standardize features for feature selection, but not for model

In my pipeline I use the SelectFromModel step with a LinearRegression model for feature selection. It requires standardization of the input features using the StandardScaler. But I would like to maintain the features in their original units for the final model step.

Is there a way to accomplish this in a single sklearn pipeline? So that I only allow StandardScaler for the feature selection, but not for the model step to maintain the coefficients of the model in the original unit?