Western Blot Imaging: Film vs CCD vs Scanner

No. A laser scanner works by exciting a fluorophore point by point and reading the filtered emission, so an ECL blot gives it nothing to excite, and its emission filters reject the wavelengths ECL produces. Chemiluminescence needs an integrating detector — film or a camera imager. This is a category mismatch rather than a performance limitation, and it is the most common reason a newly purchased imager turns out not to fit a lab’s existing workflow.

Because film’s response is described by the photographic characteristic curve: a toe where almost nothing develops, a midsection where density rises with the logarithm of exposure, and a shoulder where the grains are exhausted. There is no region where density is proportional to photon count. Worse, the developed sheet carries no record of where on that curve a band landed, so a saturated band and a strong band look identical and image-analysis software assigns them the same value. Bell’s 2016 BMC Biology commentary works through exactly this scenario, in which a saturated tubulin loading control concealed a genuine difference in sample loading.

For faint bands, generally yes. A cooled sensor integrating for the same duration collects the same photons with a characterised noise floor, whereas film has a threshold below which exposure produces essentially no density. Taylor et al. (2013) measured a linear range of four dilutions on film against seven on a cooled CCD for the same blot. Film is not the sensitivity advantage it is sometimes assumed to be; its real advantages are spatial resolution and requiring no capital equipment.

For this decision, nothing that matters. Most instruments now sold in this category use CMOS or sCMOS sensors, and "CCD imager" has become a partly legacy label for the class. Both are linear photoelectric integrating sensors, so the linearity, cooling, noise-floor and saturation-visibility arguments apply to both. Do not treat a CMOS specification sheet as describing a different kind of instrument from the "cooled CCD" systems in the older method literature.

They fail differently. Chemiluminescence has enzymatic amplification — one HRP molecule turns over many substrate molecules, so signal accumulates with time, which is a real advantage for a genuinely scarce target. Fluorescence has no amplification step, so its floor is set by background rather than by the label. For the faintest possible band, chemiluminescence on a cooled camera; for quantification across a range of abundances, fluorescence.

Visible-channel fluorescent blotting works, but it fights membrane autofluorescence. Nitrocellulose and PVDF, along with residual biological material and some blocking agents, autofluoresce appreciably in the visible range and much less in the near-infrared. Because a fluorescence limit of detection is set by background rather than by the label, moving the measurement into a spectral region where the membrane is dark lowers the floor more effectively than improving the detector would.

On a camera imager or scanner, the acquisition software flags saturated pixels in the raw file — check before quantifying, not after. On film you cannot know, which is the argument against it. The general validation is a two-fold dilution series imaged under one fixed exposure setting: plot background-subtracted density against micrograms loaded and identify where density stops tracking the load. That procedure, and the requirement that target and normaliser be linear at the same load, is set out on the loading-control page.

Only if the acquisition conditions were identical and both blots sat inside the validated linear range. Auto-exposure is the usual thing that quietly breaks this, because it optimises each image separately. On a scanner, a changed PMT gain does the same. Record exposure time, gain, binning and filter set, hold them fixed across an experiment, and prefer an internal normaliser measured on the same lane to any cross-blot comparison.