You may also force the planner to start by evaluating nodes for a label instead of an index. This is particularly useful if you are querying for nodes having two labels and you know that one of the labels is more restrictive than the other. In this example, we could have a billion locations in Japan so why look all of them to see if they have the Tower label? It's better to do it the other way.
MATCH (t:Location:Tower)USING SCAN t:TowerWHERE t.country='JPN'RETURN t