MS stands for Mid/Side, and is traditionally used as a recording technique with the aid of a MS microphone. M stands for Mid or Mono, and is nothing more than the sum of the left and right channel. S stands for Side and is the difference channel, or left minus right. After LR->MS processing the left channel becomes L+R  (M) and the right channel becomes L-R (S). Decoding a MS signal back to stereo is a similar process. The following equations are being performed:

Encoding:

L out = M = L + R -> center information

R out = S = L – R -> side information

(The plugin actually stores the M channel in the left channel and the S channel in the right channel.)

Decoding:

L out = M + S

R out = M - S

When you change the balance of the M and S signals before decoding, the width of the stereo image changes. You are actually changing the level of the common information of the L en R channels! So, when for example you want to widen the image, just reduce the gain of the M channel because this reduces the common or mono information. When the S channel gain is reduced, the image shrinks because of the opposite effect. This is just a example of what you can do with this technique, more examples can be found in the helpfile.

You can for instance process the M and S channels individually. This is a mastering trick. For instance, compress only a voice signal in the middle, or enhance the stereo width of a cymbal by using eq on the S signal only.

Note:
When decoding from MS to stereo, a gain reduction of 6 dB is performed by the plugin to compensate for the inherent 6dB gain of the decoding:

L out = M + S = L + R + L – R = 2L

and

R out = M - S = L + R - L + R = 2R