When subtracting background from a Western blot image, is it better to use a single region or multiple regions?

I am working on Western blot image analysis and I am trying to refine my method of subtracting the background from my target protein bands on my blot. The way that I am currently subtracting the background is to draw a rectangle (that is the same area as the rectangle outlining my protein band of interest) at the top of each lane, and subtract the integrated density of this background rectangle from the integrated density of my protein band.

However, I have read online that some people select an empty lane where no protein was loaded, and outline a rectangular region in that lane. This rectangular region is considered as the background, and is used for all protein bands.

I am wondering if this method is better than creating a background rectangle for each lane? The reason why I am making a background rectangle for each lane is because some of my blots have splotchy/uneven background, due to non-specific binding and air bubbles. But many of my blots have a uniform background.

Any insights are appreciated.