Alterated Talar and Navicular Bone Morphology Is Associatedwith Pes Planus Deformity: A CT-Scan Study

Koen Peeters,1 Julien Schreuer,2 Fien Burg,1,3 Catherine Behets,2 Saskia Van Bouwel,4 Greta Dereymaeker,1,4

Jos Vander Sloten,1 Ilse Jonkers3

1Biomechanics Section, Department of Mechanical Engineering, KU Leuven, Belgium, 2Institut de Recherche Experimentale et CliniqueMorphology, Department, Universite Catholique de Louvain, Leuven, Belgium, 3Faculty of Kinesiology and Rehabilitation Science, KULeuven,Tervuursevest 101, 3001 Leuven, Belgium, 4Department of Orthopedics, University Hospital Antwerp (UZA), Antwerp, Belgium

Received 20 March 2012; accepted 13 August 2012

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22225

ABSTRACT: We compared bone and articular morphology of the talus and navicular in clinically diagnosed flatfeet and evaluated theirpotential contribution to talo-navicular joint instability. We used CT images to develop 3D models of talus and navicular bones of 10clinically diagnosed flatfeet and 15 non-flatfeet. We quantified their global bone dimensions, inclination and dimensions of the articularsurfaces and their curvatures. Additionally, ratios of six talar and navicular dimensions were calculated. The values for these para-meters were then compared between both groups. In flatfeet, the talar head faced more proximal and its width was larger compared tonon-flatfeet. Also the navicular cup faced more proximal and its depth was significantly increased. Furthermore, we observed a moreprotruding talar head compared to the navicular cup in the control group with the articular surface depth being relatively larger forthe navicular cups when compared to the talus in flatfeet. The ratio of the talar and navicular articular surface height was decreasedin flatfeet, suggesting increased height of navicular cups relative to the articulating talar heads. Our results show that flatfoot deformi-ty is associated with morphological changes of talar and navicular articular surfaces that can favor medial arch collapse and forefootabduction. � 2012 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res

Keywords: flatfoot; pes planus; bone morphology; navicular; talus

Currently, no clear quantitative anatomical characteri-zation of the flatfoot or pes planus deformity exists.The diagnosis of flatfoot in usually made based on clini-cal observation.1 Decrease of the medial foot arch, fore-foot abduction, and a pronounced valgus of the heel arethe most commonly observed characteristics.2 Comple-mentary investigations like the weightbearing footprintand different angles on radiographs are used to assessflatfoot deformity and foot bone alignement.3,4 Theseobservations are easy to perform but are somewhatsubjective.

Several authors tried to introduce more reproducibleexams, such as measurement of the bone-to-bone rela-tionships on MRI or weightbearing CT-scan images, todefine foot function during weightbearing conditions.5,6

CT-scan studies led to several outcomes, such as 3Dchanges in relationships among bones,7,8 and altera-tions in joint contact characteristics.9–11 These studiescontributed to the conclusion that flatfoot deformity isassociated with changes in bone-to-bone relationships.

Evidence exists that anatomical configuration andmorphology per se are associated with the developmentof flatfoot deformity. Dyal et al.12 found that manypatients with symptomatic flatfeet also presented a de-creased foot arch on the contralateral, asymptomaticside. However, few authors explored further thesemorphological changes until now. Anderson et al.13

investigated alterations of talar bone morphology inflatfeet and found significant differences with respect tocontrols. These differences concerned the ratios of bonelength/width, bone length/height, and head length/width. Flatfoot tali were narrower in width and shorterin height and had more oval shaped heads. Further-more, Grasso et al.14 observed different calcaneal mor-photypes when comparing different plantar archalterations, including pes planus deformity.

Based on these observations we investigated if, inaddition to the aberrant relationships between bonespreviously documented in flatfeet, a difference alsoexists in bone and articular surface morphology of thetalo-navicular joint. We hypothesized that a decreasedcongruence of articular surfaces in flatfeet comparedto a control group is associated with increased talo-navicular joint mobility. More specifically, we expectedan increase of the talar articular surface dimensionswith relatively smaller navicular cups, flattening ofthe articular surfaces associated with a decreased cur-vature, and alterations of the articular surface orienta-tion in the talus and navicular, all favoring a lateraland dorsal shift of the navicular bone. This might re-sult in talo-navicular joint instability, favoring forefootabduction and medial arch collapse, both characteris-tics for pes planus deformity.

MATERIALS AND METHODSSubjectsCT-scans of 25 feet from 25 patients were performed in anonweightbearing position on a Light Speed VCT 64 slicescanner (GE Medical Systems, Waukesha, WI) with a0.625 mm slice thickness and a 512/512 matrix resolution.Ten feet were classified as flatfeet (4 males, 6 females, age:12–56, with an average of 31); 15 were classified as non-

Additional supporting information may be found in the onlineversion of this article.Conflicts of interest: There is no conflict of interest for any of theauthors that could bias our results.Correspondence to: Ilse Jonkers (T: þ32-16-329-105;F: 32-16-329197, E-mail: ilse.jonkers@faber.kuleuven.be)

� 2012 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.

JOURNAL OF ORTHOPAEDIC RESEARCH 2012 1

flatfeet (control group, 6 males, 9 females, age: 22–68, withan average of 48). Classification of flatfoot was based on clini-cal evaluation by an experienced foot surgeon under weight-bearing conditions. Inclusion criteria were: a decreasedmedial foot arch; a pronounced valgus of the heel (an angleof >78 between the axis of the calcaneus and the axis ofthe calf); and abduction of the fore- and the mid-foot relativeto the heel (in posterior view of the non-flatfoot only the5th and maybe the 4th toe were visible). For the non-flatfeet,CT-scans were used from patients suffering from trauma ofthe phalanges, but who did not comply with the inclusioncriteria.

Data AnalysisThe CT-scans were analyzed using commercial softwareto reconstruct (Mimics1, Materialise, Belgium) and clean(Magics1, Materialize) 3D models of the talar and navicularbones, and custom software to extract the articular surfaces.The morphology of the bones and articular surfaces involvedin the talo-navicular joint were then quantified by 36 para-meters that were compared between flatfeet and the controlgroup through statistical analysis.

In step 1, CT data were loaded into Mimics. Individualbones were segmented from other tissues using a manualgray value thresholding procedure. The segmented data werethen used to generate a 3D model of the bone shell using anautomated reconstruction algorithm. The extracted 3D modelwas smoothed before being saved as an STL file. In step 2,automated and manual correction tools, available withinMagics were used to correct for surface irregularities such asinverted normals, remaining noise shells and bad edges. Ifdata related to a left foot, the cleaned mesh was mirrored toobtain a set of right feet for which parameter extraction wasperformed next.

In step 3, we used a custom built, Matlab based software(The MathWorks) to extract the articular surface patch usingits curvature characteristics (Fig. 1). The STL file was loadedin the software, which was used to calculate the KoenderinkShape Index (KSI) in each point of the mesh.15 The KSI cha-racterizes local shape by combining the principal curvaturesin a point of the STL mesh. The KSI has values in the range[�1, þ1]; when plotted using a color scheme (Fig. 1), thisinduces a perceptual segmentation of convex, concave, andhyperbolic areas. Next, a thresholding procedure was used toisolate a region of interest with the KSI between selectedthresholds. This region ideally coincided with the articularsurface of interest, that is the head of talus and navicularcup. If needed, superfluous parts were deleted by manualselection of mesh triangles until a clean surface contour wasobtained. Thereafter, we selected three landmarks on eachbone model (Fig. 2). For the talus: (1) lateral: tip of the

lateral malleolar facet, (2) posterior: lateral tubercle of theposterior process, and (3) medial: anterior tip of the medialmalleolar facet. For the navicular: (1) lateral: most lateralpoint of the lateral surface, (2) medial: tip of the tuberosity onthe medial facet, and (3) anterior: at mid-distance between theprevious two landmarks and central on the anterior aspect ofthe bone. Landmark selection and extraction of the surfacepatch was repeated 3� (twice by a first observer and a thirdby a second observer) to evaluate the consistency of observedparameter differences between flatfeet and the control groupwithin one observer and between multiple observers.

For both bones, an anatomical reference frame (ARF) wascalculated using the three manually selected reference points(Fig. 2). For each bone, the x-, y-, and z-axis were perpendi-cular to frontal, transverse, and sagittal planes, respectively.To exclude the effect of observer-dependent errors in selec-ting reference points and the resulting effects on the anatom-ical frame location, we also used a reference frame usingprincipal components analysis (Fig. 2). For each bone, theprincipal components were calculated, indicating its maindirections. The principal component based reference frame(PCRF) was defined with its origin coinciding with that ofthe anatomical reference frame and the axes parallel to theprincipal directions.

In step 4, we quantified bone and articular surface geome-try and articular surface orientation. A complete parameterlist is added as supplementary material (S–List 1). To evalu-ate differences in bone dimensions, we calculated:

– length, height, and width, that is the distances alongthe sagittal, longitudinal, and transverse axes, respec-tively, of both ARF and PCRF. These distances werethe differences between point coordinates of the verticeson the bone surface with minimal and maximal co-ordinate values measured along each coordinate axis(P1–P3 and P7–P9); and

– ratios of these dimensions, that is length/height, length/width and height/width (P4–P6 and P10–P12).

To quantify articular surface dimensions, we determined:

– relative depth, height, and width of the articular sur-face, calculated as the distance along the principaldirections of the surface patch, divided respectively bybone length, height or width, measured in both ARFand PCRF (P22–P24 and P25–P27). These distanceswere calculated as the difference between point coordi-nates of the vertices on the surface patch with minimaland maximal coordinate values measured along eachcoordinate axis;

– ratios of the articular surface dimensions depth/height,depth/width and height/width (P28–P30);

Figure 1. CT slices were used to generate a smoothed 3D reconstruction of the bone surface. Next, the KSI was calculated for eachpoint on the mesh, highlighting different regions.15 After thresholding, the region of highest curvature, indicated in red, appear aswhite in the next image. Finally, the articular surfaces were segmented using a region-growing algorithm.

2 PEETERS ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH 2012

– a measure of the average global articular surface curva-ture (Fig. 3), calculated as the inverse sphere radiusof the spherical approximation of the surface (P19).The approximation was calculated using a least squaresoptimization for the sphere radius and center coordi-nates to find the optimal fit of the point cloud contain-ing all vertices on the articular surface and its sphereapproximation; and

– curvatures in two perpendicular directions, quantifyingthe articular surface curvature in perpendicular planes,each calculated as the inverse radius of a circularapproximation of the intersection of a plane with thearticular surface (P20–21) as shown in Figure 3.

To quantify the articular surface orientation, the anglesbetween the normal of the planar approximation of the artic-ular surface and the three coordinate planes of both the ARFand PCRF were calculated (Fig. 4, P13–15 and P16–18). Forquantification of the articular surface congruency, ratios oftalar to navicular dimensions were compared, more specificratios of the global curvature (P31), the two perpendicularcurvatures (P32–P33), and the ratios of talar and navicularheight, depth, and width (P34–P36).

Statistical AnalysesWe analyzed the differences in the parameters between theflatfeet and control groups using the Mann–Whitney-U-test(Statistica1). Differences with p < 0.05 were assumed signifi-cant. Differences with 0.05 � p � 0.1 were considered border-line. Tables presenting average and standard deviations forall parameters in both groups are included as supplementary

material (S-Tables 1–3). In the Results, only significant andborderline differences are reported. Parameters are presentedas different if they are significantly different or have a border-line difference over all three observations. For parameters,calculated for both the ARF and PCRF, 1 of 6 comparisons(2 frames, three observations/frame) was allowed to haveinsignificant differences for the parameters to be presentedas different. For the navicular, we relaxed this criterion andstate that parameters were different if they were different forthe PCRF for all three observations, independently of what isobserved for the ARF.

RESULTS

Talus (S-Table 1)For the talus, only the angle between the articularsurface normal and the sagittal axis in the sagittalplane of the ARF (P13, Fig. 4) were significantly re-duced in flatfeet, indicating that the talar head facedmore proximally. This trend was confirmed for similarmeasures in the PCRF (P16, Fig. 4), where the diffe-rences were borderline for the three observers.

None of the global bone dimensions were signifi-cantly different between flatfeet and controls, nor werethe differences related to the articular surface curva-tures (global or perpendicular, i.e., P19, P20, and P21;Fig. 3). For the articular surface dimensions however,the articular surface width was larger in flatfeet inboth the ARF and PCRF (P24 and P27). Borderlinedifferences were confirmed for all three observations.

Figure 2. Contact surfaces, landmarks for anatomical reference frame definition and different reference frames for talus (top row)and navicular (bottom row). Note that talus and navicular are presented at different scales.

BONE MORPHOLOGY ALTERATIONS IN FLATFEET 3

JOURNAL OF ORTHOPAEDIC RESEARCH 2012

Navicular (S-Table 2)For the navicular, the angle in the sagittal plane ofthe PCRF (P16, Fig. 4) was significantly reduced inflatfeet for all observations. This finding is indicativefor a more proximal facing articular surface.

None of the global bone dimensions were significantlydifferent between flatfeet and controls. Articular surfacecurvatures of the navicular cup showed no significantdifferences (P19, P20, P21) between the two groups.Conversely, the navicular cup depth was significantlyincreased in flatfeet, both relative to navicular thicknessmeasured in the ARF and PCRF (P22 and P25) and re-lative to navicular cup height and width (P28 and P29).

Talus Relative to Navicular (S-Table 3)Among the three parameters reflecting articular sur-face dimensions (P34, P35, P36), the increased ratioof talar articular surface depth to navicular articularsurface depth (P34) was borderline, but suggests amore protruding talar head compared to the navicularcup in the controls. The ratio of talar articular surfaceheight to navicular articular surface height (P35) wassignificantly decreased in flatfeet for the three obser-vations; thus, the talar patch height was increased re-lative to the navicular patch height in the controls.The ratios of articular surface curvatures showed nodifferences.

Figure 3. The top row shows the approxima-tion of the talar head and navicular cup by asphere for calculating the global curvature (P19in S-List 1). The bottom row illustrates the cal-culation of the curvatures in perpendiculardirections (P20 and P21 in S-List 1). After seg-mentation of the articular surface, a coordinateframe was defined, with the axes parallel to theprincipal directions of the segmented pointcloud. These principal directions were calculatedusing principal component analysis on the pointcoordinates of vertices on the surface patch. Inthis coordinate frame, with the z-axis the axiswith minimal inclination with respect to thesagittal axis of the foot, the articular surfacewas represented by a thin-plate spline.18 Next,the intersection of the thin plate spline with thetwo coordinate planes, most inclined with re-spect to the articular surface, was calculated.Therefore, the x and y coordinates in the thin-plate spline expression were put to zero. Theintersections were sampled, and these sampledcurves were approximated by circles using aleast squares optimization for the circle radiusand center coordinates to fit the sampled curveto its circle approximation. The inverse radii ofthese circles were then used to quantify curva-tures in two perpendicular directions.

Figure 4. Articular surface orientation wascalculated by approximating the surface with aplane, then projecting the plane normal in oneof the reference frame coordinate planes, afterwhich the angle between the projected normaland a reference axis was calculated. Negativeangles were indicated in the sagittal, frontal,and transverse plane.

4 PEETERS ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH 2012

DISCUSSIONDespite existing evidence for the aberrant anatomicalconfiguration of bones to be associated with the deve-lopment of flatfoot deformity,12 few authors exploredthe morphological changes in flatfeet. We evaluateddifferences in bone and articular surface morphology,focusing on the talo-navicular joint in flatfeet. Wehypothesized that prominent features in flatfeet, fore-foot abduction and medial arch collapse, are associatedwith talo-navicular joint instability. This instability isexpected to be associated with an increase of the talararticular surface dimensions with relatively smaller na-vicular cups reducing coverage of the talar head by thenavicular cup, flattening of the articular surfaces asso-ciated with a decreased curvature, and alterations ofthe articular surface orientation in the talus and navi-cular. In flatfeet, our results confirm that the articularsurface orientation at the talo-navicular joint is alteredso as to encourage a plantar shift of both talus andnavicular. The observed reduction of the angle betweenthe articular surface normal and the sagittal axis inflatfeet can be assumed to be associated with the talarneck axis being more aligned with the sagittal footaxis. This indirectly suggests that the forefoot will bepositioned more in line with the talar sagittal axis ofthe ARF and thus suggests a medial arch collapse(Fig. 5). Likewise, our data show that the sagittal planeangle between the navicular cup normal and the navi-cular sagittal axis are decreased in flatfeet. Thesechanges reflect a more horizontal inclination of themedial bone column and flattening of the longitudinalarch.

The suggested relation between bone geometryand the relation to medial arch collapse complementfindings of Kitaoka et al.7 They reported an averagenavicular-to-talar position difference of 13.68 in abduc-tion, 10.58 in dorsiflexion, and 8.18 in eversion.The average metatarsal-to-talar position difference is11.78 in abduction, 10.48 in dorsiflexion, and 10.98 ineversion. We confirm their conclusions by stating thatour reported changes in bone geometry clearly asso-ciate with altered bone-to-bone relationships that willalter the foot arch loading, making it more susceptibleto a medial collapse of the longitudinal arch.

Apart from the altered orientation of the articularsurfaces of the navicular and/or talar bone, changes inthe articular surface dimensions could contribute to adecreased congruence of talar and navicular articularsurfaces that evokes joint instability. However,changes in our parameter values are suggestive of bet-ter coverage of the talar head by the navicular cup;the articular surface of the talar head is wider andnavicular cups are deeper in flatfeet, even whenaccounting for the increased bone length. Also relativeto the articulating talar heads, navicular cups aredeeper and higher in flatfeet. Also, articular surfacecurvatures were not significantly different. Based onthese findings, we conclude that changes in articularsurface orientation, rather than decreased congruency

of the articular surfaces at the talo-navicular joint,contribute to joint instability and thus associate withmedial arch collapse.

The altered bone morphology, inducing medial archcollapse, completes the physiopathology described byLeemrijse et al., who state that, by an anterior drift ofthe talus relative to the calcaneus, the medial bonecolumn will be relatively increased in length, thusprovokings forefoot abduction. We found, however,no significant parameter differences that relate tochanges in morphology that may induce the hypo-thesized forefoot abduction.

In addition to the observed differences, some previ-ously reported differences in flatfeet morphology didnot reach significance in our population. In agreementwith Anderson et al.13 the talus was longer relativeto its height and width in flatfeet when compared tocontrols in both reference frames. Furthermore, theincreased articular surface dimensions of the talarhead surface patch (particularly with the surfacewidth for which we found a borderline difference) cor-responds to the more oval-shaped form of the headpreviously described by Anderson et al.13

Bones are dynamic structures that adapt their ge-ometry to mechanical loading.16 We are therefore leftwith a ‘‘chicken or egg’’ dilemma: It is unclear if theobserved adaptations in bone geometry are induced bythe medial arch collapse that affects bone loading andconsequently induces geometrical bone remodeling.Alternatively, changes in morphology could be presentinitially and predispose to medial arch collapse.In support of the first hypothesis, Anderson et al.13

advanced the hypothesis that a lateral drift of the cal-caneus decreases the support of the talar head. Thus,the forefoot abducts, and the load on the dorsolateralaspect of the talo-navicular articulation increases.This could then provoke the formation of new bone onthe dorsal and lateral part of the head, inducing amore oval shape. A similar process can be responsiblefor inducing the wider navicular cups in flatfeet.

Some shortcomings of our study must be mentioned.First, we used small-sized samples that did not allowconcluding on a general population of flatfeet patients.Second, we used nonweightbearing CT-scans to alloweasier identification of the border between individualbones and facilitate the segmentation and consequentparametrization of individual bone geometries andarticular surfaces. Although, this approach favorscharacterization of bone and articular geometries, ourfindings cannot be extrapolated to bone-to-bone rela-tionships during loading. Third, our data processingrelied on operator-dependent bone segmentation and amanual selection procedure of the landmarks requiredfor definition of the ARF that might introduce variabi-lity in the data. To account for this possibility, theanalysis was performed three times, twice by a firstobserver and once by a second observer. Furthermore,to exclude the effect of variability in landmark selec-tion between observers, the PCRF was also used. The

BONE MORPHOLOGY ALTERATIONS IN FLATFEET 5

JOURNAL OF ORTHOPAEDIC RESEARCH 2012

PCRF does not rely on subjective anatomical point se-lection, but altered bone geometry will primarily affectaxis definition and may be less prominent in the calcu-lated parameters, therefore concealing differences inbone morphology between flatfeet and controls. In con-trast, the latter accounts for the subject’s anatomy andrelates the parameters to specific anatomical referencepoints on the bone, but this approach is more sensitiveto variability introduced by the observer, in particularfor the navicular as there are no pronounced featuresthat allow easy and repeatable landmark detection. Bycomparing the results from the two frames, we canconfirm that the observed results were consistentfor both approaches for the talus. For the navicular,significant differences for the analysis results in thePCRF are considered decisive, independent from theobservations in the ARF, because of the lack of pro-nounced features for landmark detection and thusconsistent ARF definition over different observers.Fourth, 4 subjects between 12 and 17 years-old wereincluded in the flatfoot group. Although, we did notobserve substantial differences with other flatfeet, thegrowth plates may not have been completely fused inthese feet, which might allow additional adaptation inbone growth and therefore articular inclination. Thismay introduce additional variability in morphologycompared. However, we expect this effect to be mini-mal, as growth plates close at age 14 for females and16 for males, with markedly decreased growth afterage 12.17

To conclude, this study highlights morphologicalchanges of talar and navicular bones in flatfeet. Themain changes with respect to the control group includemore proximal facing articular surfaces of the talusand navicular, a deeper navicular cup, an increasedarticular surface width for the talus, a less protrudingtalar head compared to the navicular cup, and ahigher navicular articular surface relative to the arti-culating talar head height. All these changes can berelated to medial arch collapse and add to previouslyobserved changes inducing forefoot abduction, twomajor characteristics of flatfoot deformity. As this isa cross-sectional study, it fails to elucidate if thesechanges are present from the onset of flatfoot diseaseor if they develop over the course of time due to thealtered loading of the relevant bone structures.

ACKNOWLEDGMENTSThis work was funded by the Chair Bergman–Dereymaekerand the Research Foundation Flanders.

REFERENCES1. Resnick D, Niwayama G. 1981. Diagnosis of bone and joint

disorders. Philadelphia: W.B. Saunders Company; 4944 p.2. Leemrijse T, Valtin B. 2009. Pathologie du pied et de la

cheville. Paris: Elsevier; 825 p.3. Tareco JM, Miller NH, MacWilliams BA, et al. 1999. Defin-

ing flatfoot. Foot Ankle Int 20:456–460.4. Younger AS, Sawatzky B, Dryden PJ. 2005. Radiographic

assessment of adult flatfoot. Foot Ankle Int 26:820–825.5. Ferri M, Scharfenberger AV, Goplen G, et al. 2008. Weight-

bearing CT scan of severe flexible pes planus deformities.Foot Ankle Int 29:199–204.

6. Stindel E, Udupa JK, Hirsch BE, et al. 1999. A characteriza-tion of the geometric architecture of the peritalar joint com-plex via MRI: an aid to the classification of foot type. IEEETrans Med Imag 18:753–763.

7. Kitaoka HB, Luo ZP, An KN. 1998. Three-dimensional anal-ysis of flatfoot deformity: cadaver study. Foot Ankle Int19:447–451.

8. Ledoux WR, Roehr ES, Ching RP, et al. 2006. Effect offoot shape on the three-dimensional position of foot bones.J Orthop Res 24:2176–2186.

9. Ananthakrisnan D, Ching R, Tencer A, et al. 1999. Subluxa-tion of the talocalcaneal joint in adults who have symptom-atic flatfoot. J Bone Joint Surg Am 81:1147–1154.

10. Friedman MA, Draganich LF, Toolan B, et al. 2001. Theeffects of adult acquired flatfoot deformity on the tibiotalarjoint contact characteristics. Foot Ankle Int 22:241–246.

11. Greisberg J, Hansen ST, Sangeorzan BJ. 2003. Deformityand degeneration in the hindfoot and midfoot joints of theadult acquired flatfoot. Foot Ankle Int 24:530–534.

12. Dyal CM, Feder J, Deland JT, et al. 1997. Pes planus inpatients with posterior tibial tendon insufficiency: asymp-tomatic versus symptomatic foot. Foot Ankle Int 18:85–88.

13. Anderson JG, Harrington R, Ching RP, et al. 1997. Altera-tions in talar morphology associated with adult flatfoot. FootAnkle Int 18:705–709.

14. Grasso A, Scarfi G. 1993. Computed tomography demonstra-tion of calcaneal morphotypes in plantar arch alterations.La Radiologia Medica 85:384–388.

15. Koenderink J, van Doorn A. 1992. Surface shapes and curva-ture scales. Image Vision Comput 10:557–565.

16. Wolff J. 2010. Das Gesetz der Transformation der Knochen,1892. Reprint. Berlin: Pro Bus; 300 p.

17. Sarrafian SK. 1983. Anatomy of the foot and ankle. Philadel-phia: J.B. Lippincott Company; p 30–32.

18. Boyd SK, Ronsky JL, Lichti DD, et al. 1999. Joint surfacemodeling with thin plate splines. J Biomech Eng 121:525–532.

Figure 5. The red lines in the left mostimages represent the surface inclination in flat-feet relative to the surface inclination in non-flatfeet, represented by the black lines. The redarrows indicate the assumed position to whichthe navicular will move with respect to the talusas a consequence of the observed changes inmorphology, relative to the position in non-flat-feet represented by the black arrows. Arrows onthe right most images present the assumed di-rection bones will move to as a result of the ob-served changes in morphology.

6 PEETERS ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH 2012