Time-Resolved Multi-Spectral X-ray Computed Tomography of Cryoprotectant Diffusion Into Biomimetic Material

TL;DR

Multi-spectral X-ray computed tomography is introduced to noninvasively quantify the spatiotemporal distribution of cryoprotectants diffusing into a tissue mimicking phantom, and heterogeneous diffusion of the cryoprotectants into the tissue mimicking hydrogel is observed.

Abstract

Cryopreservation via vitrification requires loading cryoprotective cocktails. Insufficient loading may lead to freezing, precluding successful recovery; overloading is toxic. Yet, existing in situ measurements of cryoprotectant permeation remain largely unvalidated and do not resolve individual cryoprotectant concentrations. We introduce multi-spectral X-ray computed tomography (MSCT) to noninvasively quantify the spatiotemporal distribution of cryoprotectants diffusing into a tissue mimicking phantom. A developed photon-energy bin selection algorithm achieves sensitivity to low contrast cryoprotectants without contrast agents or fluorescence edges. The technique is validated with a dimethyl sulfoxide, glycerol, and water solution, resolving cryoprotectant volume fractions to within 5% accuracy. We observe heterogeneous diffusion of the cryoprotectants into the tissue mimicking hydrogel, a phenomenon not observable with conventional techniques. MSCT improves upon existing X ray CT methods because it is not underdetermined for multicomponent solutions and does not implicitly assume homogeneous diffusion. These advancements enable the systematic development of cryoprotectant loading protocols and provide diagnostics to assess vitrifiability before cryopreservation.

Figures (10)

• Figure 1: (a) The X-ray source is a tungsten-target YXLON FXE225.99 X-ray source with the directional head installed. The source emits a polychromatic cone beam with a 15° angle. The detector in this study is a Detection Technologies X-Card ME3 line detector capable of up to 128 photon energy bins. The approximate line detection region is indicated with a red line on the detector. (Not shown in figure \ref{['fig:setup']} is a custom cooling system to keep the ME3 detector cool during operation.) The sample is in a 15 mL conical tube, which is connected from above to a stack of rotation and centering stages. The source to detector distance, $d_{sd}$, is 1,367 mm; the source to object distance, $d_{so}$, is 140 mm. (b) The measured spectrum and the energy metric, $M(\varepsilon)$, for the range of emitted photon energies. Note that the balanced strategy does not necessarily choose the bins with the highest flux. (c) The linear attenuation coefficients for the three liquid components: DMSO, glycerol, and water. At low photon energies the DMSO has great contrast with glycerol and water, but glycerol and water have low contrast with one another. The balanced energy bin contrast metric considers this in choosing the optimal bins while the maximum contrast strategy does not.
• Figure 2: The mean measured volume fractions are plotted against the known true volume fractions for the two validation solutions using the different bin selection strategies. The developed energy bin selection algorithm, the balanced strategy, outperforms the maximum flux and maximum contrast strategies for the same regularization parameter, $\lambda = 7.91 \times 10^{-4}$, across both solutions. The maximum flux strategy performs comparably to the balanced strategy in solution 2, but is outperformed in solution 1. The maximum contrast strategy is consistently outperformed. Especially in the case of solution 1 the maximum contrast strategy struggles to accurately decompose the glycerol and DMSO.
• Figure 3: Volume fraction measurements of two well mixed solutions of known composition using the selected balanced energy bins compared to the known, true value. MSCT is able to measure the volume fraction accurately to within 5% error. The error bars show the 5th and 95th percentiles of volume fraction values of each component within the decomposed reconstructions, showing the precision of the decomposition. No significant image artifacts are observed.
• Figure 4: Volume fraction measurements of three representative trials immediately after inserting the hydrogel and one hour after inserting the hydrogel. The DMSO and glycerol are not observed to diffuse uniformly. Some diffusion is observed over the course of the 15 min of the first scan. Interestingly, glycerol is observed to diffuse more quickly than DMSO in this hydrogel. The volume fraction of water decreases as glycerol and DMSO diffuse into the hydrogel.
• Figure 5: The average radial change in volume fractions inside and outside the gel. DMSO and glycerol are seen to increase within the gel over time whereas the water decreases. Lighter shades are at later times. The gel edge is denoted with a vertical black line. The volume fraction of the hydrogel is measured to be constant, as expected, since the hydrogel is not being transported.