|Home | About | Journals | Submit | Contact Us | Français|
To develop a dedicated RF coil for high-resolution MR imaging of finger joints at 3T to improve diagnostic evaluation of arthritic diseases.
A dedicated cylindrical RF receive coil was developed for imaging finger joints at 3T. A planar coil, a saddle coil and a 1.5T coil with similar design as the dedicated coil were also constructed to compare imaging performance with the dedicated coil. A phantom was used for quantitative evaluation. Three-dimensional images were obtained on four subjects and a cadaver finger specimen using isotropic resolution of 160 microns in 9:32 minutes. The images were reviewed by 2 musculoskeletal radiologists.
The dedicated finger coil provided higher signal-to-noise and greater signal uniformity than the other coils. It supported high-resolution imaging that demonstrated anatomical details of the entire finger joint, and in the subject study, revealed abnormalities not detectable by traditional clinical resolution.
The dedicated finger coil optimizes the potential advantages of 3T scanners compared to lower field magnets. Use of this coil should facilitate early diagnosis, improve assessment of treatment response and provide better understanding of basic mechanisms that underlie arthritis.
Finger joints are often the first joints affected in rheumatoid (RA) and psoriatic arthritis (PsA) (1), and they are also commonly involved in osteoarthritis (OA) (2). In RA and PsA, the early months of joint inflammation are critical because during this period irreversible damage can occur (3, 4). Moreover, destruction of musculoskeletal structures is associated with loss of function (5). For example, the presence of bone erosions at the time of diagnosis is related to poor long-term radiographic and functional outcomes.
MRI is a widely used imaging modality for the diagnostic evaluation of arthritic diseases. While conventional radiographs detect advanced erosive changes in arthritis and visualize late signs of preceding disease activity, MRI can detect erosive changes with greater sensitivity, and allows direct visualization and assessment of synovitis, tendinitis and bone edema (6). Unfortunately, its use in the evaluation of inflammation in small joints such as fingers has been limited by inadequate image resolution (7, 8). Early joint changes such as small erosions and bone edema changes may be missed due to partial volume averaging (4, 9). Thus, there is a need to improve MRI resolution in finger joint imaging for early diagnosis of arthritis. Indeed, accurate diagnosis at the first presentation of arthritis may predict future radiographic damage and allow appropriate treatment to be targeted to patients with aggressive disease (4). Furthermore, higher MRI resolution will improve monitoring of therapeutic responses in not only the joints but also in entheses (10) and nail lesions (11) that are observed in PsA. Early diagnosis of arthritis of the hand joints is of particular relevance given the fact that highly effective, targeted therapies are available for the treatment of RA and PsA. In addition, differentiating PsA from erosive OA is important from a therapeutic perspective since anti-TNF agents are effective for the former but not the latter (12). Besides diagnostic and clinical trial evaluation, high-resolution MRI is also important for studies centered on the pathogenesis of inflammatory joint disorders and osteoarthritis (7, 8, 13).
RF coils with high signal sensitivity are required to support the high spatial resolution needed for the accurate evaluation of arthritis in fingers. Currently there is no commercially available dedicated RF coil for the fingers, and general-purpose coils do not provide the optimal signal and resolution for finger imaging. Though high-resolution MRI studies of fingers had been conducted using small surface coils (2, 14-16), the design and performance of these coils may not be optimal for finger imaging. Thus our objective was to develop a specifically designed finger coil that provides optimized signal-to-noise ratio (SNR), signal uniformity and image resolution for finger joints. As 3-Tesla (T) MRI scanners are becoming more available in diagnostic imaging environment, it is advantageous to conduct high-resolution finger MRI at 3T where SNR is twice that of conventional 1.5T. To date, few published studies of arthritic fingers at magnetic fields higher than 1.5T have been reported (17). Therefore, we conducted this pilot study on a 3T MRI scanner, and we tested for the highest image resolution achievable with a clinically acceptable scan time.
The study was conducted on a Siemens TRIO 3T whole-body MRI scanner (Siemens HealthCare, Erlangen, Germany). The maximum gradient amplitude was 40mT/m and the maximum slew rate was 200mT/m/ms. A specialized RF receive coil was developed for imaging finger joints (Figure 1a). It consisted of a cylindrical loop with inner diameter of 2.9cm and length of 1.2cm made with copper tape (3M, St. Paul, Minnesota, USA), and was tuned to 123.2 MHz. Its circuit diagram is shown in Figure 1c, in which L1, C1 (75pf), diode and choke form the detuning circuit to decouple the finger receive coil from the body transmit RF coil. C2 (270pf) is the output capacitor. The output cable has a characteristic impedance of 50 ohms, half-wavelength long (80cm), and includes a cable trap to reduce the common-mode current. The pre-amplifier was custom-made with a low noise figure of 0.4dB. During scanning, the finger was inserted through the coil and oriented transversely to the magnetic field, with the finger joint being imaged positioned at the center of the coil. Both in vivo and in vitro finger imaging was conducted. The study was approved by the Research Subject Review Board of our institution, and written informed consent was obtained from all subjects after the nature of the procedure had been fully explained.
To compare the imaging performance of the new dedicated coil with those used by other researchers in high-resolution finger MRI (2, 15, 16), we also built a planar circular RF receive coil (Figure 1b) and a linear saddle coil (not shown in figure). The planar coil was built using gauge-14 copper wire, and had the same diameter as the dedicated coil to allow fair comparison. A saddle coil was used in this study since it was simpler to construct and had similar imaging properties as the birdcage coil that had been used for finger imaging (15, 16). The saddle coil was made with 6mm wide copper tape. It had the same diameter as the dedicated coil, but had a longer length of 2.6cm (end-to-end) to provide similar longitudinal signal coverage as the dedicated coil. The imaging performances of the three coils were evaluated and compared quantitatively using a uniform phantom in a vinyl syringe tube with 2.2cm inner diameter. The phantom was made up of sodium chloride solution (0.38%) to simulate the coil loading of a finger. During scanning, the phantom was either inserted through the coils, or placed underneath the planar coil. Spin echo images were acquired in all three orthogonal planes using TR 500ms, TE 25ms, in-plane resolution 156μm, slice thickness 0.5mm and receive bandwidth 130Hz/pixel. The images obtained from the three coils were then compared with respect to SNR and signal uniformity.
To evaluate the advantages of imaging at 3T, the phantom study was also conducted at 1.5T for comparison. A 1.5T coil with the same design and dimensions as the dedicated 3T finger coil was built. Imaging test was conducted on a GE 1.5T whole body MR scanner (GE HealthCare, Waukesha, WI) using the same phantom and similar pulse sequence and parameters including the receive bandwidth.
In vitro imaging study was conducted on a cadaver finger specimen obtained via the National Disease Research Interchange from an adult male donor (age 56) with no record of arthritis. The specimen was kept in a tissue freezer, and was exposed to room temperature six hours before being imaged. The specimen, with the palmar side facing down, was scanned using both the dedicated finger coil and the planar coil. The planar coil was placed on top of the finger joint, as in the studies of Tan and other groups (2). High resolution 3D-FLASH sequence was used to obtain T1-weighted (T1W) coronal images with repetition time (TR) 33ms, echo time (TE) 9ms, flip angle 30°, field-of-view (FOV) 4cm × 2.25cm, matrix 256×144, in-plane resolution 156×156μm, slice thickness 160μm, 120 slices and scan time 9:32 minutes.
For the in vivo study, four subjects (one male, three female, age 26 to 64) with no clinical history of arthritis were included. The index finger proximal interphalangeal (PIP) joint of each subject was imaged using the dedicated finger coil. During scanning, the subjects were in a prone position with the forearm extended in front of the body and bended at right angle. T1W 3D-FLASH images were obtained using the same imaging parameters as in the in vitro study. Both fat-suppressed and non-fat-suppressed images were acquired. The scan time for each sequence was 9:32 minutes. Additional images were acquired using a clinical 2D T1W spin echo (SE) sequence protocol on a 56-year old female subject who had been experiencing tenderness in several finger joints including the one imaged. The imaging parameters for this SE sequence were TR 700ms, TE 15ms, FOV 6cm, matrix 256×256, slice thickness 2mm and scan time 3:03 minutes.
After scanning, the 3D images were reformatted and displayed in the three orthogonal planes of the imaged finger joint using the “3D” task card on the scanner console. The in vivo images were evaluated by two experienced radiologists for the depiction of anatomical structures and for the presence of abnormalities.
Using an Agilent E5061A network analyzer (Agilent Technologies, Santa Clara, CA, USA), the unloaded and loaded Q-factor of the dedicated finger coil was measured to be 246 and 187 respectively.
For the phantom study, SNR values measured in the transverse and longitudinal cross sections are plotted in figure 2. A circular region-of-interest 1.5mm in diameter was used to evaluate the signal values along the lines indicated in the figure, and noise standard deviation was measured from the image background. The results showed that the dedicated finger coil provided higher SNR and higher signal uniformity than the other coils in comparison. For the planar coil placed on top of the phantom, SNR at the position of the coil was comparable to that of the dedicated finger coil but dropped off quickly with distance to 40% and 17% of its original value at 1cm and 2cm from the coil respectively (Figure 2a). For the linear saddle coil, the SNR at the center of the coil was 61% of that of the dedicated coil. Though a quadrature saddle coil will in theory increase signal by 41%, the SNR will still be lower than that of the dedicated coil. Furthermore, the saddle coil has a similar sensitivity profile in the longitudinal direction as the dedicated coil, confirming the need of its longer length to provide similar signal coverage as the dedicated coil. For the 1.5T coil with the same geometry as the dedicated coil, the SNR at the center of the coil was 55% compared to the 3T dedicated coil. This result is consistent with the generally accepted theory that SNR increases linearly with magnetic field strength in clinical MRI systems.
The finger specimen study showed that the dedicated finger coil supported the high resolution used, and provided signal coverage for the entire joint (Figure 3). In contrast, the steep signal drop-off of the planar coil placed on top of the finger resulted in insufficient SNR in the middle and palmar side of the finger joint. Thus, structures such as the flexor tendon cannot be adequately visualized with the planar coil.
In the in vivo study, the dedicated finger coil was used successfully on all four subjects. It supported the high isotropic resolution of 160μm for the T1W 3D-FLASH sequence. Subsequent image reformation enabled the 3D data to be visualized in any planes to facilitate image evaluation. These high-resolution images were able to demonstrate anatomical details of the articular cartilage, subchondral bone, trabecular structures, joint capsule, collateral ligaments, musculotendinous structures and entheses (Figures (Figures44--6).6). Abnormalities were detected on two subjects. Figure 5 shows that the high-resolution images are able to reveal early osteophyte formation on the subject experiencing tenderness in the finger joint. These osteophytes were not observed on the conventional SE images. In figure 6, an evolving subchrondral cyst was seen on the images of a 64-year old asymptomatic male subject whose career involves routine machinery work. A cortical fissure was observed by the side of the lesion, which might not be routinely seen with standard imaging.
Recent literature has emphasized that effective instruments to diagnosis, monitor and determine prognosis are critical tools in the treatment of patients with early arthritis (3-5). In particular, MRI is a sensitive imaging modality for the detection of erosive changes and quantitative assessment of inflammation in the joint and surrounding structures. However, insufficient image resolution in MRI of finger joints leads to partial-volume artifacts that can simulate erosions and obscure small arthritic changes, thus hindering accurate evaluation (4, 7-9). In this pilot study, we demonstrated significant improvement in the resolution of finger MRI by development of a dedicated finger RF receive coil for imaging at 3T.
In order to obtain optimal signal sensitivity and uniform coverage of the whole finger joint, the new finger coil consists of a cylindrical loop constructed using copper tape. Cylindrical loop coil design had been shown to provide higher SNR than planar wire loop or flat strip design (18). Combining our special coil design with the SNR benefit of imaging at 3T, we obtained high quality in vivo and in vitro images of finger joints with isotropic resolution of 160μm within clinically acceptable time of 9:32 minutes. This resolution is significantly higher than those in published literature. Compared with a study (2) in which the acquired resolution was 156μm × 196μm × 1000μm (displayed resolution was 80μm × 100μm), our resolution (in terms of the inverse of voxel volume) is about 8 times higher. Furthermore, since our coil encompasses the whole finger joint, it provides uniform signal over the entire joint structure. In contrast, the planar coil placed on top of the finger joint resulted in signal decay and poor visualization of the middle and palmar regions of the finger joint. This may hinder the detection of abnormalities such as flexor tenosynovitis and dactylitis, two relatively common features in PsA patients (19).
Recently, another finger coil design was reported for MR angiography of fingers with systemic sclerosis (14). Though that coil provides uniform signal and moderately high resolution of 160μm × 210μm × 1200μm, its design is not suitable for imaging arthritic fingers because it has a relatively small diameter of 2.5cm which may not accommodate swollen finger joints and its long (about 6cm) cylindrical geometry may prohibit imaging arthritic fingers that cannot be straightened. On the contrary, a finger coil with similar geometrical design as the dedicated coil in this study was used successfully at 1.5T on all six PsA patients with different disease stages (20). For patients whose imaged finger could not be straightened, scanning was performed with the patient lying in a supine position with the palmar side of the finger facing upward.
Compared with a modified birdcage coil that was developed for finger imaging (15, 16), coil sensitivity plots indicate that our coil has higher signal uniformity. Besides, cylindrical coils that orient parallel to the static magnet field, such as birdcage coil and saddle coil, suffer from rapid signal drop-off towards both ends of the coil. As a result, they need to be significantly longer than the dedicated finger coil to provide the same signal coverage along the finger, which we have demonstrated by comparing the image performance of a saddle coil with that of the dedicated coil. This leads to 1) larger coil volume that reduces SNR and 2) longer coil length that may prohibit the coil from being used on arthritic fingers that cannot be straightened.
The 3D isotropic high-resolution acquisition obtained with the dedicated coil allows images to be reformatted and displayed in any planes with similar image quality. This is especially beneficial for the diagnosis of inflammatory arthritis since evaluation of erosive changes requires images in at least two orthogonal planes, as stated in the Outcome Measures in Rheumatology (OMERACT) Rheumatoid Arthritis MRI Scoring System (RAMRIS) (21). The 3D isotropic acquisition avoids the need for multiple acquisitions in different planes. In addition, it also facilitates quantitative analysis by providing information regarding the volumetric and spatial relationship between different MRI findings. Volumetric measurements of joint structures, such as the synovium, joint fluid and cartilage, are useful for the evaluation and prognosis of joint inflammation and damage (3).
The level of image resolution used in this study revealed detailed joint structures not observable in conventional MRI so images should be compared with a standard reference. We are in the process of conducting histology on the cadaveric finger specimen to correlate with the MR images in order to provide a standard reference. We will use a histological method that allows accurate matching of the histological sections with the corresponding MR images (22).
An obvious limitation of the dedicated finger coil is that it images one finger joint at a time. However, there are some clinical applications in which it is sufficient to image only one finger joint. One example is the differentiation of erosive OA from PsA, which is a common clinical problem and is an important one as anti-TNF treatment is effective for PsA but not OA. Another clinical utility in which it is adequate to image one finger joint is the evaluation of bone erosion and new bone formation in arthritis. For the future, we plan to develop a phased array coil that covers all the interphalangeal joints and the MCP joints from the second to the fifth fingers, while providing high SNR and uniformity to support high-resolution imaging.
In conclusion, we have demonstrated the potential of the new dedicated finger coil for high-resolution imaging of finger joints. The coil optimizes the potential advantages of 3T scanners over lower field magnets in imaging the small structures of finger joints, as was demonstrated in the comparison with a similarly designed coil at 1.5T. Our technique should be useful for the early diagnosis and assessment of treatment response of arthritis, and should also facilitate studies on the pathogenesis of degenerative and inflammatory joint disease.
The authors would like to thank Pat Weber for assistance in the subject scanning.
Grant Support: NIH/NIAMS 5R03AR048949