The first component of the imaging pipeline is the Modality LUT transformation, which compensates for any modality-specific presentation. 

There are two alternatives: the information is scaled in a linear manner or in a non-linear manner.  The non-linear modality LUT is sent a table, which is very common for CR/DR modalities.  For linear scaling, only a rescale/slope value is needed.  In the second case, a complete look-up table is sent, specifying for each pixel value the output value that it should be mapped into. 

Scaling the data is rather common when the presentation of negative values is required.  CT data is a good example of having a negative values. 

The value of a CT pixel is represented in Hounsfield values, after the inventor of the CT, Sir Godfrey N. Hounsfield (He actually received a Nobel prize for his invention).  This value relates the x-ray attenuation of water, which has a value of 0 Hounsfield.  Any matter that is more dense and attenuates the X-ray more than water, e.g., bone, will be represented with a positive pixel value;any matter that is less dense with a negative value.  Most modalities try to accommodate software implementers and shift or "scale" the data so that there are only positive pixel values represented internally for the software to work with.  However, if so stored, the data has to be rescaled prier to presentation.  A typical value for the rescale might be (-1024), to shift the data back 1014 units; a typical value for the slope might be 1, indicating a linear mapping (1 is the slope of the conversion formula).