Test
Q: What kind of image format do I need?
A: Your images must be in binary format (black and white) with a known pixel-to-pixel distance. They must show an object with an uninterrupted boundary.
Q: What kind of output will I get?
A: You get the values of 3 dimensioned variables, area, perimeter, and length of the major axis, plus values for 31 dimensionless variables. Their definitions are given here CLICK HERE.
Q: Are all the variables size-independent?
A: We supply the values of 3 dimensioned variables, area, perimeter, and length of the major axis. The remaining variables are dimensionless, either because they were dimensionless at the outset or rendered dimensionless by normalization to one of these 3. Although they are formally size-independent, our experience with cells shows that there is some size dependency left in the data. For our purposes, the most size-invariant variables are MINP, CAVS, PTOM, BMPS, WDTH, CENT, and SDFD.
Q: Are the data useful for quality control?
A: Yes. There are several ways of using the data in this way. If you plot the value of the variable with its 95% confidence interval, you can tell whether the mean of another dataset falls outside the CI bands. Or you can test the mean of the new sample to determine whether it differs significantly from that of the old sample. In practice, principal components may be more informative about what is changing (see below).
Q: Can I extract phenotypes that can be used to classify the types of objects I have?
A: Yes. For eukaryotic cells, the primary variables are not useful for defining phenotypes. You may need to do experiments with your own datasets to determine what works for them. A link to variables that are particularly good for solving cell features and phenotypes can be found here CLICK HERE.
Q: Can I compare two sets of objects where the pixel-to-pixel distance differs?
A: Yes, but you have to upload the images in different batches so that we receive the correct pixel dimensions. This doesn't affect size-independent variables' values but ensures your dimensioned variables are output correctly.
Q: What can I do if I am not receiving output from the processing stream?
A: We have suggestions for troubleshooting, if your IP stream fails to give results. CLICK HERE.