K e v i n R. S t o n e , M . D . , D a v i d W. S to l le r , M . D . , S t e v e n G. I rv ing , Ca ro l i n E l m q u i s t , R . T . , a n d G i n n y G i l d e n g o r i n , P h . D .
Summary: Meniscal rep lacement by allograft and meniscal regenerat ion through collagen meniscal scaffolds have been recently reported. To evaluate the effectiveness of a replaced or regrown meniscal cartilage, a method for measuring the size and function of the regenerated tissue in vivo is required. To solve this problem, we developed and evaluated a magnetic resonance imaging (MRI) technique to measure the volume of meniscal tissues. Twenty-one intact fresh cadaver knees were evaluated and scanned with MRI for meniscal vol- ume sizing. The sizing sequence was repeated six times for each of 21 lateral and 12 medial menisci. The menisci were then excised and measured by water volume displacement. Each volume displacement measurement was repeated six times. The MRI technique employed to measure the volume of the menisci was shown to correspond to that of the standard measure of volume and was jus t as precise. However , the MRI technique consistently underestimated the actual volume. The average of the coefficient of variation for lateral volumes was 0.04 and 0.05 for the water and the MRI measurements, respectively. For medial measurements it was 0.04 and 0.06. The correlation for the lateral menisci was r = 0.45 (p = 0.04) and for the medial menisci it was r = 0.57 (p = 0.05). We conclude that 3D MRI is precise and repeatable but not accurate when used to measure meniscal volume in vivo and therefore may only be useful for evaluating changes in meniscal allografts and meniscal regeneration templates over time. Key Words: Meniscus- -MRI--Siz ing--Al lograf t .
Significant meniscal loss leads to progressive de- generative arthritis in the knee (1). The amount of tissue loss bears directly on the load bearing func- tion of the knee and the stress concentration on the tibial plateau (2-4). We have pursued a program to develop meniscal regeneration techniques and other investigators have pursued replacement of the me- niscus with allograft tissue (5,6). Since the size of the meniscus is related to its function, any effort to replace or regrow the cartilage must be accompa- nied by measurement techniques to determine me- niscal size.
Previous attempts to measure the size of the me- niscus have varied from simple ruler measurements
From The Stone Clinic (K.R.S., S.G.I., G.G.), and California Advanced Imaging (D.W.S., C.E.), San Francisco, California, U.S.A.
Address correspondence and reprint requests to Kevin R. Stone, M.D., The Stone Clinic, 3727 Buchanan Street, San Fran- cisco, CA 94123, U.S.A.
to more complex techniques involving sterophoto- grammetry, moldings, and ultrasound (7-14). No technique has been directly applied to the meniscus cartilage or validated with direct measurements. To address this issue, we chose to initially measure meniscal volume and compare that measurement to actual volume measurements by water displace- ment. Volume was specifically chosen after our pi- lot studies to evaluate measurements of meniscal size yielded variable results due to operator error in outlining the true meniscai borders. This study eval- uates whether 3D magnetic resonance imaging (MRI) volume measurement at this stage of devel- opment can be used as a tool for in vivo meniscus volume determinations.
MATERIALS AND METHODS
Twenty-one fresh frozen cadaveric knees were thawed for 24 h. Each knee was scanned with a 1.5
641
642 K. R. STONE ET AL.
Tesla General Electric magnetic resonance imager (Milwaukee, WI, U.S.A.) with a dedicated send and receive extremity coil. Imaging protocols used a TR (repetition time) of 55 ms, TE (echo time) of 15 ms, and a flip angle of 11 °. A 16-18 cm FoV (field of view) and 256 × 256 acquisition matrix was utilized. 3D volume acquisitions were performed at 0.7 mm slice thickness for a total of 32 images. 3D MRI raw data were archived on a Dimensional Medicine In- corporated (DMI) disc storage system. Meniscal cartilage is disarticulated from the joint, thereby isolating the menisci from surrounding tissue (Fig. 1). Tissue density thresholds were then established. Meniscal volumes were co-rendered with surface and internal fibrocartilage detail. Those menisci with previous meniscectomies or other damage were not included. For the full study, 21 lateral and 12 medial menisci were available for the MRI vol- ume measurement technique. The MRI measure- ments were repeated six times each prior to menis- cal excision for water displacement measurements. The object was to measure the precision of MRI sizing as compared to that of water volume dis- placement. Precision measurements were deemed important since MRI is an operator-dependent eval- uation.
Volume displacement measurement was per- formed according to the principles of Archimedes. Each knee's medial and lateral meniscus was ex- posed through a mid-line incision. The meniscus was excised and all nonfibrocartilage soft tissue de-
FIG. 1. Axial view of 3D MRI of an in vivo knee joint meniscus cartilage for meniscal volume measurement.
tached. The meniscus was placed into a 50-cc cali- brated beaker filled with saline. The saline was then micropipetted out to return saline to the original level. The amount of volume change before and af- ter placement was deemed to be the volume of the total meniscus. Each displacement measurement was repeated six times.
Statistical analysis Descriptive statistics were computed and analy-
sis performed for the 3D MRI and water displace- ment measurements of the lateral meniscus and the medial meniscus, as shown in Table 1. Student's t test for paired data was done on the coefficient of variation measurements to compare the precision of the MRI to the water displacement measurements. To compare the mean volume measurements of the two procedures, an analysis of variance model was used with repeated measures. In addition, data were plotted and a t test was performed to test for a significant correlation between the two measure- ments. A significance level of 0.05 was used for all statistical tests.
RESULTS
For the 21 lateral menisci and the 12 medial me- nisci both MRI and volume displacement measure- ments were repeated six times to compare the abil- ity to quantitate volume precisely and to examine any relation between the two techniques. As a mea- sure of precision, the coefficients of variation for MRI were not found to be significantly different from those for volume displacement data for the medial menisci (p = 0.168) and for the lateral me- nisci (p = 0.892) (Table 1). The average of the co- efficients of variation for the lateral volumes was 0.04 and 0.05 for water and MRI measurements, respectively, and 0.04 and 0.06 for the medial vol- umes.
The correlation for the lateral menisci was r = 0.45 (p = 0.042) and that for the medial menisci was r = 0.57 (p = 0.05) (Figs. 2,3). These results along with the difference in the mean volumes show MRI measurement to be significantly smaller than those of water displacement.
Equations of the linear regression lines fit to the data expressing the volume measurements as a function of the MRI data are:
Medial menisci(Volume) = 1.03 + 0.35*MRI
Lateral menisci(Volume) = 0.79 + 0.23*MRI
Arthroscopy, Vol. 10, No. 6, 1994
3D MRI VOLUME SIZING OF MENISCUS CARTILAGE 643
TABLE 1. Means, standard errors, and coefficients of variation for medial and lateral menisci
* Significantly different from the MRI data (p < 0.05).
The MRI technique employed to measure volume of the menisci of cadavers is shown to correspond to that of the standard measure of volume. In addi- tion, statistical analysis shows that the MRI tech- nique appears to be just as precise as that of the standard technique. However, MRI consistently underestimates the true volume of the meniscus.
DISCUSSION
Accurate measurement of meniscal characteris- tics would be helpful as a clinical evaluation tool for following meniscal replacement and regeneration techniques. There is no validated technique for in vivo evaluation of the size of the meniscus or its replacement. This study documents that volume measurements of meniscal cartilage by the 3D MRI technique described are linearly correlated with the volume displacement measurement but underesti- mate the true volume of the meniscus. The error is
most likely due to the complex multiplanar geome- try of the meniscus and the planar geometry of the MRI slices. Additionally, the edge of the meniscus at the meniscosynovial junction is difficult to pre- cisely determine. Because of the slice thickness, there is partial volume averaging between the me- niscus and the adjacent tissue. Thus an operator outlining the meniscus may err by being conserva- tive in choosing the border Where the meniscus is obvious and not in the " fuzzy" boundary between meniscus and synovial attachments. This error oc- curred in our previous efforts to measure meniscal size as well. This most likely explains the consistent undersizing by MRI as compared to water displace- ment. An edge detection algorithm may improve this measurement. If the clinical use of noninvasive MRI sizing of the meniscus is to follow either shrinkage, as has been noted in allografts, or tissue regeneration as had been noted through regenera- tion scaffolds, then the relative volumes may be
4
o o 3
, m
• ~ 2 eq
, , , , , ,
0
0
r = . 5 " 7
p= .052
n= 12
0
o o
o o
/ O
] ! I I
1 2 3 4 5
Volume (cc ]
FIG. 2. Medial meniscus volume by 3D MRI vs. water volume displacement.
o o 3
w
:E "m 2
r = .45
p= .042
n= 21
o O
O
0 I I I ........... I
0 1 2 3 4 5
Volume (ec} FIG. 3. Lateral meniscus volume by 3D MRI vs. water volume displacement.
Arthroscopy, Vol. 10, No. 6, 1994
644 K. R. STONE ET AL.
clinically useful. For this reason we performed the repeated measures analysis of the coefficients of variation. This analysis establishes that in the same patient, the MRI technique is consistent and thus possibly useful for following patients over time. The authors note that there are significant improve- ments that could be made to this technique before widespread application is possible. In our study we used 0.7-mm images, which may be beyond the nor- mal slice thickness for some MRI magnets. Addi- tionally, although this study demonstrated minimal variation in sequential measurements by the MRI technician, only a highly experienced orthopedic MRI technician with a high level of interest in this anatomy will be able to reliably produce the quality of images required for precise measurements. Lastly,for certain evaluations , shape and size mea- surements, not just volume, may be useful. We are currently working to automate this data acquisition and measurement process to facilitate widespread use.
REFERENCES
I. Burr DB, Radin EL. Meniscal function and the importance of meniscal regeneration in preventing late medial compart- ment osto-arthrosis. Clin Orthop 1982;171:121-6.
2. Ahmed AM, Burke DL. In vivo measurement of static pres- sure distribution in synovialjoints. J Biomech Eng 1983;105: 216.
3. Seedhom B. Loadbearing function of the menisci. Physio- therapy 1976;62.
4. Shoemaker SC, Markolf K. The role of the meniscus in the anterior posterior stability of the loaded anterior cruciate deficient knee. Effects of partial versus total excision. J Bone Joint Surg [Am] 1986;68:71-9.
5. Stone KR, Rodkey WG, Webber RJ, McKinney LA, Stead- man JR. Meniscal regeneration with copolymeric collagen scaffolds. In vitro and in vivo studies evaluated clinically, histologically, and biochemicaUy. Am J Sports Med 1992;20: 104-11.
6. Milachowski K, Weismeier K, Wirth CJ. Homologous me- niscus transplantation. Experimental and clinical results. Int Orthop 1989;13:1.
7. Alteshian GA, Soslowsky LJ, Froitnson MI, Lai WM, Mow VC. Stereophotogrammetric determinations of pateUar car- tilagc thickness and surface geometry. Adv Bioeng 1988;8: 167-70.
8. Ghosh SK. A close-range photogrammetric system for 3-D measurements and perspective diagramming in biomechan- ics. J Biomech 1983;16:667-74.
9. Huiskes R, Kremers J, Lange A, Woltring HJ, Selvik G, Rens TJ. Analytical stereophotogrammetric determination of three dimensional knee-joint geometry. J Biomech 1985; 18:559-70.
10. Modest VE, Murphy MC, Mann RW. Optical verification of a technique for ultrasonic measurement of articular cartilage thickness. J Biomech 1989;22:171-6.
11. Mow VC, Kuei SC, Lai WM, Armstrong CG. Biphasic creep and stress relaxation of articular cartilage theory and exper- iment. J Biomech Eng 1980;102:73-84.
12. Rushfeldt PD, Mann RW, Harris WH. Improved techniques for measuring in vitro the geometry and pressure distribution in the human acetabulum-I. Ultrasonic measurement of ac- etabular surfaces, sphericity and cartilage thickness. J Bio- mech 1981;14:253-60.
