European Radiology

, Volume 14, Issue 9, pp 1568–1573

Fixed-flexion radiography of the knee provides reproducible joint space width measurements in osteoarthritis


  • Manish Kothari
    • Department of Scientific Client ServicesSynarc, Inc.
    • Department of Reading ServicesSynarc, Inc.
  • Gabriele von Ingersleben
    • Department of Reading ServicesSynarc, Inc.
  • Yves Miaux
    • Department of Reading ServicesSynarc, Inc.
  • Martine Sieffert
    • Department of Scientific Client ServicesSynarc, Inc.
  • Jon E. Block
    • Department of Scientific Client ServicesSynarc, Inc.
  • Randall Stevens
    • Hoffmann-La Roche
  • Charles G. Peterfy
    • Department of Scientific Client ServicesSynarc, Inc.

DOI: 10.1007/s00330-004-2312-6

Cite this article as:
Kothari, M., Guermazi, A., von Ingersleben, G. et al. Eur Radiol (2004) 14: 1568. doi:10.1007/s00330-004-2312-6


The validity of a non-fluoroscopic fixed-flexion radiographic acquisition and analysis protocol for measurement of joint space width (JSW) in knee osteoarthritis is determined. A cross-sectional study of 165 patients with documented knee osteoarthritis participating in a multicenter, prospective study of chondroprotective agents was performed. All patients had posteroanterior, weight-bearing, fixed-flexion radiography with 10° caudal beam angulation. A specially designed frame (SynaFlexer) was used to standardize the positioning. Minimum medial and lateral JSW were measured manually and twice by an automated analysis system to determine inter-technique and intra-reader concordance and reliability. A random subsample of 30 patients had repeat knee radiographs 2 weeks apart to estimate short-term reproducibility using automated analysis. Concordance between manual and automated medial JSW measurements was high (ICC=0.90); lateral compartment measurements showed somewhat less concordance (ICC=0.72). There was excellent concordance between repeated automated JSW measurements performed 6 months apart for the medial (ICC=0.94) and lateral (ICC=0.86) compartments. Short-term reproducibility for the subsample of 30 cases with repeat acquisitions demonstrated an average SD of 0.14 mm for medial JSW (CV=4.3%) and 0.23 mm for lateral JSW (CV=4.0%). Fixed-flexion radiography of the knee using a positioning device provides consistent, reliable and reproducible measurement of minimum JSW in knee osteoarthritis without the need for concurrent fluoroscopic guidance.


KneeOsteoarthritisRadiographyJoint space width


A broad-based research and development effort is underway to test and commercialize a variety of pharmacological agents aimed at inhibiting the pathogenetic processes related to the breakdown of articular cartilage and/or underlying subchondral bone associated with knee osteoarthritis [14]. Consequently, a critical need has emerged for accurate and precise radiographic outcome measures to evaluate the comparative effectiveness of these agents among large patient populations participating in multi-center clinical trials. A variety of methods have been proposed, but standard radiographic assessment of joint space width (JSW) is generally accepted as the most relevant [2]. Indeed, Buckland-Wright et al. [5] demonstrated that JSW measurements made in the weight-bearing tunnel view reliably reflected medial compartment cartilage thinning as determined by double contrast macroarthography.

Longitudinal changes in JSW measured on serial films are small relative to inherent biological or treatment response variability. Thus, maintaining a high degree of JSW measurement precision is critical. A number of studies have verified that knee radiographs taken in the traditional weight-bearing position with full knee extension provide inadequate JSW reproducibility for the evaluation of chondroprotective agents in multi-center trials of knee osteoarthritis [611]. Most of this inherent measurement error can be attributed to irreproducible image acquisition, particularly with respect to consistent positioning of the knee [12].

Several authors have confirmed that even minor alterations in image acquisition such as the amount of knee flexion or foot rotation, non-standardized X-ray beam alignment and variable distance between the knee and the X-ray cassette can have a substantial influence on JSW measurement precision [79, 1215]. Therefore, standardized image acquisition is essential, and some studies have shown that automated image analysis may provide additional improvements in reproducibility [7, 16, 17]. Importantly, imaging the knee in approximately 20–30° of flexion engages the central posterior aspect of the femoral articular cartilage, the region most markedly thinned in the arthritic disease process [18, 19]. However, achieving reproducible joint flexion even with the assistance of fluoroscopic guidance remains a formidable challenge in knee radiography.

The current paper introduces a technique for image acquisition and analysis in knee osteoarthritis. This method employs a specially designed positioning frame and calibration phantom that fixes knee flexion and positions the patient’s feet reproducibly. To assure that the X-ray beam is tangential with the floor of the medial tibial plateau, a 10° caudal X-ray beam angulation is used in all cases. These standardized radiographic acquisition parameters guarantee that the most affected region of articular cartilage is imaged reproducibly on initial and follow-up films. This study estimates the short-term JSW reproducibility of this technique and assesses whether the addition of an automated image analysis software method (SynaVu) that corrects for magnification and parallax effects offers further improvements in precision compared with manual JSW measurements.

Patients and methods


Four hundred and fifty seven patients were screened for entrance eligibility to participate in a randomized (Hoffman-La Roche trial), prospective multicenter (58 centers) trial of a chondroprotective osteoarthritis drug. Informed consent and IRB approval were obtained for all patients and all sites. Two hundred and thirty eight patients qualified for inclusion based on clinical and radiographic evidence of osteoarthritis including:
  • Chronic knee pain

  • An Osteoarthritis Research Society International (OARSI) grade of 1 or 2 [20]

  • A minimum JSW measurement in the medial compartment of between 2.0 and 4.0 mm determined by manual measurement

Approximately 70% (165/238) of these patients from 47 centers elected to participate and were monitored prospectively with serial radiography of both knees and follow-up clinical assessment. The average age of these patients was 60.5 years (±8 years), and most patients were female (119/165, 72%). The baseline radiographs of these patients represent the study material for the current investigation and include a random sample of 30 knee radiographs (30 patients) obtained from 14 geographically dispersed centers where standardized knee radiography was repeated twice within a 2-week period to estimate short-term JSW measurement precision. The mean age of this patient subsample was 59.4 years (±6.8 years) and 53% (16/30) of the cases were female.

Radiographic technique

A standardized radiographic technique was utilized to image the knee in all cases. Weight-bearing, posteroanterior (PA) radiography was undertaken without concurrent fluoroscopic guidance. Uniform anatomical alignment of the knee joint was facilitated by use of a proprietary positioning frame and calibration phantom (SynaFlexer) that places the patient’s feet reproducibly in 5° external rotation. The beam was centered on the back of the knees at the level of the joint line, defined by the horizontal skin crease of the popliteal fossa. Use of this positioning device requires that the thigh, patella and pelvis remain flush with the frame, coplanar with the tips of the great toes, resulting in a fixed knee angulation of approximately 20° flexion (i.e., “fixed-flexion”). Importantly, 10° caudal X-ray beam angulation provides correct projection of the tibial spines between the femoral condyles and superimposition of the anterior and posterior tibial rims. Fig. 1 illustrates the standardized patient positioning achieved with the SynaFlexer frame. A representative fixed-flexion PA radiograph showing concurrent imaging of magnification markers in the positioning frame is provided in Fig. 2.
Fig. 1

Schematic illustration of the non-fluoroscopic fixed-flexion radiographic knee protocol with 10° caudal beam angulation to ensure alignment of the beam with the medial tibial plateau. A standardized degree of knee flexion (20°) and external foot rotation (5°) are achieved with use of the SynaFlexer calibration and positioning frame

Fig. 2

Typical PA fixed-flexion radiograph showing simultaneous imaging of magnification markers incorporated in the SynaFlexer calibration and positioning frame. Note the proper alignment of the tibial plateau with superimposition of the anterior and posterior tibial rims

Analytical methods

Minimum JSW was determined by a trained radiologist from all radiographs using a manual measurement technique with a graduated magnifying lens. Medial and lateral compartment measurements were made separately in all cases. JSW measurements were also conducted by a second trained radiologist using an automated edge detection and analysis software system (SynaVu). This automated image processing engine has been previously validated [21, 22] and was licensed from the University of California at San Francisco (UCSF). This automated analysis was repeated by the same radiologist blinded to the initial results 6 months later. The analysis software suite utilizes calibration information obtained from the SynaFlexer phantom during radiography to verify beam angulation and knee side (right or left) and to correct for magnification differences and parallax errors between films. In all cases, automated results are presented with magnification correction. This image acquisition and analysis system provides automated film digitization and the software edge enhances the digital images using unsharp masking to maximize the delineation of cortical and trabecular bone (Fig. 3). Image quality also can be optimized using window and level adjustment, zoom and panning functions, a robust edge-detection algorithm to demarcate the weight-bearing articular surfaces as well as to calculate minimum JSW, average JSW and joint space area. To enhance methodological integrity during clinical trials, the system provides reader and chronological blinding in addition to automated databasing and archiving of images and JSW results.
Fig. 3

The SynaVu image analysis system showing automated JSW measurement. In addition to automatically verifying knee side and beam angulation and correcting magnification and parallax, the software automatically traces the articular surfaces of the femorotibial joints and localizes and calculates minimum medial and lateral JSW as well as joint space area. The three vertical lines represent the capability of the software to restrict the analysis to the desired areas of interest

Statistical methods

Utilizing the baseline films from the 165 patients participating in the prospective treatment study of knee osteoarthritis, the concordance and association between manual JSW measurements and the initial automated measurements with magnification correction were estimated using the intra-class correlation coefficient (ICC) [23, 24] and Pearson’s correlation coefficient [25], respectively, for medial and lateral compartments separately. Using the same study material, the initial and repeat automated measurements were compared using similar statistical procedures as above, and the reproducibility was estimated by calculating the corresponding coefficient of variation (CV). The short-term reproducibility of minimum JSW by the magnification-corrected automated analysis procedure was estimated by calculating the average standard deviation (SD) between measurements from the subset of paired radiographs (n=30) taken 2 weeks apart. The corresponding CV was calculated as an estimate of the magnitude of between-subject variability in JSW relative to the mean of JSW.


A high degree of concordance and association was demonstrated between manual and magnification-corrected automated baseline JSW measurements in the medial compartment among the study group of 165 patients (330 knees). The ICC and Pearson’s correlation coefficient were both 0.90 for this comparison. A lesser degree of concordance and association was observed for lateral compartment JSW measurements with an ICC of 0.72 and a Pearson’s coefficient of 0.74, respectively. Fig. 4 illustrates superimposed frequency distributions of medial JSW measurements for manual and automated measurements separately. Inspection of these distributions shows that only four baseline measurements (2.4%) were <2.0 mm by automated measurement, and these cases would not have qualified for study participation based on the study inclusion criterion of minimum medial JSW of 2.0–4.0 mm by manual measurement.
Fig. 4

Superimposed frequency distributions for 165 patients (330 knees) of minimum medial JSW by manual and automated measurement

There was excellent concordance and association between automated JSW measurements repeated 6 months apart by the same reviewer using the SynaVu software system with magnification correction. For the medial compartment, the ICC and Pearson’s coefficient were 0.94 and 0.95, respectively. The corresponding CV was 3.9%. For the lateral compartment, the ICC and Pearson’s coefficient were 0.86 and 0.89, respectively, and the corresponding CV was 5.6%.

Among the subsample of 30 cases with fixed-flexion knee radiography repeated twice in a 2-week period, the short-term reproducibility in the medial compartment using automated analysis was estimated by an average SD of 0.14 mm and a corresponding CV of 4.3%. In the lateral compartment, differences measured between repeat radiographs resulted in an average SD of 0.23 mm and a corresponding CV of 4.0% using automated analysis.


The standardized radiographic technique described in the current study fixes the knee reproducibly in approximately 20° flexion, thereby engaging the articular aspect of the medial compartment most severely affected in knee osteoarthritis, the central posterior region [18, 19]. By standardizing radiographic acquisition parameters such as the amount of knee flexion, degree of foot rotation and X-ray beam angulation and by controlling evaluation factors including magnification errors, the current automated technique either corrects or attenuates the known variability associated with many of the factors identified to degrade measurement precision [9, 14]. The fixed-flexion technique is essentially similar to the Lyon Schuss view [10], but without the necessity of fluoroscopic guidance to assist in aligning the X-ray beam. Piperno et al. [10] reported a short-term JSW reproducibility with a corresponding CV of 3.5% for the Lyon Schuss view, an estimate that compares favorably with the precision demonstrated herein using the fixed-flexion method. However, the fixed-flexion method does not require a dedicated technologist trained in fluoroscopic examination and spares the patient the added cost and radiation exposure associated with fluoroscopy.

Other radiographic methods have been developed to measure minimum JSW and to allow for some degree of knee flexion. A semi-flexed (i.e., 7–10°) anteroposterior view with fluoroscopic guidance has demonstrated reasonable short-term reproducibility (CV=5.5%) in 25 arthritic cases imaged at a single center [7], but considerably worse measurement precision (CV=8.7%) when the same technique was employed across five clinical centers [13]. Center-to-center variability was common with almost 30% of paired radiographs judged unsatisfactory with respect to the degree of knee flexion or foot rotation [13].

A non-fluoroscopically assisted protocol utilizing the same semi-flexed position with a PA horizontal beam has been proposed [8, 26]. In this method, the first metatarsophalangeal (MTP) joints are positioned directly beneath the front surface of the X-ray film cassette; patellae remain in contact with the cassette and aligned vertically with the first MTP joints. External foot rotation is standardized at 15° and maintained reproducibly with the aid of individual foot maps. Short-term reproducibility has been reported to be excellent (CV=1.6%) in a single center series of cases [8] and considerably worse (CV=8.0%) in a field test of the procedure [26]. In both reports of this technique, parallel alignment of the tibial plateau and the X-ray beam occurred in only about 30% of cases [8, 26]. This problem, coupled with the fact that the slight degree of flexion achieved (i.e., 7–10°) fails to capture radiographically the most affected articular region, raises some concern about its use in the serial assessment of chondroprotective agents.

We noted somewhat worse inter-technique and intra-reader concordance and reliability for lateral compared to medial compartment JSW measurements. However, short-term reproducibility estimates for the subsample of repeat knee radiographs demonstrated similarly low imprecision for medial (CV=4.3%) and lateral (CV=4.0%) compartment measurements. Buckland-Wright et al. [7] likewise noted inconsistencies with respect to JSW measurement precision between the medial and lateral compartment, with the medial compartment generally offering somewhat better reproducibility overall. Nevertheless, minimum medial JSW is uniformly considered the standard measurement parameter for clinical study eligibility and for monitoring longitudinal articular changes [12].

Using 2.0-mm medial JSW as the lower cutoff eligibility criterion for study participation, we found that only four out of 165 patients that were eligible by manual measurement would have been excluded by the higher precision automated SynaVu analysis system with magnification correction. The results of our correlation analysis indicated that manual measurements of JSW, when performed by an experienced radiologist, were concordant and associated with automated measurement methods. In contrast to longitudinal assessments of joint space narrowing (i.e., change in JSW), in which measurement precision is critical, the rate of incorrect inclusion or exclusion in patient selection was not substantially impacted by the choice of measurement technique in this study.


The current study provides initial validation results for the fixed-flexion radiographic technique in determining JSW in knee osteoarthritis. A high degree of concordance was demonstrated between manual and automated measurement techniques as well as when the automated measurements were repeated by the same reader. Moreover, short-term measurement reproducibility was high. These findings were particularly impressive in that the study material of knee radiographs was obtained from 47 clinical centers overall and from 14 centers for the subsample of cases with repeat radiographs. This suggests that the SynaFlexer positioning frame aids in eliminating many of the image acquisition errors that plague non-standardized radiographic protocols used to evaluate JSW.


This image acquisition for this study was funded by Hoffman-La Roche. Authors thank Melissa M. Ta from Synarc for providing the drawings.

Copyright information

© Springer-Verlag 2004