A Mn[Cl.sub.2]-based MR signal intensity linear response phantom.

Age-related intervertebral disk (IVD) degeneration is a major contributor to spinal morbidity in modern society) Previous authors have suggested that physiological (ie, nontraumatic) degenerative changes in the IVDs likely begin as early as the second decade of life and progress until death. (2) While degradation of proteoglycan metabolism and expression pathways have been heavily implicated in these changes, (3) the precise mechanism by which the process occurs remains unclear. (4) Independent of the finite molecular interactions, a clear link has been made between progressive degeneration of the IVDs and a reduction in relative disk water content. (2,5-7)

Orthopedic and biomechanical research using a variety of ex vivo and postmortem models has established a strong positive correlation between IVD hydration (independent of age) and physiological spinal performance. (8-10) While such a link suggests the potential appropriateness of rehydrating therapies and interventions to promote or restore function in otherwise degenerate disks, there currently are no universally accepted means for noninvasive description and quantification of the fluid content of in vivo fibrocartilage. Thus, longitudinal assessment of the efficacy of chondro-promoting treatment regimens has been difficult and has relied heavily on subjective patient outcome reporting. Such evaluation is further complicated by the high degree of both within-patient and between-patient variation in the magnetic resonance (MR) signal presentation of the lumbar IVD. Such variability prevents the broad application of standardized ("normal") reference intervals for identifying abnormality.

Research, clinical and anecdotal evidence has long suggested a direct correlation between observed MR signal intensity (SI) and tissue hydration. (11) Such a premise is supported by basic time-relaxation characteristics: Structures rich in hydrogen ions (ie, water) appear brighter under T2-weighted imaging conditions and, conversely, structures low in hydrogen ion content appear darker. (11) Despite this subjective MR phenomenon, there remains no accepted noninvasive model for determining relative water content of tissue in vivo.

Manganese chloride (Mn[Cl.sub.2]) has been described previously as an appropriate signal-enhancing material for use in clinical studies of physiological function (12-15) and its basic application as a contrast-variable MR imaging phantom also has been explored. (15,16) Manganese chloride is a relatively inexpensive and widely available inorganic compound that is safe to handle, easy to store and has a high degree of phantom reproducibility. (15) In an aqueous state (ie, bound or free fluid) Mn[Cl.sub.2] displays basic paramagnetic polarity (17) and appears to exhibit the somewhat unusual characteristic of being concentration-dependent signal enhancing under both basic T1- and T2-weighted imaging conditions. Collectively, these attributes make Mn[Cl.sub.2] an ideal agent for a contrast-variable MR phantom model and suggest that Mn[Cl.sub.2] might be a useful and appropriate calibration tool for assessing tissue hydration.

In the absence of evidence suggesting that such work has been undertaken previously, the purpose of this study was to develop and refine a Mn[Cl.sub.2]-based MR phantom that could be incorporated into conventional imaging protocols to allow comparative acute and longitudinal evaluation of relative IVD hydration.

Methods

A prospective, multiphase, progressive refinement approach was adopted to create a pair of 8-increment phantoms that would produce a predictable, concentration-dependent SI response profile under both basic T1and T2-weighted imaging conditions.

For initial SI analysis, a Mn[C1.sub.2] concentration range was selected based on earlier work by Matsushima et al, (18) which reported visualization of phantom SI markers in a 20% solution of concentration ranging from 0.001 to 0.5 mM. A series of 8 individual phantoms was created, using a serial dilution technique approximating this range, from a stock solution of 50.0 mM. The solution concentration was separated into 5 mL polypropylene screw-top tubes (4 cm x 1.5 cm) and set within a gelatin substrate to ameliorate the potentially artifact-inducing effect of water flux during MR scanning. The combined gelatin/Mn[Cl.sub.2] mixture then underwent passive curing in an autoclavable test-tube rack (Tetra Pharmaceuticals, Sefton, Australia) for 30 minutes at 70[degrees]C. The 8 sealed phantom tubes then were seated, in sequential concentration order, inside an MR-inert custom-built cradle device (see Figure 1).

[FIGURE 1 OMITTED]

With institutional ethical approval, a surgically harvested cadaveric human spine was inserted into the bore of the cradle device for imaging, along with the phantoms, to establish a Mn[Cl.sub.2] concentration range reflecting that seen in IVDs. The cradle device, phantoms and spine specimen were imaged using a standard 1.5 T Philips Intera MR scanner (Best, Netherlands) and a transmit-and-receive knee extremity coil for signal enhancement. A multiplanar scout image was used to plan an axial acquisition block at right angles to the long axis of the phantom tubes. In total, 6 contiguous slices were obtained using both a Tl-weighted fast-field-echo sequence (TR 400 ms; TE 6 ms; slice thickness 3.0 mm; FOV 160 mm; matrix 179/512r) and a T2-weighted spin-echo sequence (TR 2000 ms; TE 50 ms; slice thickness 3.0 mm; FOV 160 mm; matrix 128/512r) for each of the lumbar IVD levels.

The images were reviewed at a workstation using the manufacturer-provided display software (EasyVision; Philips Medical Systems, Andover, Massachusetts). To ensure measurement consistency, the midaxial slice of the imaging range (corresponding to the vertical center of the phantom) was selected for review. Initial SI measurements were generated from this slice by placing an observer-defined region-of-interest (ROI) marker over the center of the cross-sectional depiction of each individual phantom (see Figure 2). A measure of SI was recorded for each phantom visible on the slice. For instances in which a phantom cross-section could not be directly visualized or when the observed SI measured was less than 150% of the recorded background noise level, a not-visually-appreciable (NVA) SI was recorded. Region-of-interest size was maintained for all measurements to ensure consistency (area = 75.0 [mm.sup.2]; SD = +1.0 [mm.sup.2]). This process was repeated for both T1- and T2-weighted acquisitions. The collected SI data were then plotted against the known Mn[Cl.sub.2] concentration (see Figure 3) to allow subjective description of the distribution pattern, and linear-regression analysis was performed.

[FIGURE 2 OMITTED]

Despite poor initial depiction (T2) and demonstration of factorial linear agreement (T1), subjective interpretation of the preliminary scans suggested that specific regions of each response curve (marked with arrows on Figure 3) did in fact pertain to linear or near-linear concentration/SI interactions. Based on this observation, 2 new series of Mn[Cl.sub.2] phantoms were prepared to reflect new concentration ranges of 1.0 to 27.0 mM and 0.01 to 0.50 mM for T1- and T2-weighted imaging, respectively. Each new phantom, along with the original phantom as a control and indicator of measurement repeatability, was then subjected to T1- and T2-weighted imaging, SI derivation and linearity plotting.

[FIGURE 3 OMITTED]

The process described above was repeated progressively using a 2-phantom approach until incremental signal linearity corresponding to increasing relative phantom hydration was achieved. The final Mn[Cl.sub.2] concentrations for the 2 phantoms (1 specific for T1 imaging and 1 specific for T2 imaging) are shown in Table 1. Axial cross-sections demonstrating the MR appearance of the final phantom concentrations and elliptoid ROI placement for SI measurement are shown in Figure 4. The collected SI data were plotted again against the known Mn[Cl.sub.2] concentration and linear-regression analysis was performed.

Results

The graphs of SI vs known Mn[Cl.sub.2] concentration for phase 1 of the study are shown in Figure 3. Assessment of the T1 image plot demonstrated a nonlinear profile, suggesting a poor interaction relationship between the 2 variables (ie, Mn[Cl.sub.2] concentration and SI), while the T2 image plot only demonstrated 4 of the 8 phantoms, indicating that the selected concentrations fell outside of the optimal SI capture range.

The graphs of SI vs known Mn[Cl.sub.2] concentration for the final phase of the study using a 2-phantom approach are shown in Figure 5. The plots demonstrate a nearlinear profile, subjectively suggesting a consistent interaction relationship between the 2 variables within the selected concentration ranges. Linear-regression testing confirmed this association, demonstrating a 2-variable agreement coefficient of 0.968 for T1- and 0.942 for T2-weighted data plots.

Discussion

The aim of this study was to develop a Mn[Cl.sub.2]-based imaging phantom that would demonstrate progressive incremental MR signal response under both fundamental T1and T2-weighted imaging conditions, allowing correlation with tissue SI characteristics.