For filter measurements, method detection limit is reported. However, some take MDL as equivalent to average + 3 standard of blank filters while some treat MDL as just 3 standard deviation of filter blanks...Would be happy to know why this difference in approach exists and when one prefers over another?