Analytical sensitivity

[Ben Buse]

Hi,

[With regard to] Analytical sensitivity maps, are these based on an average of neighboring pixels or for each pixel[?] The reason I ask is I have a crystal where the net counts/quant are bobbing around zero (some pixels plus, some negative) and [in] the analytical sensitivity map there is no z value across the crystal [...?]

Thanks

Ben

[John Donovan]

[With regard to] Analytical sensitivity maps, are these based on an average of neighboring pixels or for each pixel[?] The reason I ask is I have a crystal where the net counts/quant are bobbing around zero (some pixels plus, some negative) and [in] the analytical sensitivity map there is no z value across the crystal [...?]

Hi Ben,
I think you are asking if the CalcImage analytical sensitivity maps are based on each pixel or the average of several neighboring pixels...  and the answer is the former.

It's *exactly* the same analytical sensitivity calculation as is used for single point analyses in PFE as described here:

http://probesoftware.com/smf/index.php?topic=93.msg341#msg341

Also, when you say "there is no Z value across the crystal" do you mean it displays a gray color instead?  If so, yes, that is because the analytical sensitivity calculation is limited (as is documented) to non trace level concentrations. For trace level concentrations, please utilize the detection limit map output.

Some nice examples are attached below.  The gray pixels in the map indicate "no calculation".

[Ben Buse]

Hi John,

Thank you that's right its a gray color. And yes the detection limit maps work well.

Out of interest what's the cut off for analytical sensitivity. Is it based on X weight percent or when the analytical sensitivity exceeds a particular value

Many thanks

Ben

[John Donovan]

Thank you that's right its a gray color. And yes the detection limit maps work well.

Out of interest what's the cut off for analytical sensitivity. Is it based on X weight percent or when the analytical sensitivity exceeds a particular value

The numerator in the expression linked to above, has to be greater than (or equal to) zero to take the SQRT.

Try it on your calculator!   
john

[John Donovan]

Thank you that's right its a gray color. And yes the detection limit maps work well.

Out of interest what's the cut off for analytical sensitivity. Is it based on X weight percent or when the analytical sensitivity exceeds a particular value

The numerator in the expression linked to above, has to be greater than (or equal to) zero to take the SQRT.

Try it on your calculator!   
john

Hi Ben,
Actually I find that in addition to the above limitation, I am skipping the analytical sensitivity calculation for concentrations below 1 wt. % (elemental) exactly as you suspected:

Code: [Select]

' Store analytical sensitivity (percent error)
For i% = 1 To CalcImageOldSample(1).LastChan%
If CalcImageAnalysis.WtPercents!(i%) < MINIMUM_CONC_ANAL_SENS! Then  ' skip analytical sensitivity value if concentration is less than 1 percent
tPercentErrors!(i%) = BLANKINGVALUE!
Else
tPercentErrors!(i%) = CalcImageAnalysis.CalData!(1, i%)
End If
Next i%

The reason being that below 1 wt. % the analytical sensitivities tend to exceed 100% for typical mapping pixel dwell times.