With a pre-calibrated camera and optimized software settings, we claim that IRIS gives less variable results and can be used by anyone. In this technote, we asked a customer to put that claim to the test.
Purpose
Our aim of the present study was to compare Mabtech IRIS with another commercial FluoroSpot reader requiring user-defined settings, to evaluate whether user-defined input and experience in FluoroSpot analysis influences spot counts.
Method
Mabtech IRIS has fixed camera settings and RAWspot technology, whereas other commercial readers require user-defined settings of the camera, algorithm and XY-table. To assess the effect of user-defined input on analysis, an external organization was contracted and five operators of varying experience levels were given a FluoroSpot plate with wells in four replicates of an unknown sample. The operators were instructed to analyze the plate using two different readers: (i) Mabtech IRIS and (ii) a competitor FluoroSpot reader.
Results
Five operators counted unknown sample in four replicates. With IRIS, the operators’ average spot-forming units (SFU) ranged between 191-202 spots, resulting in a coefficient of variance (CV) of 3%. With the competitor reader, the operator's average SFU range was between 24-160 spots (CV 59%). A summary of the results is shown in the graph.
Conclusion
Mabtech IRIS minimizes the user-defined subjective input of the analysis, thereby reducing the influence of the operator on the end result.
Plug-and-play minimizes user-bias
Five operators were instructed to read the same plate on the Mabtech IRIS and a competitor reader
Accurate spot center is key
The algorithm RAWspot and pre-calibrated reader settings accurately detects each spot center and gives precise spot counts.