NIR Technology Australia
Bankstown NSW
2200
Tel: 02 9708 5068
Fax: 02 9708 5537
Application Note 03: Cropscan 2000G Performance Evaluation
February 2000
A performance evaluation was carried out at the Bread Research Institute (BRI) in Sydney, Australia on the Cropscan 2000G Portable Whole Grain Analyser to measure protein and moisture content in wheat. This evaluation was conducted at the request of NIR Technology Australia to have an evaluation of the Cropscan 2000G based on the USA National Institute of Standards and Testing (NIST) for Near Infrared (NIR) analysers. The testing protocol included:
Accuracy and Reproducibility
Ten (10) samples (equilibrated to room temperature) were analysed in
duplicate. The results were compared against a reference analyses for protein
and moisture. The Standard Error of Prediction (SEP) and Bias were calculated
as well as the Standard Deviation of Differences (SDD) of the duplicate
analyses.
Repeatability
The same ten samples above were analysed in duplicate over three (3)
consecutive days to assess the day-to-day repeatability. After comparison of
the results over the three days, the SDD was computed for each sample against
the average of the three daily readings.
Temperature Stability
The instrument was equilibrated in a refrigerated incubator at 5oC
for 4 hours along with five samples. The instrument was removed, switched on
and allowed to equilibrate for 10 minutes at room temperature. The five cold
samples were analysed in duplicate along with five samples kept at room
temperature and the results were compared against the average of the results
for the three previous days. The average bias was computed for each sample against
the three-day average.
In a similar experiment, the instrument and five samples were placed in the incubator set at 45oC for 4 hours. The samples along with five kept at room temperature were analysed as above and the average bias was computed.
Results and Specifications
The following table shows the results of the evaluation and the
performance specifications proposed by NIR Technology Australia.
|
Test |
Variable |
Cropscan 2000G Results (%) |
Specification (%) |
|
|
Accuracy |
Wheat Protein |
Bias |
0.17 |
0.20 |
|
|
|
SEP |
0.22 |
0.35 |
|
|
|
SDD |
0.15 |
0.15 |
|
|
Wheat Moisture |
Bias |
-0.34 |
0.20 |
|
|
|
SEP |
0.16 |
0.25 |
|
Reproducibility |
Wheat Protein |
SDD |
0.15 |
0.15 |
|
|
Wheat Moisture |
SDD |
0.05 |
0.15 |
|
Repeatability |
Wheat Protein |
SDD |
0.10 |
0.20 |
|
|
Wheat Moisture |
SDD |
0.06 |
0.20 |
|
Temperature Stability |
Wheat Protein |
Cold Bias |
0.13 |
0.35 |
|
|
|
Hot Bias |
-0.04 |
0.35 |
|
|
Wheat Moisture |
Cold Bias |
-0.13 |
0.30 |
|
|
|
Hot Bias |
-0.17 |
0.30 |
Conclusion
The Cropscan 2000G Whole Grain Analyser was calibrated for measuring protein
and moisture in wheat and subsequently used to measure these properties in independent
samples to an accuracy and precision consistent with NIST standards. The
Cropscan 2000G is an accurate NIR analyser providing stable results within a
wide temperature range.