Development and Testing of a Hybrid Synthetic-Biologic Phantom for the Optimization of Mr Sequences for Articular Cartilage
MR: Magnetic Resonance;
MRI: Magnetic Resonance Imaging;
TKA: Total Knee Arthroplasty;
SE: Spin Echo;
TSE: Turbo Spin Echo;
GR: Gradient Echo;
DESS: Dual Echo Steady State;
TRUFI: True Fast Imaging With Steady-State Free Precession;
TR: Time Of Repetition;
TE: Time Of Echo;
FA: Flip Angle;
THK: Slice Thickness
use of high-field magnetic resonance (MR) scanners with increased signal-to-noise ratio, allowed increasing spatial resolution and acquisition times [3-5]. Moreover, the development of new imaging techniques such as three-dimensional (3D) acquisition and spectral fat suppression  further increased diagnostic values of MRI.More recently, specific MRI imaging relaxation protocols for articular cartilage mapping allowed to delineate and quantify early stage alterations related to the macromolecular structure of the cartilage (T2, T2*and T1ρ mapping) [6,7]. Indeed, compositional MRI reveals biochemical and microstructural changes in cartilage before morphologic alterations are detectable . However, the optimization of sequence parameters and the validation process required to assess the diagnostic reliability of these MRI protocols are needed before their introduction in the diagnostic routine. Since optimization and validation processes are excessively time consuming to be executed on patients in vivo, there is the need for reliable phantoms able to represent properly the complex morphology and chemical composition of the articular cartilage.
This study aimed at developing and validating a hybrid phantom, composed by synthetic components and biological tissue, to be used in the optimization of MRI protocols for the morphological and structural evaluation of the knee articular cartilage.
Materials and Methods
Patients and MRI protocol
Two patients have been enrolled for this study (case A: 67 year old female, case B: 74 year old male). Both of them were scheduled for total knee arthroplasty (TKA) because of advanced knee osteoarthritis without a history of rheumatoid arthritis or other joint-related infection. The study has been approved by the local Ethics Committee and patients gave informed consent.
In vivo MRI study of the knee cartilage was performed 12 hours prior to surgery (in-vivo MRI) on a 1.5 T scanner (Magnetom Aera®, Siemens Medical Systems, Erlangen, Germany). Sagittal images for morphological tissue evaluation were collected with spin echo (SE) T1-weighted (TE 12 ms, TR 500 ms THK 3 mm) dual echo steady state (DESS) (TE 7 ms, TR 19 ms, FA 25, THK 2 mm) and three dimensional true fast imaging with steady-state free precession (TRUFI 3D) (TE 4 ms, TR 10 ms, FA 28, THK 0.7 mm) sequences. Compositional mapping were obtained from SE T2 (5 echoes at TE at 13.8, 27.6, 41.4, 55.2, 69.0 ms, TR 1030 ms, THK 3 mm) and GR T2* (5 echoes at TE 4.2, 11.3, 18.5, 25.6, 32.7 ms, TR 420, FA 60, THK 3 mm) sequences.
Figure 1. Cartilage specimen collection from patients A undergoing TKA. a) Surgical field showing severe cartilage degeneration especially in the medial compartment of the knee. b) The seven osteochondral fragments obtained from surgery. From top to bottom: anterior central and posterior condyle fragments and tibial plate immediately after resection.
Both patients underwent total knee arthroplasty, where the seven osteochondral fragments (anterior medial and lateral condyle, central medial and lateral condyle, posterior medial and lateral condyle, tibial plateau) were collected (Figure 1), and identified for anatomical orientation adapting the method proposed by Li et al . Case A fragments were washed from blood excess with sterile saline and preserved by immersion in 10% formalin buffered solution. Case B fragments were washed and kept hydrated by using sterile saline moistened gauze in a plastic airtight container. Collected fragments were then used in setting-up the hybrid synthetic-biologic phantom as reported below.
The phantom consisted of a 1.8 L transparent polyethylene box filled with echographic gel (Cogel Ultrasound, Comedical, Italy).
A calibration squared grid was placed on the bottom of the container. Moreover, two “L” shaped polypropylene supports were installed for positioning the biological phantom components according to knee anatomy (Figure 2a and 2b). All osteochondral tissue fragments obtained from a single patient were fitted in the gel taking care of reproducing in-vivo orientation (Figure 2c). The phantom was completed by sealing the container with an airtight lid.
Phantom assemblage and MRI was perform 12 months after surgery for formalin-preserved case A fragments (fixed tissues), and 20 minutes after surgery for case B fragments (fresh tissues).
In-vitro MR data were acquired by scanning the above-described phantom with same imaging protocols used for the in vivo imaging. Positioning of the phantom within the MR scanner was realized to emulate patient’s knee orientation (Figure 2d).
T2 maps were reconstructed with a software specifically developed on Matlab programming platform, by using the sequence of five spin-echo magnitude images Mi acquired at echo times TEi ranging from 13.8 to 69.0 ms. A noise bias correction scheme for low signal-to-noise ratio MR images, adapted from the method proposed by McGibney and Smith  and Miller and Joseph , was employed. Assuming a Rician distribution of noise, the unbiased estimate of the power signal was computed, where the noise variance σ was estimated as the second order moment on a sample in a background region (air) of the image:|
The model for the power signal decay was then fitted on the Pi sequence, finding the optimal A0 and T2 values that minimize the sum of square errors, under the constraints of positive A0 and T2. The minimization was performed by a fast dyadic progressive refinement search, overcoming the caveats that small, or even negative, Pi values could pose to the conventional log-linear interpolation. A similar approach was applied
to obtain T2* maps.
Phantom was validated by comparing MR data of the same knee osteochondral tissue obtained in patient before TKA surgery (in-vivo) and in the phantom (in-vitro). Moreover, comparison of data obtained from in-vitro MRI of case A and case B, representative for formalin-fixed and fresh tissues respectively, allowed to assess any alteration to MR signal due to aldehydic fixation and prolonged storage time before phantom set-up.
The following aspects were considered for the evaluation of the in vivo and in vitro images:
1) Correct identification of the most significant anatomical sites (lateral and medial tibial articular surface and tibial spine morphology).
2) Morphological damage grading according to semi-quantitative scale (0: normal cartilage, 1: cartilage signal inhomogeneities with preserved tissue thickness, 2: cartilage tissue loss <50%, 3: cartilage tissue loss >50%, 4: exposition of subchondhral bone).
3) Consistency of color-coded maps of T2 and T2* relaxation times. The evaluation was performed blindly by two radiologists with expertise in musculoskeletal RM imaging and inter-observer agreement was assessed.
The viscosity of the gel and the optical clarity of the polymeric components used in the phantom facilitated the spatial positioning of the full set of osteochondral fragments emulating their anatomical location.
The correct anatomical orientation was evaluated by recognizing the anatomy of the tibial articular surface on in-vitro MR images and, more specifically, by discriminating the convexity of the lateral tibial plateau and the planarity of the medial tibial plateau. Moreover, morphology of tibial spine was also checked for consistency between in-vivo and in-vitro collected MR images. According to both radiologists, excellent concordance existed between the in-vivo anatomical morphology of the knee articular knee surfaces and the in-vitro corresponding reconstructed anatomy (Figure 3a and 3b, Figure 4a and 4b). A minor discrepancy was observed in the latero-lateral alignment of the condyles with respect to the tibial fragments within the phantom due to lack of anatomical links (cruciate ligaments) between fragments that were removed during TKA surgery.
The in-vitro MR imaging within the phantom allowed the evaluation of the whole articular surface of femoral condyles and tibial plateau, with the only exception of the trochlear groove region, which is not usually integrally available in the resection material during TKA.
Inter-observer agreement between radiologists was very strong for both location and grading of chondral lesion in the tissue imaged in-vivo and in-vitro. MR findings were also confirmed by evaluating the gross pathology aspects of the osteochondral fragment by an independent researcher (Figure 3c and Figure 4c).
In case A, the thickness of the cartilage was preserved in the lateral compartment and markedly reduced in the medial compartment. Focusing on the lateral compartment, the anterior portion of the condyle cartilage showed a diffuse hypointense tissue signal in T1 weighted images (Figures 3a and 3b) with a corresponding alteration of the values of the T2 relaxation time in the T2 maps (blue area in Figure 3d and 3e). The corresponding area evaluated by gross pathology showed the preservation of the cartilage surface integrity and thickness (evaluated by obtaining several osteochondral sample by a 2.5 mm Jamshidi needle), but a loss of the hyaline aspect typical of the healthy cartilage (Figure 3c) was also present.
In case B, a complete erosion or the cartilage with exposition of the subchondral bone was found on the medial compartment and on the posterior part of the lateral condyle. This feature was obvious at the gross pathology analysis of the lateral condyle fragments (Figure 4c). From the in-vitro MR images, a chondral elevated flap was recognized in between the central and posterior lateral condyle that was put on evidence during the surgical resection of the osteochondral samples (Figure 4b). This detail was not evident at the in-vivo MR analysis (Figure 4a).
Figure 3. Case A: comparative evaluation between in-vivo (a and d) and in-vitro formalin-fixed tissue (b and e). The macroscopic inspection of the surgical fragment (c) showed that the cartilage thickness is preserved but a discoloration is present in the central and anterior portion. T1 weighted images (a and b) and reconstructed T2 maps (d and e) refers to the imaging plane indicated by the red straight line on the tissue fragment in the macrophotography (c).
Bone cavities due to sampling with Jamshidi needle to evaluate tissue thickness are visible in (c) Red circles (d and e) indicateareas in the cartilage showing a variation in the relaxation time T2 both for in-vivo and in-vitro.
Images obtained in-vitro with DESS and TRUFI 3D sequences showed relevant limitations for the proper morphological evaluation of the cartilage surface because of their sensitivity to magnetic field inhomogeneity. This could be related to metallic
debris released by the surgical procedure or to the presence of small air bubbles within the phantom gel. It is therefore of theuttermost importance to avoid bubble formation when fitting osteochondral fragments in the phantom gel. Cartilage surfaceshould be deeply scrutinized before imaging the phantom and air bubbles should be removed.
Figure 4. Case B: comparative evaluation between in-vivo (a and d) and in-vitro fresh tissues (b and e). The macroscopic inspection of the surgical fragment (c) showed the preservation of the cartilage thickness in the anterior portion of the lateral condyle but a variation in the intensity of the MR signal is in accordance with the discoloration of the tissue. In the central portion, the cartilage delamination visible at the macroscopic inspection of the fragment (c) is more evident in-vitro (b) than in-vivo (a). Clearly T1 weighted images (a and b) and reconstructed T2 maps (d and e) refers to the imaging plane indicatedby the red straight line on the tissue fragment in the macrophotography (c).
The qualitative and quantitative comparison of relaxation time in T2 maps obtained from in-vitro MR imaging (Figures 3e and4e) were found spatially and quantitatively equivalent to those realized in-vivo (Figures 3d and 4d). As an example, the variation of the relaxation time T2 found in the cartilage of the anterior portion of the lateral condyle in case A in-vivo (Figure 3d) presented the same features in-vitro (Figure 3e).
Similarly to DESS and T2 TRUFI 3D sequences, T2* mapping of the tissues in-vitro resulted severely limited by sensitivity ofT2* images to local field inhomogeneity.
MRI of in-vitro fresh and formalin-preserved fragments resulted in similar morphology and composition (T2 relaxation times) of the chondral tissue. Minor variations in the relaxation times were present for the bone tissue in-vitro, especially when formalin fixation was performed (case A). However, these variations did not interfere with the evaluation of T2 map for chondral tissue. Taking these results into consideration, it seems feasible and reliable to realize the phantom with osteochondral tissue fragments fixed in formalin after a medium-long period of storage (several months), preserving the quality of the MR signal for both morphological and compositional analysis. However, considering that this explorative feasibility study has been conducted on a limited number of subjects, further research should be provided to support these preliminary data.
One of the major advantages of the phantom here described is the large availability of the osteochondral resections due to theincreasing number of TKA in the last years that facilitate the possibility of realizing a number of phantoms with differentchondral characteristics both in term of morphology and composition (relaxation times). Previous studies using osteochondralresection obtained from TKA were limited by the need of analyzing only fresh tissue [9,12]. Our data indicate that formalinfixed tissues has equivalent morphological characteristics to fresh tissue and T2 relaxation times for the chondral tissue are sufficiently preserved.
In a summary, the study showed a good correspondence between MR data obtained by using the proposed hybrid synthetic-biologic phantom and the in-vivo results, giving equivalent information about both the knee articular cartilage morphology and composition. The phantom can be therefore considered as a valid tool for the optimization of specific MR sequences for the articular knee cartilage.
Authors are grateful to the technical staff of the Division of Diagnostic Radiology at Rovereto Hospital for assistance in collecting MR data.
The study has been supported in part by the Healthcare Research
and Innovation Program (IRCS) of the Autonomous
Province of Trento.
The authors declare that they have no competing interests.
- Recht M, Bobic V, Burstein D, Disler D, Gold G, Gray M et al. Magnetic resonance imaging of articular cartilage. Clin Orthop Relat Res. 2001, 391:S379–396.
- Recht MP, Goodwin DW, Winalski CS, White LM. MRI of articular cartilage: revisiting current status and future directions. AJR Am J Roentgenol. 2005,185:899–914.
- Link T, Stahl R, Woertler K. Cartilage imaging: motivation, techniques, current and future significance. Eur Radiol. 2007,17:1135–1146.
- Gold G, Chen C, Koo S, Hargreaves B, Bangerter N. Recent advances in MRI of articular cartilage. AJR Am J Roentgenol. 2009,193:628–638.
- Burstein D, Gray M, Mosher T, Dardzinski B. Measures of molecular composition and structure in osteoarthritis. Radiol Clin North Am. 2009,47: 675–686.
- Crema MD, Roemer FW, Marra MD, Burstein D, Gold GE, Eckstein F et al. Articular cartilage in the knee: current MR imaging techniques and applications in clinical practice and research.
- Kijowski R, Blankenbaker DG, Munoz Del Rio A, Baer GS, Graf BK. Evaluation of the articular cartilage of the knee joint: value of adding a T2 mapping sequence to a routine MR imaging protocol.
- Roemer FW, Kijowski R, Guermazi A. Editorial: from theory to practice – the challenges of compositional MRI in osteoarthritis research. Osteoarthritis Cartilage. 2017, 25(12) 1923-1925.
- Li X, Cheng J, Lin K, Saadat E, Bolbos RI, Jobke B et al. Quantitative MRI using T1ρ and T2 in human osteoarthritic cartilage specimens: correlation with biochemical measurements and histology. Magn Reson Imaging. 2011,29(3): 324-334.
- McGibney G, Smith MR. An unbiased signal-to-noise ratio measure for magnetic resonance images. Med Phys. 1993, 20(4): 1077-1078.
- Miller AJ, Joseph PM. The use of power images to perform quantitative analysis on low SNR MR images. Magn Reson Imaging. 1993,11(7):1051-1056.
- Bittersohl B, Mamisch TC, Welsch GH, Stratmann J, Forst R, Swoboda B et al. Experimental model to evaluate in vivo and in vitro cartilage MR imaging by means of histological analyses. Eur J Radiol. 2009 Jun;70(3):561-569.