This paper deals with the diagnosability analysis of intermittent faults in discrete-event systems using a diagnoser-based approach. Intermittent faults are defined as faults that can automatically recover once they occur. Firstly, we discuss the modeling of intermittent faults in finite state automata and revisit two existing diagnosability properties regarding the detection of fault/recovery occurrences and the identification of system status. In the current work, we introduce a diagnoser variant, which consists in separating normal states, faulty states and recovered states in each diagnoser node. Such a structure serves to keep tracking the nominal, faulty, and recovered traces separately and more efficiently based on the diagnoser paths. Furthermore, a necessary and sufficient condition for checking the diagnosability is derived on the basis of the diagnoser structure. Besides, a systematic procedure for checking such a condition without needing any intermediate model is developed.