Matlab 2013a

Low-energy structures were identified using conformational searches consisting of 1000 individual steps using Macromodel 9.1 (Schrödinger Inc., Portland, OR, U.S.A.) using the OPLS2005 force field. A single search was done for small clusters, whereas up to five conformational searches starting with different initial structures were done for the larger clusters. Between two and five low-energy structures were reoptimized at the B3LYP/6-31++G** level of theory, followed by a harmonic frequency analysis. The water binding energy of H₂O to Gdm(H₂O)⁺ was obtained from various low-energy isomeric structures of Gdm(H₂O)₂⁺, correcting for the basis set superposition error using the counterpoise method. Q-Chem 4.0 (Q-Chem, Inc., Pittsburgh, PA, U.S.A.)^{51 (link)} was used for all quantum chemical computations. Relative Gibbs free energies as a function of temperature were determined from the rotational constants, unscaled harmonic frequencies and electronic ground state energies using an in-house Matlab 2013a (The MathWorks, Natick, MA, U.S.A.) routine.

Single molecule localization was performed using an ImageJ ThunderStorm plugin64 (link). The default values were used for the analysis (B-spline wavelet filter—order 3 and scale 2.0, approximate localization by eight-neighbourhood local maximum, Subpixel localization by PSF integrated Gaussian with the weighted least squares fitting method with a 3 pixels fitting radius and 1.6 pixels initial sigma). Particle coordinates and statistical properties were exported and further analysis was conducted using Matlab 2013a (MathWorks). Image drift and vibrations were corrected by mean image displacement values using a custom code in MATLAB.

For each acquisition series, the central lumen of the aortic root was segmented semi-automatically (Vitrea fX version 6.0, Vital Images, Inc.), to yield a 5 mL vascular volume-of-interest (VOI). This VOI was used to generate an enhancement curve that was automatically fit with a gamma variate function (LSQCurveFit, MatLab 2013a, MathWorks) [13 , 14 (link)], as shown in Eq. 1.
$\mathrm{A}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\left(\mathit{t}\right)=\mathit{A}\times {\left(\frac{\mathit{t}}{\mathit{\tau }}\right)}^{\mathit{B}}\times {\exp }^{\mathit{B}\mathbf{\times }(\mathbf{1}-\frac{\mathit{t}}{\mathit{\tau }})}+\mathit{C}$
A is the maximum enhancement, t is time, τ is the peak time, B is the growth factor, and C is the initial pre-contrast blood pool enhancement. The resulting fit curves were then used to identify the volume scan at the aortic root peak, after which the reference aortic root enhancement, coronary enhancement, and coronary CNR were determined. Specifically, the aortic VOI was used to measure the mean and standard deviation of the aortic root enhancement. Next, volumetric segments of the proximal left main (LM) and right coronary (RCA) arterial lumens were segmented semiautomatically (Vitrea fX version 6.0, Vital Images, Inc.), to measure the mean coronary enhancements. Finally, the CNR of the LM and RCA were calculated as the mean coronary enhancement minus the surrounding tissue enhancement normalized by the standard deviation of the aortic root enhancement. Finally, each aortic fit curve was used to simulate the optimal and standard CCTA protocols.

Analysis of temperature and motion was performed with dedicated software (MatLab 2013a, MathWorks, Natick, MA) by loading the grayscale bitmap files. For each animal, the individual average minimum and maximum temperature of all frames per time point were determined. The 15-min interval 2-min time points in all groups and the 1–9 min interval 1-min time points in the denervation-, sham-, and hypoxia control groups comprised 360 and 180 frames, respectively. Subsequently, the mean T_s was calculated per time point per group.
For motion analysis, the same thermographic images were used as for temperature analysis. Fluctuations in grayscale pixel intensity of consecutive images were determined and averaged per time point. A pixel was considered to reflect animal movement when the grayscale intensity difference exceeded 7 on a scale of 0 to 255, accounting for the background scatter. This cut-off was determined in pilot analyses of 11 120-s video sequences. Values were expressed as the mean ± SEM amount of pixels with ‘motion’ per group per time point.