Simulation of MRI Related Tissue Changes Occurring during Formaldehyde Fixation of Human Brain Hemispheres

Introduction: Postmortem MRI of the human brain allows for invasive examination of the tissue specimen immediately following the MR scan, a practice that is not possible with living subjects. An increasing number of researchers have exploited this opportunity, correlating MR results from cadaveric brains with the findings of histological examination. However, previous investigations have shown that MR properties of the tissue vary with both time and position as fixation progresses, causing misinterpretation of MRI results. In this study, computer simulation was used to model the fixation process in cadaveric brain hemispheres immersed in formaldehyde solution. The resulting changes in the simulated T2 values were compared with previous experimental observations. 2 Methods: A realistically shaped 3D hemisphere model was created by masking an available T2 map from Reference 2. This tissue was treated as a homogeneous mass and was assumed to initially consist of 100 percent unfixed tissue (UT). A pool of formaldehyde formed a boundary around the tissue and diffused into the hemisphere in a Fickian manner (finite differences, voxel size = 1.25 × 1.25 × 1.5 mm, timestep = 1.0 s), with a diffusion coefficient of 8.0 × 10 cm·s, which is within the range of reported values. In each timestep, a portion of UT in each voxel was allowed to transform into either decomposed tissue (DT) or fixed tissue (FT) via autolysis and protein cross-linking, respectively. The rates of these processes were governed by rate constants and depended on the local concentrations of UT, FT, DT, formaldehyde, and autolytic enzyme. The T2 of each voxel was then calculated based on the state of the tissue (UT, FT, and DT composition) for each point in time. The simulated changes in T2 throughout the hemisphere were then compared to the actual T2 changes that had been observed previously. 2 Results & Discussion: The simulated patterns of T2 change were very similar to experimental results (Fig. 1). The main difference was that the changes occurred approximately twice as fast in the simulations compared to experiments, probably because the simulations treated the hemisphere as a homogeneous mass and also did not account for changes in the diffusion coefficient over time. However, experimental results and simulations showed the same behaviors for T2 values both near the edge of the hemisphere and in deep tissue. Near the edge of the hemisphere, T2 values dropped rapidly and then stabilized within days postmortem (Fig. 2). In the deep tissue of the hemisphere, T2 values also decreased, but took approximately two weeks rather than days to reach a minimum. According to simulation results, the T2 decrease was caused by the conversion of UT to FT (i.e., formation of protein cross-links) shortly after death (Fig. 3). The protein cross-links confer rigidity to the tissue, reducing the mobility of water molecules and reducing T2 values. Simulations also showed that formaldehyde was delayed in reaching the deepest parts of the tissue because of its relatively far distance from the surface, and consumption of the fixative by more superficial tissue. Therefore, the rate of fixation in deep tissue was slow compared to tissue near the surface. Additional decomposition during this slow fixation led to an increase in DT (Fig. 3) and allowed the water molecules to become more mobile, leading to the gradual increase in T2 (Fig. 2). This T2 increase has also been observed experimentally (Fig. 1). Finally, the deep tissue T2 values reached a plateau by approximately six weeks of fixation (Fig. 2), after the simulated values of UT, FT, and DT had largely stabilized (Fig. 3). The models used to simulate formaldehyde fixation in this study provide insight to the possible mechanisms of T2 change at different locations in the hemisphere over time. Therefore, in studies that combine histology and postmortem MR imaging of human brain hemispheres, it may be possible to account for the time dependent and location dependent variations of the MR properties of the tissue using models such as this one.