Matlab r2020a

The computer code shared within OpenEP has been under continual development for over a decade and is actively used within our research groups to analyze data from the major electroanatomic mapping platforms. This active use permits its ongoing development. The software described here is made available under the Apache License 2.0 and can be freely used for academic research.
Inspection of data exported from Velocity, Precision and Carto3 electroanatomic mapping system revealed two categories of electroanatomic mapping data – surface data and electrogram data. Individual exported datatypes representing all geometric and electrical data acquired by the mapping system were grouped into each of these categories. An etymology was designed categorizing each datatype into subgroups within these categories (see “Supplementary Material”). An implementation of OpenEP was developed using MATLAB R2020a (The MathWorks, Inc.).

A 3-dimensional motion capture system (VICON ver. 2.5, Oxford Metrics Ltd., Oxford, UK) was employed. It was equipped with ten infrared cameras (VICON Bonita, Oxford Metrics, UK) to capture joint kinematic data during lumbopelvic control training movements. These data were sampled at a rate of 120 Hz. Forty-five spherical retro-reflective markers (14 mm in diameter) were used, strategically positioned according to the Plug-in-Gait model’s anatomical landmarks [15 (link),16 (link)]. Our chosen motion analysis system was VICON, renowned as the golden standard [17 (link)]. It uses infrared detection to track the position of reflective markers accurately. The pelvic angles were calculated using the 3-dimensional motion capture system. A customized MATLAB R2020a software (MathWorks, Natick, MA, USA) was used to determine the range of anterior–posterior tilt (AP), upward–downward obliquity (UD), and internal–external rotation (IE), including the minimum and maximum values of the pelvic angles [13 (link)]. The joint kinematics data were filtered using a 2nd-order low-pass Butterworth filter with a 10 Hz cut-off frequency [14 (link)].

Since no baseline methods exist for comparison, we developed an approach to assess the extent to which deep learning outperforms traditional imaging informatics methods. We developed a tracking program in MATLAB R2020a (Mathworks) where cells are cropped from timepoint t and assessed on a pair-wise basis to identify whether its’ nearest neighbors on t + 1 correspond to the same cell at timepoint t. To determine this association, we employed a few simple heuristics and rules: (1) successful tracking required a structural similarity index (SSIM) greater than 0.2 between cropped volumes from different timepoints. SSIM is an indicator of similarity that considers structure, intensity, and contrast-based differences between images. We applied the assumption that if a cell exists at t + 1, the overall local environment should look rather similar at timepoint t, thus a correct association would have a moderate to high SSIM. (2) Similar to the post-processing used for Track-CNN, we estimated the average vector of all nearest neighbors to model local tissue movement in a cropped field of view from t to t + 1. This allowed us to evaluate if the current track from t to t + 1 flows in the same direction as the local shift of neighboring tracked cells. If the proposed track does not align with the local shift of neighboring tracked cells, then the track is terminated.

We applied the SBM method for CT estimation. All participants' data were processed with the CAT12 toolbox (http://www.neuro.uni-jena.de/cat/; version 1700) within SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12) and MATLAB R2020a (Mathworks, Natick, Massachusetts). Initially, all volumes were automatically segmented, where CT estimation and central cortical surface creation for each hemisphere were performed using the projection-based thickness (PBT) method (Dahnke et al., 2013 (link)). Topological correction (Yotter et al., 2011a (link)), spherical mapping (Yotter et al., 2011b (link)), and spherical registration (Ashburner, 2007 (link)) were subsequently carried out within one step. The output thickness data was resampled and smoothed in a subsequent step, applying 15 mm FWHM smoothing kernels.
Additionally, we extracted CT values of regions of interest (ROIs) from each individual surface, based on Desikan–Killiany Atlas (DK40) (Desikan et al., 2006 (link)). Mean values inside the referred ROIs were applied for correlation analysis.

Data was recorded using
TofPilot v.2.11.6.0.190ff674 (TOFWERK AG, Thun, Switzerland). The
LA-ICP-TOFMS data were saved in the open-source hierarchical data
format (HDF5, www.hdfgroup.org). Post-acquisition data processing was performed with Tofware v3.2.2.1,
which is a TOFWERK data analysis package and used as an add-on on
IgorPro (Wavemetrics Inc., Oregon). The data processing comprised
the following steps: (1) drift correction of the mass peak position
in the spectra over time via time-dependent mass calibration (2) determining
the peak shape and (3) fitting and subtracting the mass spectral baseline.
Data was further processed with HDIP version 1.6.6.d44415e5 (Teledyne
Photon Machines, Bozeman, MT). An integrated script was used to automatically
process the files generated by Tofware and to generate two-dimensional
(2D) elemental distribution maps. For calibration, signal responses
for each mass channel monitored during ablation of a single spiked
droplet were integrated using HDIP. The integrated signal intensities
and the absolute masses of the respective elements within the gelatin
micro-droplet standards were used to set up calibration curves.
Data processing for the semiquantitative calibration and custom-developed
semiquantitative calibration script was packaged in an online app
by MatLab R2020a (MathWorks, Natick, MA). Image processing and visualization
were performed in ImageJ 1.53.

All analyses were performed using MATLAB R2020a (version 9.8.0.1396136; The MathWorks, Inc., Natick, MA, USA). Categorical variables were compared using Chi‐square analysis or Tukey's honest significant difference test, as appropriate. P values ≤0.05 were considered statistically significant.