Background Subtraction

Formalin-fixed paraffin-embedded (FFPE) samples and certain tissue types such as liver or brain may have high auto-fluorescence that significantly impact signal-to-noise ratio (SNR). In turn, low SNR may reduce the accuracy of gating, clustering and downstream spatial analysis. To improve SNR, blank cycles corresponding to the estimated auto-fluorescence can be acquired and subtracted from raw marker intensities.

Successful background subtraction requires 1) a start blank cycle, 2) an end blank cycle and 3) channel exposure times.

Cycle	Channel 1	Channel 2	Channel 3	Channel 4
1	DAPI-1	Blank	Blank	Blank
2	DAPI-2	Pan-cytokeratin	Ki67	CD68
3	DAPI-3	CD4	CD21	SMA
4	DAPI-4	Blank	Blank	Blank

Biomarkers must not contain 'Blank' in its name. 'Empty' channels will not be used for background subtraction. If variable exposure times are used, use the shortest exposure time of the corresponding channel to reduce over-subtraction.

Cycle	Channel 1	Channel 2	Channel 3	Channel 4
1	10 ms	500 ms	350 ms	500 ms
2	10 ms	500 ms	350 ms	600 ms
3	10 ms	500 ms	500 ms	500 ms
4	10 ms	500 ms	350 ms	500 ms

As a reference, default exposure times for Akoya-validated CODEX® conjugates are listed here.

Background signal is determined from the blank cycle. A multitude of CODEX experiments have demonstrated that over time (i.e., as cycle number increases) background can either increase or decrease, depending on tissue type and tissue preparation conditions. Thus, we subtract background from signal channels using both start and end blank channels to account for changes in autofluorescence during a CODEX experiment (Figure 1).

Additionally, we recognize that differences in exposure times between start blank, end blank, and signal channel may affect background subtraction. To account for differences in exposure time we use a ratio of signal exposure time to blank exposure time (Figure 2).

Figure 1. Background subtraction as a function of cycle number. Prior to subtraction, the intensity of start and end blank channels (i.e., autofluorescence at the beginning and end of the experiment) is scaled linearly by cycle number to account for changes in background intensity during the CODEX experiment.

Figure 2. Background subtraction as a function of cycle number and exposure time. To account for differences in exposure time between blank channels and signal channels background subtraction is applied using a ratio of signal exposure time to blank exposure time. We recommend setting blank exposure to the minimum signal channel exposure time.