Computers in Biology and Medicine 39 (2009) 280 -- 285

Contents lists available at ScienceDirect

Computers in Biology andMedicine

journal homepage: www.e lsev ier .com/ locate /cbm

Reliability of a navigation system for intra-operative evaluation of antero-posteriorknee joint laxity

Nicola Lopomoa,b,∗, Simone Bignozzia, Sandra Martellia, Stefano Zaffagninia, Francesco Iaconoa,Andrea Visania, Maurilio MarcacciaaIstituto Ortopedico Rizzoli, Laboratorio di Biomeccanica, via di Barbiano, 1/10, Bologna BO 40136, ItalybDipartimento di Bioingegneria, Politecnico di Milano, Milano, Italy

A R T I C L E I N F O A B S T R A C T

Article history:Received 18 June 2007Accepted 5 January 2009

Keywords:Knee laxityACL lesionNavigation systemsArthrometerRolimeterKIN-Nav

Background: The purpose of this study was to investigate about the reliability of measuring antero-posterior laxity within-subjects for in-vivo studies using a navigation system.Methods: The analysis was performed by enroling 60 patients undergoing anterior cruciate ligament ACLreconstruction, and assessing AP laxity during the Lachman and drawer tests.Results: For the navigation system standard deviation for intra-trial measures was 0.7mm, thus the intra-trial repeatability coefficient was 2.2mm; standard deviation for intra-trial measure was 1.2mm, whilethe reference inter-trial repeatability coefficient between expert surgeons was 3.4mm.Conclusions: In conclusion, this study suggests that KIN-Nav may represent a new method to measure anddocument AP laxity intra-operatively with improved accuracy and test the effect of surgical treatmentin-vivo with higher sensitivity than in the past and this study quantify its reliability for within-subjectsstudies performed by a single expert surgeon.

© 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The importance of the evaluation of passive antero-posterior (AP)knee laxity in order to assess both knee anterior cruciate ligament(ACL) injuries and the efficacy of relative surgical reconstruction hasbeen widely discussed and validated [1–3]. The necessity to have aquantification of this laxity led to the development of several in-struments and methodologies used for objective measurements; inparticular arthrometers [4], bioimages (radiological methodologies[5,6], RSA [7], dynamic MRI [8,9], and fluoroscopy [10]), electrogo-niometers [11], electromechanic devices [12] and 3D trackers [13]were widely used.

Nevertheless, only a few techniques have proven to be suitablefor intra-operative use, basically due to the strict restraints for theequipment that can be used in the operating room (OR), including theneed to be sterilisable and the requirement to minimise surgeons'fatigue, additional surgical time and patientmorbidity. At present themost widely spread tools used to quantify manually AP displacement

∗ Corresponding author at: Istituto Ortopedico Rizzoli, Laboratorio di Biomecca-nica, via di Barbiano, 1/10, Bologna, BO 40136, Italy. Tel.: +390516366520;fax: +39051583789.

E-mail address: n.lopomo@biomec.ior.it (N. Lopomo).

0010-4825/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.compbiomed.2009.01.001

of the tibia with respect to the femur are basically arthrometers thatcan be used either at maximum stress [14–16] or with a known force[17–19]. These devices are very simple and are able to provide anobjective quantitative evaluation of AP laxity, both before and aftersurgical treatment.

However the measurements performed using these devices areaffected by several sources of errors, such as the impossibility to fixthe tool directly on bones during acquisition, the uncertainty aboutthe initial knee position (degree of flexion/extension), test condi-tions during the evaluation, and variability due to tester's experi-ence [4,12,16,18,20,21]. Therefore, further studies on these devicesor on new methodologies that quantify AP laxity would be useful toimprove reliability of knee kinematics estimation. Recently, the usein the OR of 3D tracking devices, and in particular their applicationin navigation systems, seems useful to overcome some of the prob-lems reported with arthrometers. Since their intrinsic accuracy ishigh (0.350mm is the highest threshold in the identification of op-tical marker position) and the trackers are directly fixed on bones,they can be used to assess the relative displacement of bones with-out concern for soft tissue artefacts; in addition navigation systemsare suitable to improve the test reproducibility, allowing the surgeonto check knee placement and test execution [22–24].

Although this methodology is expensive and slightly invasive(due to the need to use sensors that are fixed to the bones),

N. Lopomo et al. / Computers in Biology and Medicine 39 (2009) 280–285 281

navigation systems-having the robust capability to evaluate jointkinematics-are very promising in this field of application and there-fore deserve further investigation. Previous studies have assessedtheir reliability in the acquisitions of anatomical landmarks or in theevaluation of specific surgical techniques [25–29], but the analysisof their performance for measuring kinematic values is still lacking.

The objectives of this study were thus to statistically determinethe reliability of a navigation system in the measurement of APlaxity clinical tests (Lachman and drawer tests) performed by thesame surgeon (intra-tester reliability) and by different operators(inter-tester reliability), and also to compare the computer-assistedprocedure with a known widely used method, such as the clin-ical evaluation performed with a commercial knee arthrometer(Rolimeter�, Aircast Europe, Germany) in the same test conditions.A preliminary validation of the navigation system for translationaland rotational laxity measurements was also performed by the au-thors in [30], however the statistical analysis and the experimentaldesign was improved in this study for AP laxity in order to un-derstand how to interpret clinical results obtained with navigationsystems and compare them with those obtained in literature withother methodologies. Therefore the major novelty of this work is toprovide a complete analytical method, able to assess the clinical co-herence of the data coming from different measuring system and toevaluate the degree of consistency between the measurements per-formed both by the same examiner and by different examiners too.

2. Material and methods

2.1. Subjects and testers

This study involved a cohort of 60 consecutive patients that un-derwent arthroscopic ACL reconstruction performed in our institutebetween January and September, 2005. All patients were recruitedif they complained of anterior knee instability, with a diagnosis ofisolated ACL ligament injury and no previous surgery on the affectedknee. Patients were included also if a torn meniscus was associatedwith the ACL injury (12 cases); however, they were excluded if theyhad any co-existing disease in the knee. Among this group of patients,all those who voluntarily agreed to take part in the research proto-col were enroled. The sample set included 53 men and 7 women;the mean age of study patients was 31.0 ± 10.8 years (range: 16–59years); 32 knees were right and 28 left; all lesions were consideredas sub-acute or chronic lesions of traumatic origin with a mean timefrom injury to surgery of 12 ± 10.8 months (range: 3–48 months).

This study design was preferred to a randomised trial with amuch longer completion time to minimise biases introduced by thelearning curve, differences in surgical equipment, and the operatingenvironment.

All patients underwent an arthroscopic double-bundle or single-bundle ACL reconstruction with hamstring tendon graft [31,32]. AllACL lesions were arthroscopically confirmed to match the inclusioncriteria of this study.

Three investigators from the surgical team of the institute wereselected for the necessary examinations used for this study. The firstexaminer (surgeon A) was an experienced surgeon, who has beenin charge of regular clinical routine and research services for morethan 10 years and 200 navigated interventions; he was consideredto be an expert user of both the navigation system and other clin-ical evaluation methods. The second examiner (surgeon B) was anexperienced surgeon with more than 5 years' experience in clinicalpractise and 50 navigated interventions; he was considered to bean intermediate user of the navigation system and an expert user ofthe clinical evaluation methods. The third investigator (surgeon C)was considered to be a novice user of all evaluation methods, as hewas medically qualified but had little experience with diagnostic and

Fig. 1. KIN-Nav set-up.

surgical examinations (less than 2 years) and no previous experiencewith the navigation system.

The study designwas approved by the Institutional Ethical ReviewBoard of the Institute prior to data collection, and all patients gavetheir informed consent.

2.2. Experimental procedure

After spinal anaesthesia and with tourniquet inflated, a prelim-inary clinical pre-operative knee evaluation was performed using aRolimeter knee arthrometer before entering the OR. The evaluationwas performed by a single examiner, the most experienced investi-gator.

ACL reconstruction was performed arthroscopically by the same,most experienced surgeon, using or a single-bundle or a double-bundle technique with hamstring tendons.

The intra-operative assessment of the knee AP laxity wasperformed using the KIN-Nav navigation system, developed at ourinstitute, consisting of an optical localizer (Polaris, Northern Dig-ital, Waterloo, ON, Canada) and dedicated software running on acommercial laptop (Fig. 1). This system was tested and optimisedprior to this study and was deemed suitable for surgical routine[23,24,26,33].

In this study we recorded Lachman and drawer tests in ACL-deficient and reconstructed knees, thus estimating AP laxity, respec-tively, at 30◦ and 90◦. The surgeon performed the evaluation of kneestability as in clinical practise, by applying maximum manual loadat the level of the tibial tuberosity perpendicular to the tibial axis,similarly to tests with the Rolimeter [14–16]. The leg was positionedat the corresponding degree of flexion supported by a sterile drape,placed under the thigh, and the corresponding flexion angle waschecked in real-time on the KIN-Nav interface (Fig. 2).

AP laxity was measured also after the reconstruction, as in ACL-deficient state, both with KIN-Nav and the Rolimeter. The mean in-terval time between the pre-operative Rolimeter evaluation and thefirst KIN-Nav evaluation of the knee was 30 minutes, while, due toOR necessity, post-operative evaluations were consecutive. All sur-geons witnessed the measures and were blinded to the laxity valuesobtained with KIN-Nav and Rolimeter.

In order to estimate KIN-Nav's ability to assess AP laxity wetested inter-trial and intra-trial repeatability. In an effort to limit theadditional surgical time required for this study the intra-trial and

282 N. Lopomo et al. / Computers in Biology and Medicine 39 (2009) 280–285

Fig. 2. The user interface of KIN-Nav system. The figure shows in details the interfacewith each single parameters available in the system (pre- and post-operative out-comes of general laxity tests—including varus/valgus and internal/external rotationtest here not discussed—and limb position). For this study the results of the testswere covered for a blinded evaluation: only the flexion/extension angle was visible.

inter-trial repeatability measures were performed in two separategroups of patients. On the first 30 out of 60 cases the senior operat-ing surgeon performed AP displacement tests at manual maximumforce three times at 30◦ of knee flexion and three times at 90◦ of kneeflexion. After each trial the surgeon stopped and the initial knee flex-ion angle was adjusted, according to KIN-Nav's measure of the kneeposition. On the next 30 cases, two of the three surgeons performedanterior displacement tests with a manual maximum force. Each ofthem performed one evaluation at 30◦ knee flexion and one evalu-ation at 90◦ knee flexion checking knee position with the KIN-Navsystem. Once the knee was tested by one examiner, another surgeonperformed the same evaluation. The selection of two surgeons outof three and the order of the surgeons for consecutive tests wererandomized for each subject.

2.3. Data analysis

Descriptive statistical analysis was performed on pre-operativeevaluations on ACL-deficient knees and post-operative evaluationson reconstructed knees obtained with bothmeasurement techniques(KIN-Nav and Rolimeter) at 30◦ and 90◦ of flexion.

Analysis of variance was performed on KIN-Nav data in orderto verify the intra-trial and inter-trial variability and clinical reli-ability of the navigation system. Specifically, inter-trial and intra-trial repeatability was estimated by evaluating within-subject SD,item inter-class correlation coefficient (ICC), Cronbach � test values(alpha), Bland & Altman repeatability coefficients (RC) and Pearsoncorrelation coefficient (r) [34–39]. In particular, in the intra-trial re-peatability the three repeated test performed by the same surgeonin different knee conditions were compared. For inter-surgeon re-peatability we correlated the two measures, taken on each patientby two of three surgeons, dividing the data according to the cou-ple of surgeon that performed the measurement. In this manner wewere able to correlate the results of expert, intermediate and noviceusers. The data sample was not divided according to the knee con-ditions in order to have an adequate sample size for the analysis.

Analysis of variance was also performed for the comparison be-tween KIN-Nav and Rolimeter measurements. In particular, limits ofagreement [37] and linear regression were evaluated for all the casesand knee conditions.

2019181716151413121110

987654321

AP

(mm

)

PRE POST POSTPREAP 90AP 30

KIN-nav

Rolimeter

Fig. 3. Boxplot of the pre-operative evaluations on ACL-deficient knees (PRE) andpost-operative evaluations (POST) on reconstructed knees obtained with both mea-surement techniques (Kin-Nav and Rolimeter) at 30◦ (AP 30) and 90◦ (AP 90) offlexion.

For all tests the level of significance was set at p = 0.05 (both forcomparison of techniques and KIN-Nav reproducibility estimation).Statistical analysis was performed using Analyse-it software 1.73(Analyse-It Software, Ltd., Leeds, UK) and SPSS 12 (SPSS Inc, Chicago,Illinois).

3. Results

3.1. Descriptive statistics

Themean and standard deviation of the AP laxity values producedby the KIN-Nav and Rolimeter measurement techniques across allsubjects and in the four knee conditions are shown in Fig. 3. Thesevalues showed that laxity values obtained with both methods forthe examined sample are consistent with previous studies in theliterature [4,5,12,14–18].

3.2. Repeatability of KIN-Nav

The data obtained with the KIN-Nav were in a range from 1.1to 19.2mm. Moreover the SD of individual repeated measures wasindependent from the average (Kendall Tau = 0.154 with p = 0.014and Spearman's Rho = 0.22, p = 0.011), so that also Bland & Altmananalysis could be applied.

The intra-trial variability of navigated AP laxity evaluation withKIN-Nav by the most experienced surgeon during AP test (includingLachman and drawer test in ACL deficient and reconstructed knees)was low: the within-subject SD was 0.7mm. The calculation of ICCfor intra-trial data showed that there was an excellent correlationamong the three repeated measures (ICC = 0.95). Also the reportedPearson correlation coefficient and Cronbach � test value were high(respectively, r = 0.96 and alpha = 0.98). We obtained also a globalrepeatability coefficient (RC) of 2.2mm during all the tests, accordingto Bland & Altman analysis (Table 1).

The inter-trial variability of navigated AP laxity evaluations bydifferent examiners was greater than that obtained by the most ex-perienced surgeon; in particular the within-subjects SD was 1.2mmbetween surgeon A and surgeon B, 1.1mm between surgeon A andsurgeon C, and 2.1mm between surgeon B and surgeon C. Somedifferences among the testers were present, in particular between

N. Lopomo et al. / Computers in Biology and Medicine 39 (2009) 280–285 283

intermediate and novice surgeons (the worst Pearson correlationcoefficient r = 0.69 and Cronbach � test value alpha = 0.82). Thereference RC (surgeon A vs. Surgeon B) we obtained was 3.4mm(Table 1). A summary with all the values of ICC and correlation coef-ficients for intra-trial and inter-trial analysis is reported in Table 1.

3.3. Comparison between KIN-Nav and Rolimeter

According to the Bland & Altman agreement analysis the KIN-Navevaluations appeared to be slightly higher than the Rolimeter ones(especially for very lax knees and at 30◦ of flexion in ACL-deficientknees). It was also noticed that data obtained with Rolimeter rangedfrom 2 to 15mm in most cases (only one outlier = 20mm), while theKIN-Nav evaluation provided laxity values between 16 and 20mmmore often (8 cases), although these conditions were exceptionalcases. The mean difference between the KIN-Nav laxity (as the meanvalue of repeated measures of the expert surgeon) and Rolimeterlaxity for measurements was 1.1mm, with the computed limits ofagreements [7]. The difference between the two methods appeared

Table 1Standard deviations (SD), interclass correlation coefficient (ICC), Bland & Altmanrepeatability coefficient (RC), Pearson correlation coefficients (r) and Cronbach �test value (alpha) for intra-trial (Surgeon A vs. Surgeon A) and inter-trial (SurgeonA vs. Surgeon B, etc.) analysis.

Surgeon A Surgeon B Surgeon C

SD = 0.7mm SD = 1.2mm SD = 1.1mmICC = 0.95 ICC = 0.93 ICC = 0.83

Surgeon A RC = 2.2mm RC = 3.4mm RC = 3.1mmr = 0.96 r = 0.94 r = 0.86Alpha = 0.98 Alpha = 0.97 Alpha = 0.93

SD = 1.2mm SD = 2.1mmICC = 0.93 ICC = 0.68

Surgeon B RC = 3.4mm – RC = 5.8mmr = 0.94 r = 0.69Alpha = 0.97 Alpha = 0.82

SD = 1.1mm SD = 2.1mmICC = 0.83 ICC = 0.68

Surgeon C RC = 3.1mm RC = 5.8mm –r = 0.86 r = 0.69Alpha = 0.93 Alpha = 0.82

Fig. 4. (A) Bland & Altman limits of agreement for all the laxity values. (B) Bland & Altman limits of agreement for laxity values < 12mm.

to be higher and dependant on the average when the laxity was> 12mm (Fig. 4A). In the range of 0–12mm laxity Bland & Altmanhypotheses for agreement analysis were verified and the computedlimits of agreement were [−3.7, 4.9mm] with a mean difference of0.6mm (Fig. 4B).

Linear regression calculated on all data gave the following results:

KIN-Nav (mm) = 0.62 ∗ Rolimeter (mm) + 2.85 (mm)

with an R2 = 0.45 (adjusted R2 = 0.45) and a standard error of2.40mm (p < 0.0001).

4. Discussion

4.1. Repeatability of KIN-Nav

The results of the performed analysis of KIN-Nav laxity mea-surements are very encouraging. The mean standard deviation ofthe differences among repeated measures for an expert surgeon is0.7mm (smaller than those reported in intra-repeatability studieswith Rolimeter [14–16] or KT1000 [17–19,40]). This result may bedue to the fact that the navigation system has a more sensitivescale than arthrometer and also to the fact that for within-subjectstudies the navigation system is maintained on the limb while thearthrometer needs to be setup again for each measurement. Inthis study we used the Bland & Altman method to estimate thediscriminating power of the navigation system. The RC associated tothe KIN-Nav system used by the expert surgeon was 2.2mm, whichrepresents the minimal detectable difference with this methodologyand is significant in discriminating individual changes in a singlepatient. We noticed that this result appears to be dependant on thesurgeon's experience, and the least experienced surgeon may haveworse repeatability than more expert ones, as for arthrometers[18,41]. These results need to be confirmed with future analy-sis, as one limitation of the present study is the small number ofmeasurements for the estimation of inter-trial repeatability.

4.2. Comparison between KIN-Nav and rolimeter

The comparison between KIN-Nav and Rolimeter AP laxityevaluation showed a huge interval of agreement between the twomethods, as defined by Bland and Altman [37]. These high valuesdepended on the fact that both KIN-Nav and Rolimeter have high

284 N. Lopomo et al. / Computers in Biology and Medicine 39 (2009) 280–285

confidence limits for repeatability. However it is interesting to noticethat in this analysis showed that KIN-Nav had a bias of 1.1mm withrespect to Rolimeter evaluation and produced more values greaterthan 12mm. This may be due to the fact that KIN-Nav computesthe exact bone displacement, and is not affected by the soft tissueartefacts, that the Rolimeter is subject to. In addition, the KIN-Navprovides an infinite range of values and similar accuracy for all val-ues, while Rolimeter evaluation or reading may be difficult near thescale end. The fact that the comparison of navigated laxity and theRolimeter arthrometer values was not statistically significant be-cause the errors of both techniques have the same magnitude of thedifferences we would like to detect, suggests that the reliability ofnavigation systems for kinematic evaluations should be assessed byonly specific repeatability studies with each system and all surgeonsusing it during the clinical study.

5. Conclusions

This study investigates the reliability of a specific navigation sys-tem, KIN-Nav, for intra-operative kinematic evaluation of AP kneelaxity and compare it with an established method, i.e. the Rolimeterknee arthrometer [14–16]. It represents the first large study that aimsat statistically characterizing the reliability of a navigation systemused for measuring knee kinematics in normal and ACL injured kneeconditions, as previous studies were performed with small samplesizes, with simplified statistical analysis or on the general outcomeof surgery [22,27,28,32].

The use of a navigation system for kinematic values providesthe surgeon with a quantitative and reproducible estimation ofthe knee conditions, which is a critical factor for arthrometers.Therefore the navigation system may offer new interesting po-tentialities for within-subject intra-operative studies, such as thecomparison of different surgical techniques or associated injuries bymeasuring the individual changes from pre-operative to post-operative laxity/stability or the understanding of knee kinematicsin pathological and ACL-reconstructed knees by measuring laxitiesat different knee conditions in large set of patients. A further im-provement to the clinical reliability of KIN-Nav AP evaluation couldbe obtained by refining the intra-operative test execution, regardingthe loading conditions of the knee. At present the navigation systemis used to measure manual manoeuvres, not instrumented or guidedones. A better standardisation of this step of the laxity evaluationmight improve the inter-trial and inter-subject reproducibility ofKIN-Nav AP evaluation. Several doubts remain in the literatureabout both the suitable amount of load to be used and the load ap-plication conditions for instrumented laxity tests [13–15,17,20,40].In addition, some authors have shown that AP manual tests at max-imum force have a satisfactory repeatability with respect to testingwith a standardized load; therefore further data are necessary to un-derstand whether navigation systems perform better with manualor instrumented AP laxity tests [13–15,17,20,40].

In conclusion, this study suggests that KIN-Nav may represent anew method to measure and document AP laxity intra-operativelywith improved accuracy and test the effect of surgical treatmentin-vivo with higher sensitivity than in the past and this study quan-tify its reliability for within-subjects studies performed by a singleexpert surgeon. Moreover we defined new methodological approachto analyse the clinical reliability of a medical device, even if furtherinvestigations of its reliability for inter-surgeons and inter-patientsrepeatability should be performed for among-subjects studies ormulti-centric trials.

Conflict of interest statement

None declared.

Acknowledgement

We would like to acknowledge the help of the staff of the operat-ing theatre of the Traumatological division at Villa Toniolo, Bologna,Italy, for their effective collaboration in the optimisation and execu-tion of the experimental acquisitions. In particular we are gratefulto Pietro Guerra and Mirella Farina for their professional support. Inaddition, we thank the Carmelo Carcasio for his help in maintenanceof the custom tools, Giampaolo Bernagozzi for the preparation ofthis manuscript.

The authors would also like to thank AirCast Europe for the kinddonation of the Rolimeter device used in this study.

References

[1] A. Benjaminse, A. Gokeler, C.P. van der Schans, Clinical diagnosis of an anteriorcruciate ligament rupture: a meta-analysis, J. Orthop. Sports Phys. Ther. 36(2006) 267–288.

[2] S.J. Kim, H.K. Kim, Reliability of the anterior drawer test, the pivot shift test,and the Lachman test, Clin. Orthop. Relat. Res. 317 (1995) 237–242.

[3] R.J.P.M. Scholten, W. Opstelten, C.G. van der Plas, D. Bijl, W.L.J.M. Deville,L.M. Bouter, Accuracy of physical diagnostic tests for assessing ruptures of theanterior cruciate ligament: a meta-analysis, J. Fam. Pract. 52 (2003) 689–694.

[4] D.M. Daniel, M.L. Stone, R. Sachs, L. Malcom, Instrumented measurement ofanterior knee laxity in patients with acute anterior cruciate ligament disruption,Am. J. Sports Med. 13 (1985) 401–407.

[5] J.F. Benvenuti, J.A. Vallotton, J.L. Meystre, P.F. Leyvraz, Objective assessment ofthe anterior tibial translation in Lachman test position. Comparison betweenthree types of measurement, Knee Surg. Sports Traumatol. Arthrosc. 6 (1998)215–219.

[6] H.U. Staubli, B. Noesberger, R.P. Jakob, Stressradiography of the knee. Cruciateligament function studied in 138 patients, Acta Orthop. Scand. 249 (Suppl.)(1992) 1–27.

[7] J. Karrholm, G. Selvik, L.G. Elmqvist, L.I. Hansson, H. Jonsson, Three-dimensionalinstability of the anterior cruciate deficient knee, J. Bone Joint Surg. Br. 70(1988) 777–783.

[8] M.C. Logan, A. Williams, J. Lavelle, W. Gedroyc, M. Freeman, Tibiofemoralkinematics following successful anterior cruciate ligament reconstruction usingdynamic multiple resonance imaging, Am. J. Sports Med. 32 (2004) 984–992.

[9] M.C. Logan, A. Williams, J. Lavelle, W. Gedroyc, M. Freeman, What really happensduring the Lachman test? A dynamic MRI analysis of tibiofemoral motion, Am.J. Sports Med. 32 (2004) 369–375.

[10] M.T. Thompson, M.A. Conditt, S.K. Ismaily, A. Agarwal, P.C. Noble, Brief report:validation of a system for automated measurement of knee laxity, Clin. Biomech.(Bristol, Avon) 19 (2004) 308–312.

[11] A. Vergis, M. Hindriks, J. Gillquist, Sagittal plane translations of the knee inanterior cruciate deficient subjects and controls, Med. Sci. Sports Exercise 29(1997) 1561–1566.

[12] K.L. Markolf, A. Kochan, H.C. Amstutz, Measurement of knee stiffness and laxityin patients with documented absence of the anterior cruciate ligament, J. BoneJoint Surg. Am. 66 (1984) 242–252.

[13] B.S. Uh, B.D. Beynnon, D.L. Churchill, L.D. Haugh, M.A. Risberg, B.C. Fleming,A new device to measure knee laxity during weightbearing and non-weightbearing conditions, J. Orthop. Res. 19 (2001) 1185–1191.

[14] H. Balasch, M. Schiller, H. Friebel, F. Hoffmann, Evaluation of anterior knee jointinstability with the rolimeter. A test in comparison with manual assessmentand measuring with the kt-1000 arthrometer, Knee Surg. Sports Traumatol.Arthrosc. 7 (1999) 204–208.

[15] J. Hatcher, A. Hatcher, J. Arbuthnot, M. McNicholas, An investigation to examinethe inter-tester and intra-tester reliability of the rolimeter knee tester, and itssensitivity in identifying knee joint laxity, J. Orthop. Res. 23 (2005) 1399–1403.

[16] T. Muellner, W. Bugge, S. Johansen, C. Holtan, L. Engebretsen, Inter- andintratester comparison of the rolimeter knee tester: effect of tester's experienceand the examination technique, Knee Surg. Sports Traumatol. Arthrosc. 9 (2001)302–306.

[17] I.W. Forster, C.D. Warren-Smith, M. Tew, Is the kt1000 knee ligamentarthrometer reliable?, J. Bone Joint Surg. Br. 71 (1989) 843–847.

[18] D.L. Kowalk, E.M. Wojtys, J. Disher, P. Loubert, Quantitative analysis of themeasuring capabilities of the kt-1000 knee ligament arthrometer, Am. J. SportsMed. 21 (1993) 744–747.

[19] R.R. Wroble, L.A. Van Ginkel, E.S. Grood, F.R. Noyes, B.L. Shaffer, Repeatability ofthe kt-1000 arthrometer in a normal population, Am. J. Sports Med. 18 (1990)396–399.

[20] B.D. Beynnon, R.J. Johnson, B.C. Fleming, P.A. Renstrom, C.E. Nichols, M.H. Pope,L.D. Haugh, The measurement of elongation of anterior cruciate-ligament graftsin vivo, J. Bone Joint Surg. Am. 76 (1994) 520–531.

[21] D.M. Daniel, L.L. Malcom, G. Losse, M.L. Stone, R. Sachs, R. Burks, Instrumentedmeasurement of anterior laxity of the knee, J. Bone Joint Surg. Am. 67 (1985)720–726.

[22] Y. Ishibashi, E. Tsuda, K. Tazawa, H. Sato, S. Toh, Intraoperative evaluation ofthe anatomical double-bundle anterior cruciate ligament reconstruction withthe orthopilot navigation system, Orthopedics 28 (2005) s1277–s1282.

N. Lopomo et al. / Computers in Biology and Medicine 39 (2009) 280–285 285

[23] S. Zaffagnini, S. Bignozzi, S. Martelli, N. Imakiire, N. Lopomo, M. Marcacci,New intraoperative protocol for kinematic evaluation of acl reconstruction:preliminary results, Knee Surg. Sports Traumatol. Arthrosc. 14 (2006) 811–816.

[24] S. Zaffagnini, S. Martelli, S. Bignozzi, N. Imakiire, M. Marcacci, Navigatedintraoperative clinical test for kinematic assessment of acl graft behavior, in:F. Langlotz, B. Davies, D. Schlenzka (Eds.), in: Proceedings of the Fifth AnnualMeeting of CAOS-International, 2005, pp. 530–533.

[25] C. Luring, H. Bathis, M. Tingart, L. Perlick, J. Grifka, Computer assistance intotal knee replacement—a critical assessment of current health care technology,Comput. Aided Surg. 11 (2006) 77–80.

[26] S. Martelli, S. Zaffagnini, S. Bignozzi, M. Bontempi, M. Marcacci, Validation of anew protocol for computer-assisted evaluation of kinematics of double-bundleacl reconstruction, Clin. Biomech. (Bristol, Avon) 21 (2006) 279–287.

[27] F. Picard, A.M. DiGioia, J. Moody, V. Martinek, F.H. Fu, M. Rytel, C. Nikou,R.S. LaBarca, B. Jaramaz, Accuracy in tunnel placement for ACL reconstruction.Comparison of traditional arthroscopic and computer-assisted navigationtechniques, Comput. Aided Surg. 6 (2001) 279–289.

[28] R.P. Pitto, A.J. Graydon, L. Bradley, S.F. Malak, C.G. Walker, I.A. Anderson,Accuracy of a computer-assisted navigation system for total knee replacement,J. Bone Joint Surg. Br. 88 (2006) 601–605.

[29] B. Stockl, M. Nogler, R. Rosiek, M. Fischer, M. Krismer, O. Kessler, Navigationimproves accuracy of rotational alignment in total knee arthroplasty, Clin.Orthop. Relat. Res. 426 (2004) 180–186.

[30] S. Martelli, N. Lopomo, S. Bignozzi, S. Zaffagnini, A. Visani, Validation of a newprotocol for navigated intraoperative assessment of knee kinematics, Comput.Biol. Med. 37 (2007) 872–878.

[31] M. Marcacci, S. Zaffagnini, F. Iacono, M.P. Neri, I. Loreti, A. Petitto, Arthroscopicintra- and extra-articular anterior cruciate ligament reconstruction with gracilisand semitendinosus tendons, Knee Surg. Sports Traumatol. Arthrosc. 6 (1998)68–75.

[32] M. Marcacci, A.P. Molgora, S. Zaffagnini, A. Vascellari, F. Iacono, M.L.Presti, Anatomic double-bundle anterior cruciate ligament reconstruction withhamstrings, Arthroscopy 19 (2003) 540–546.

[33] S. Martelli, S. Zaffagnini, S. Bignozzi, N. Lopomo, F. Iacono, Navigation systemfor real-time intra-operative evaluation of knee kinematics, in: F. Langlotz, B.Davies, R. Ellis (Eds.), in: Proceedings of the Sixth Annual Meeting of CAOS-International, 2006, pp. 352–355.

[34] D.G. Altman, J.M. Bland, Comparing several groups using analysis of variance,Br. Med. J. 312 (1996) 1472–1473.

[35] D.G. Altman, J.M. Bland, Standard deviations and standard errors, Br. Med. J.331 (2005) 903.

[36] J.M. Bland, D.G. Altman, Measurement error and correlation coefficients, Br.Med. J. 313 (1996) 41–42.

[37] J.M. Bland, D.G. Altman, Measuring agreement in method comparison studies,Stat. Meth. Med. Res. 8 (1999) 135–160.

[38] R.G. Marx, Knee rating scales, Arthroscopy 19 (2003) 1103–1108.[39] J.A. Sterne, M. Egger, G.D. Smith, Systematic reviews in health care: investigating

and dealing with publication and other biases in meta-analysis, Br. Med. J. 323(2001) 101–105.

[40] M.E. Steiner, C. Brown, B. Zarins, B. Brownstein, P.S. Koval, P. Stone,Measurement of anterior-posterior displacement of the knee. A comparison ofthe results with instrumented devices and with clinical examination, J. BoneJoint Surg. Am. 72 (1990) 1307–1315.

[41] B.C. Fleming, B. Brattbakk, G.D. Peura, G.J. Badger, B.D. Beynnon, Measurementof anterior–posterior knee laxity: a comparison of three techniques, J. Orthop.Res. 20 (2002) 421–426.

Nicola Lopomo graduated in Biomedical Engineering at Politecnico di Milano, Milan(Italy), in December, 2003. Since June 2004 he has been working as a researcher inthe Biomechanics Laboratory at Codivilla-Putti Reserch Institute (Istituti OrtopediciRizzoli, Bologna, Italy). His research interests concern computer applications inmedicine (design and development of surgical navigation systems, instrumentations,navigated surgical techniques and system validation) and knee biomechanics. SinceJanuary 2005 he is also a Ph.D. Student in Bioengineering at Politecnico di Milano,Milan (Italy).

Simone Bignozzi graduated in Biomedical Engineering at University of Padova,Padova (Italy), in 2000. Since 2001 he has been working as a researcher in the Biome-chanics Laboratory at Codivilla-Putti Reserch Institute (Istituti Ortopedici Rizzoli,Bologna, Italy). His research interests concern computer applications in medicine(design and development of surgical navigation systems, instrumentations, navi-gated surgical techniques and system validation) and knee biomechanics.

Sandra Martelli received the Laurea in mathematics in 1987 from Pisa Univer-sity. She worked in algebra and applied mathematics at Scuola Normale Superi-ore (Pisa University, 1988) image processing in SELENIA research unit (Pisa, 1989),robotics at Scuola Superiore S. Anna (Arts Lab, Pisa, 1990–1992), computer-assistedsurgery and ACL modelling at TIMC (Grenoble University, 1993–1994). She joinedthe Biomechanics Laboratory at Codivilla-Putti Reserch Institute (Istituti OrtopediciRizzoli, Bologna, Italy) in 1994, where she is currently engaged in the developmentof computer-assisted surgery, computer simulations and knee biomechanics, RSAstudies. She is member of the International Society of Biomechanics (ISB), the In-ternational Society of Computer-Assisted Orthopaedic Surgery (CAOS), the SocietàItaliana di Chirurgia Assistita da Calcolatore, and collaborates with several Europeanuniversities.

Stefano Zaffagnini graduated in Medicine at the University of Bologna in Octo-ber 1987. He specialized in Orthopaedics and Traumatology from the University ofBologna in Jul 1992. Since 1997 he has been working as assistant orthopaedic sur-geon in the Sports Traumatology Department (now called “IX Divisione di ChirurgiaOrtopedico-Traumatologica”) of Istituti Ortopedici Rizzoli (Bologna, Italy). He wasESSKA AOSSM travelling fellow in June 1997 in USA. He also performed a 6 monthsFellowship in Sports Medicine and Orthopaedic Research at the University of Pitts-burgh (USA), under the direction of Professor F. Fu and Professor S. Woo. At presenthe works also as researcher in Biomechanics Laboratory of Codivilla-Putti ReserchInstitute (Istituti Ortopedici Rizzoli, Bologna, Italy). He took part in a lot of re-search project promoted by Istituti Ortopedici Rizzoli (Bologna, Italy), Italian HealthMinistry and European Community. He performed the lessons of “Orthopedics andTraumatology” in the School of Specialization in Sports Medicine at the Universityof Bologna as substitute of Professor M. Marcacci in the years 2004 and 2005.

Francesco Iacono graduated in Medicine and Surgery at University of Bologna(Italy), in 1989. He was specialized in Orthopaedics and Traumatology in 1994. Heis employed as a surgeon at IX Orthopaedic Division and Biomechanics Laboratoryat the Rizzoli Orthopaedic Institute. Since 1998 he has been attending the Biome-chanics Lab at Istituti Ortopedici Rizzoli. His main research interests concern hipand knee surgery, computer and robot assisted systems for knee surgery and kneebiomechanics.

Andrea Visani graduated in Medicine at the University of Bologna in January 1987.He specialized in Orthopaedics from the University of Bologna in July 1992. Atpresent he is a medical research coordinator in Biomechanics Laboratory of Codivilla-Putti Reserch Institute (Istituti Ortopedici Rizzoli, Bologna, Italy). He took part in alot of research project promoted by Istituti Ortopedici Rizzoli (Bologna, Italy), ItalianHealth Ministry and European Community.

Maurilio Marcacci graduated in Medicine and Surgery at the University of Pisain 1972. He specialized in Orthopaedics and Traumatology, in Medicine of theSport, and in Physiotherapy. He is full professor of University of Bologna, head ofIX Orthopaedic Division and Biomechanics Laboratory at the Rizzoli OrthopaedicInstitute. He is a university teacher at the Specialization School in: Orthopaedicsand Traumatology, Sport Medicine, Physical Medicine and Rehabilitation.