I think that the key to solve this is to dry as completely as possible, but with frequent intermediate weighing and associated scans.
Plotting the weights in a table will let you see when further weight loss has stopped, this will be close to zero % moisture as long as you are not cooking your sample.
– Now you can work back to estimate moisture at the earlier scans.
– Add this moisture content as an attribute measurement of each scan
– Try a Model to see if you get good results
Now try a fresh leaf, see what SCiO, predicts for moisture content, then measure the moisture by drying completely.
– If it is within a couple of percent – you are a winner! If not, more work is required.
Note 1: work on several leaves in parallel, scan each 5* at each stage. Don’t leave too much time between samplings to avoid biochemical changes. (Balance with rate of drying).
Note 2: you will need a very accurate balance, and good lab procedures (always weigh at a fixed temperature, avoid breezes etc).
Note 3: essential oils may evaporate also, keep dehydration temperatures gentle. Maybe estimate their quantity by washing a chopped leaf in ethanol (maybe ether?) dry both the washed leaf, and solvent to get base data. Repeat this for your slowly dried leaf. Remember to wash and dry a blank filter paper for comparison.
Note 4: A sachet of silica gel suspended near your leaf will help dehydration
Note 5: You could also scan a larger number of leaves and dry each completely, but the range of starting values may be too tightly clustered to get a reliable model.
Someone more skilled in lab practices or botany may laugh with scorn! Please feel free to criticise – I am not proud, and always ready to learn.