= Total ion current normalization
\b[Total ion current normalization] rescales a <mass spectrum> using its total included ion signal. For nonnegative, equally spaced spectral samples $I_k$, a common implementation computes $S=\sum_k I_k$ and reports $J_k=I_k/S$ when $S>0$. The normalized values then sum to one. Multiplication by a common reference total gives another convention with the same relative ratios. The name refers to the summed spectral signal; a stored intensity need not itself be expressed in amperes.
This gives exact invariance to a common positive multiplier: replacing every $I_k$ by $cI_k$ leaves every $J_k$ unchanged. It does not establish that every experimental disturbance acts through such a multiplier. The <mass-to-charge ratio> range, excluded peaks and treatment of background determine which signal contributes to the <total ion current> used for normalization.
A worked example exposes the effect of an unrelated dominant peak. Start with three recorded intensities $(10,10,80)$, giving normalized values $(0.10,0.10,0.80)$. If only the third intensity doubles, the spectrum becomes $(10,10,160)$ and the first normalized value falls to $10/180=1/18\approx0.0556$. The first feature's raw signal has not changed. Its normalized value has decreased because the denominator now contains more of a different feature. A spectrum of all zeros has no defined unit-sum normalization; assigning it an arbitrary denominator conceals missing information.
In <MALDI> tissue images, locally abundant peaks produced this kind of misleading attenuation, motivating comparisons with median- and noise-based normalization.<Normalization in MALDI-TOF imaging datasets of proteins: practical considerations by Sören-Oliver Deininger and others>[“A pledge for normalization”, mouse-pancreas results and Figure 4: dominant insulin peaks and apparent attenuation of other signals.]{ref} <Total ion current normalization> can remove common intensity scaling, but <ion suppression> affecting only selected compounds violates that simple model. Report both the normalization rule and what the normalized image is intended to compare.
Back to article page
