Abstract
A library of 524 low resolution steroid mass spectra was used to compare different methods of binary enconding and different similarity criteria. Recognition of spectra in the library was tested with a) noisy mass spectra, b) artificial mass spectra from steroid mixtures and c) actually recorded mass spectra. Best recognition rates were achieved by two methods: 1. Binary encoded mass spectra which contain 2 largest peaks in mass intervals of 14 mass units and a similarity criterion that does not consider peak heights. 2. Binary encoded mass spectra which contain only peaks above a threshold of 2 to 10% base peak and an “exclusive or” dissimilarity criterion.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
