Other possibly useful features in MeasureObjectSizeShape besides Form Factor would be Solidity, Compactness, and an array of Zernike features.
Besides the targeted measurement approach, an unbiased way to approach this issue would be to use MeasureObjectSizeShape to measure all of these features (as it does by default) and then use CellProfiler Analyst’s Classify tool to apply machine learning to uncover the features most relevant to your phenotype. Basically, use ExportToDatabase (sqlite is easiest) and save a CPAnalyst properties file. Run CPAnalyst, load the properties file, and train your classifier with “wavy” versus all others (presumably smooth). Then train, and inspect the rules that are produced. These are features sorted in most-to-least useful to distinguish your classes. Maybe to complicated for your use, but just a thought!