Matlab 2019a

Data analysis was performed with MATLAB 2019a (The MathWorks, Inc, Natick, MA, USA). A minimum of 4 samples were examined per experimental condition. Unless otherwise indicated, all values are expressed as mean ± standard deviation.

All tasks were executed on a Windows PC and participant responses were recorded with a keyboard or mouse. The visuospatial task was implemented in Presentation version 14.1 (Neurobehavioral Systems, Inc.) and the paired-associates task was implemented in Psychtoolbox 3.0.13 (Brainard 1997 (link); Pelli 1997 (link); Kleiner et al. 2007 ) and MATLAB 2019a (The MathWorks, Inc.).

MATLAB 2019a (MathWorks, Natick, MA, USA) and the density estimation program “PDFestimator” were used to generate all the data presented in this work. The PDFestimator is a C++ program that JF and DJ developed as previously reported [11 (link)], which has the original Java program in supporting material. Upgrades on the PDFestimator are continuously being made on the BioMolecular Physics Group (BMPG) GitHub website, Available online: https://github.com/BioMolecularPhysicsGroup-UNCC/PDF-Estimator, where the source code is freely available, including a MATLAB interface to the C++ program. An older C++ version is also available in R, https://cran.r-project.org/web/packages/PDFEstimator/index.html. The version on the public GitHub website is the most recent stable version that has been well tested.

The pendular kinematics and indexes were drawn and calculated using Matlab 2019a (The MathWorks, Inc., Massachusetts, U.S.A.). All variables were expressed as means and standard deviations. Spearmen and Pearson’s partial correlation tests were conducted between functional and spasticity parameters by using SPSS 22 (IBM, Armonk, NY, U.S.A.). The variables were age, sex, body mass index (BMI), Brünnstrom recovery stage, scores of total, balance and gait components of TT, the shortest time of TUG test, comfortable and the fastest speeds of 10-MWT, the MMT and the MAS of extensors and flexors of paretic hip, knee, and ankle segments. Age, sex and BMI have been reported correlated to function outcomes after stroke^{45 (link),46 (link)}. The significance level was set as p < 0.05. Multicollinearity was determined by the tolerance (1-R²) and variance inflation factor (VIF, 1/(1-R²)), where R² is the coefficient of determination for the linear regression of a variable on all the other variables^{47 (link)}. Tolerance less than 0.5 or VIF larger than 2 indicates the existence of multicollinearity. To identify the influence of individual spasticity rather than motor recovery on the mobility performance, Brünnstrom recovery stage was controlled in the partial correlation test.

Although we assumed that original image features could represent lung texture properties different from wavelet-based image features, some features could be linearly dependent, especially features from original and LLL images. However, according to the LASSO theory²⁴ , that linear dependence does not imply dispensability, and individual dispensability does not imply pairwise dispensability, we decided to employ the original CT images. Nevertheless, to avoid the risk of overfitting on the RP prediction model, the significant features among 486 radiomic features were reduced to a number of features using a LASSO logistic regression with MATLAB 2019a (MathWorks)²⁵ . This process was repeated 1000 times for each ROI. The radiomic features with the highest frequency were extracted from the 486 radiomic features to build the RP predictive model for each ROI^{26 (link)}. The RP grades were annotated by 1 for RP = 2 or above and 0 for otherwise, as the teacher data to be inputted into the logistic regression models. The logistic regression models were constructed with the radiomic signatures for each ROI to classify patients with and without grade ≥ 2 RP.

The ROIs for calculating radiomic features were extracted using structure data for total lung volumes excluding the GTV and dosimetric data obtained from the treatment planning data. Four ROIs were created by extracting lung volumes irradiated with more than 0, 5, 10, and 20 Gy for each patient, which were defined as LV0, LV5, LV10, and LV20. The image processing was performed using in-house software with MATLAB 2019a (MathWorks).

To control for differences in performance between groups, arithmetic skills were tested using a computerized version of the Skagerlund arithmetic test (Skagerlund et al., 2019 (link)). The test includes four subtests, one for each equation type (addition, subtraction, multiplication, and division). In each subtest, the participant is asked to complete as many arithmetic problems as possible within 120 s. The difficulty of the problems increase within each subtest by increasing the number of digits included or by requiring borrowing and carrying. Each subtest includes 54 problems, except for the division subtest that contains 27 problems. The total score is the number of correctly completed problems, with a maximum score of 189. The test was implemented in Psychtoolbox running under MatLab 2019a (The MathWorks Inc., Natick, MA). Results were analyzed as a 2 × 4 ANOVA with group (deaf, hearing) as between‐group factor and equation type (addition, subtraction, multiplication, division) as within‐subject factor. Results from this test could be important in order to interpret imaging results. For example, if deaf individuals had performed as fast as hearing individuals in multiplication but still engaged the right intraparietal sulcus, it would have indicated qualitatively different processes.