I suspect that what you want are confidence intervals for the fitted regression line, since ‘standard deviation’ does not make sense in this context. Without the independent variable (x) data, it is not possible for me to do it.
With your original (x,y) data, use the fitlm function to repeat the fit, then use the predict function to create the confidence intervals. Plotting the confidence limits as lines is then straightforward. You can also use the errorbar function to plot the error bars at each point on the regression line as an alternative.
