Definitely, the first option is Brain MRI despite that it is not available in many Health Institutes. Without having available neither MRI nor CT scans, the best method to diagnose stroke is clinical data, interrogation of relatives of the patient and physical exam.
Best strategy in this setting is a thorough clinical and neurological assessment, ruling out diferential diagnoses, and then conservative treatment for stroke according to suspected origin of stroke (ischemic stroke: cardiac origin, arrhythmia-> anticoagulation, cerebrovascular origin: duplex ultrasound carotid arteries if available, antiplatelet therapy).
You really are at a horrible disadvantage without either CT or MRI, but I guess you could Doppler the internal carotid and ophthalmic arteries to look for grossly aberrant flow velocities.
Back in the bad old days, before CT, we resorted to angiography. It's risky but would be worth it only if there was a differential diagnosis including neoplasia.
I believe that in a low resource country, you will reach a lower cost per patient if you introduce a CT system, if you consider therapy costs.
That is because scanning your patients will allow to exclude many of those who won´t benefit from the expensive thrombolytic treatment and also reach other diagnosis, hard to establish solely with clinical evaluation.