New Concept of 3D Biometrics for Hindfoot Alignment Using Weight Bearing CT

, Francois Lintz2, Cesar de Cesar Netto3, Alexej Barg4, Arne Burssens5 and Scott Ellis6

Department for Foot and Ankle Surgery, Hospital Rummelsberg, Schwarzenbruck, Germany

Foot and Ankle Surgery Centre, Clinique de l’Union, Toulouse, France

Department of Orthopedics and Rehab, University of Iowa, Iowa City, IA, USA

University Orthopedic Center, University of Utah, Salt Lake City, UT, USA

Department of Orthopedics and Trauma, University Hospital of Ghent, Ghent, OVL, Belgium

Department of Orthopedic Surgery, Hospital for Special Surgery, New York, NY, USA



Weight bearing CTPedCATTALASFoot ankle offsetHindfoot alignment


The importance of hindfoot alignment (HFA) in the setting of hindfoot surgery has been described and demonstrated in the literature due to the relationship between alignment and surgical outcome [13], as in ankle arthrodesis and arthroplasty [411] or in knee arthroplasty [1216]. HFA measures based on 2D radiographs are flawed by many anatomical and operator-related bias which have been extensively investigated in the literature [1729]. These include projection and rotation issues [30, 31] mostly with regard to the use of the tibia as a reference axis [32]. Previous studies on the HFA have focused on which technique to use and measurement methods [3245].

Furthermore, 2D measures traditionally do not account for the contribution of the forefoot in ankle biomechanics. Recently, two publications from different authors [17, 31] have used HFA measures using the forefoot rather than the tibia as a reference, which seems to be one step forward in reducing the bias related to projection and rotation. However, these techniques were still based on 2D radiographs and were complicated to measure and thus were not usable in daily practice.

Today, 3D weight bearing computed tomography (3D WBCT) provides an opportunity to solve the problems associated with 2D biometrics [46, 47]. This could be done by creating a new generation of HFA measures or 3D biometrics based on 3D coordinates rather than trying to force apply older HFA measurements based on 2D angles to this new 3D environment. The ideal criteria for this new kind of tool would be physically meaningful, designed for 3D WBCT, computerized, and verifiable by data.

In this study, we investigated the efficacy of a new system: Torque Ankle Lever Arm System (TALAS™, CurveBeam LLC, 175 Titus Ave, Suite 300, Warrington, PA 18976, USA) designed to satisfy those requirements and used to calculate a computerized, semiautomatic 3D biometric—the foot ankle offset (FAO). It represents the lever arm or the torque generated in the ankle from the combined actions of body weight and ground reaction force. This technique had been previously validated in a simulated 3D weight bearing (WB) environment on 2D radiographs, at a time when 3D WBCT was not readily available [31]. The objective of this study was to describe the distribution of FAO in a series of anonymized datasets from clinically normal, varus, and valgus cases. We hypothesized that normal, varus, and valgus cases should be significantly different and that the distribution should be Gaussian in the normal population. We also assessed the interobserver and intraobserver reliability of the FAO measurement.


This is a comparative study looking retrospectively at existing data captured as part of routine clinical care. The study was conducted with institutional review board approval.

We analyzed 135 anonymized consecutive datasets, of which 57 were from patients with normal hindfoot alignment (42%), 38 from patients with varus hindfoot alignment (28%), and 40 from patients with valgus hindfoot alignment (30%). All scans were obtained from WBCT scans using the PedCAT™ unit (CurveBeam LLC, 175 Titus Ave, Suite 300, Warrington, PA 18976, USA) installed in the outpatients sector of an orthopedic Foot and Ankle Surgery referral center. The datasets were obtained using the following cone beam scanner settings: voxel size, 0.37 mm; field of view diameter, 350 mm; field of view height, 200 mm; exposure time, 9 seconds; total scan time, 54 seconds. The datasets were extracted from the existing database, containing the 3D image data (Fig. 6.1) and patient demographics limited to side and clinical morphotype.


Fig. 6.1

3D views obtained by the WBCT built in software

Datasets were screened by two independent observers using the built-in software, CubeView™ (CurveBeam LLC, 175 Titus Ave, Suite 300, Warrington, PA 18976, USA) and collected the 3D coordinates of specific anatomical landmarks required for the software to process and calculate FAO. Two independent observers collected the 3D coordinates (x, y, z), for the following landmarks: First metatarsal head WB point (A), Fifth metatarsal head WB point (B), Calcaneus WB point (C), Talus centermost and highest point respectively in the coronal and sagittal planes (D1). The WB point was defined in the ground plane by its x and y coordinates as the lowest point (z coordinate nearest to 0) on the surface of the calcaneus and metatarsal WB surfaces. The center of the talus was defined as the highest point (z coordinate furthest to 0) on the talus in the centermost sagittal plane.

All these values were collected and stored anonymously in a spreadsheet, which a third investigator then ran the through a beta version of the Torque Ankle Lever Arm System, which calculated FAO values using an algorithm based on the inverted 3D pyramid model [31] (Fig. 6.2).


Fig. 6.2

Schematic representation of FAO in the transversal plane (see the text for abbreviations)

“A,” “B,” “C,” and “D1” are defined as, respectively, the WB points of the first metatarsal head, the fifth metatarsal head, the calcaneus, and the talus; the midpoint “E” between “A” and “B” is found; the “CE” line is found, and the “CE” distance is called “foot length”; “D” is the orthogonal projection of “D1” on the ABC plane (i.e., the position of the center of the talus in the ground plane); the perpendicular line to “CE,” which includes “D,’ is found (i.e., the line from the center of the talus to the midline of the foot on the ground plane); “F” is the intersection of these two lines, and thus “DF” is found, being positive when “D” is medial to “F” and negative when lateral; FAO is DF/CE × 100 and is given as a percentage.

FAO therefore corresponds to the offset between the hindfoot-to-forefoot midline and the talus. It is therefore given as a percentage of foot length and is theoretically representative of the torque, which is produced in the ankle by this offset. This conversion to percentage of foot length is done to normalize the FAO value to foot size so that it is comparable between feet of different sizes, since the FAO is a metric, not an angle.

Statistical Analysis

FAO values in three groups were compared using a one-way ANOVA test. Normality test was performed using the Kolmogorov-Smirnov test method on the normal foot group. An intraclass correlation coefficient (ICC) two-way mixed model was used to assess the intraobserver and interobserver reliability of FAO values. ICCs were calculated with the measurements performed twice by two independent observers on datasets in the three different groups: normal, varus, and valgus. ICC values range from 0 to 1, with a higher value indicating a better reliability. ICCs below 0.40 were considered poor, 0.40–0.59 fair, 0.60–0.74 good, and 0.75–1.00 excellent.

Data are presented as mean, standard deviation, range, and 95% confidence intervals (CI 95%). All analyses were carried out using STATA statistical software package (Version 12.0, StataCorp, 2011). A P value <0.05 was considered to indicate statistical significance.


In normal cases, the mean value for FAO was 2.3% ± 2.9 (CI 95%, 1.5–3.1; range, −3.2 to 9.0), whereas in varus and valgus cases, the mean was −11.6% ± 6.9 (CI 95%, −13.9 to −9.4; range, −26.4 to −1.0) and 11.4% ± 5.7 (CI 95%, 9.6–13.3; range, 2.0–24.5), respectively. The difference was statistically significant among the three groups (P < 0.001). FAO values distribution in the normal population was normal (P > 0.81) (Fig. 6.3).


Fig. 6.3

Gaussian distribution of FAO values in the normal population

The interobserver and intraobserver reliability of FAO measure were excellent, with small standard deviations; the global mean ICC was 0.99 ± 0.002 (CI 95%, 0.99–1.00) and 0.97 ± 0.02 (CI 95%, 0.92–1.02) for the inter- and intraobserver assessment, respectively. ICC values for the normal, varus, and valgus groups are showed in Table 6.1.

Table 6.1

Inter- and intraobserver reliability of FAO measurement in different groups




95% Cl





































Abbreviations: FAO foot and ankle offset, ICC intraclass correlation coefficient, SD standard deviation

Only gold members can continue reading. Log In or Register to continue

Apr 25, 2020 | Posted by in MUSCULOSKELETAL MEDICINE | Comments Off on New Concept of 3D Biometrics for Hindfoot Alignment Using Weight Bearing CT
Premium Wordpress Themes by UFO Themes