Problem measuring relative amounts from scan of TLC plate

I used Imagej to analyze a thin layer chromatography (TLC) plate consisting of four lanes, each lane was spotted with a single one microliter aliquot of sample solution. (A TLC plate is somewhat similar to a gel electropherogram). Lane 1 was spotted with 5 mM solution, Lane 2 was spotted with 10 mM solution, Lane 3 was spotted with 25 mM solution and lane 5 was spotted with 50 mM solution. So each lane contains a known amount of sample. The TLC plate was developed, resulting in migration of the sample across the plate, similar to gel electrophoresis development.

The TLC plate contains a fluorescent compound that fluoresces a light yellow-green color when illuminated with ultraviolet light. However, the areas of the TLC plate that contain sample exhibit quenching of the fluorescence and show up as dark blue spots.

So, this TLC plate consists of lanes, each containing spots of sample, similar to a gel containing bands of protein or nucleic acid. The TLC plate was photographed and uploaded into Imagej

I analyzed this TLC plate using the Gel Analyzer tool of Imagej and I plotted the lanes and labeled the peaks. However the percentages of the peaks did not correlate with the known amounts of sample that were applied to the TLC plate. For example, the percentages for Lanes 1 and 2 (compared to Lanes 3 and 4) were twice as high as one would expect from the amounts that were spotted on the plate.

It appears that Imagej is measuring something other than the OD of the spots. Is there a way to set up Imagej to measure in OD so that the relative amounts of sample spotted on the TLC plate can be calculated?