury reduction effectiveness of selecting running shoes based on plantar shape. J Strength Cond Res 2009;23:685-97.
- Razeghi M, Batt ME. Foot type classification: a critical review of current methods. Gait Posture 2002;15:282-91.
- Williams DS, McClay IS. Measurements used to characterize the foot and the medial longitudinal arch: reliability and validity. Phys Ther 2000;80:864-71.
- Chen CH, Huang MH, Chen TW, Weng MC, Lee CL, Wang GJ. The correlation between selected measurements from footprint and radiograph of flatfoot. Arch Phys Med Rehabil 2006;87:235-40.
- Owings TM, Woerner JL, Frampton JD, Cavanagh PR, Botek G. Custom therapeutic insoles based on both foot shape and plantar pressure measurement provide enhanced pressure relief. Diabetes Care 2008;31:839-44.
- Xiong S, Goonetilleke RS, Witana CP, Lee Au EY. Modelling foot height and foot shape-related dimensions. Ergonomics 2008;51:1272-89.
- Luximon A, Goonetilleke RS. Foot shape modeling. Hum Factors 2004;46:304-15.
- Staheli LT, Chew DE, Corbett M. The longitudinal arch. A survey of eight hundred and eighty-two feet in normal children and adults. J Bone Joint Surg Am 1987;69:426-8.
- Mickle KJ, Steele JR, Munro BJ. The feet of overweight and obese young children: are they flat or fat? Obesity (Silver Spring) 2006;14:1949-53.
- Hawes MR, Nachbauer W, Sovak D, Nigg BM. Footprint parameters as a measure of arch height. Foot Ankle 1992;13:22-6.
- Cavanagh PR, Rodgers MM. The arch index: a useful measure from footprints. J Biomech 1987;20:547-51.
- Chu WC, Lee SH, Chu W, Wang TJ, Lee MC. The use of arch index to characterize arch height: a digital image processing approach. IEEE Trans Biomed Eng 1995;42:1088-93.
- Queen RM, Mall NA, Hardaker WM, Nunley JA. Describing the medial longitudinal arch using footprint indices and a clinical grading system. Foot Ankle Int 2007;28:456-62.
- Shiang TY, Lee SH, Lee SJ, Chu WC. Evaluating different footprint parameters as a predictor of arch height. IEEE Eng Med BiolMag 1998;17:62-6.
- Rosner B. Fundamentals of biostatistics. Belmont, CA:Duxbury Press; 2005.
- Luximon A, Goonetilleke RS, Zhang M. 3D foot shape generation from 2D information. Ergonomics 2005;48:625-41.