Here is an excellent free full text review in Circulation.
A few important points from the review:
Initial imaging localizes the ischemic lesion and may identify the culprit artery.
Such imaging consists of MRI, MRA, CTA, carotid ultrasound, some combination thereof.
This initial evaluation allows placement of the ischemic stroke into one of several categories:
Based on the results of the initial neuroimaging studies, the stroke can be classified into preliminary diagnostic categories: large vessel occlusion, small vessel occlusion, potential cardioembolic, or unknown/other.
Small vessel strokes, aka lacunar strokes are defined by size (under 15 mm diameter) and subcortical location.
Secondary prevention in these cases involves antiplatelet therapy and management of vascular risk factors.
If no large artery culprit is identified, the stroke is not lacunar and the patient is not in atrial fibrillation then further investigation for a cardioembolic source is indicated.
This will consist of further monitoring for atrial fibrillation as well as TTE and/or TEE.
Intracranial large vessel stenosis is managed with antiplatelet therapy and standard risk factor modification.
Systemic anticoagulation and mechanical revascularization have not proven favorable in clinical trials.
Many patients with so called cryptogenic strokes are later found to have paroxysmal atrial fibrillation.
Outpatient event monitoring is recommended for patients whose strokes are considered cryptogenic at discharge.
Recommendations for duration of monitoring vary. The review author says at least two weeks.
Extensive thrombophilia testing is not recommended.
Mitral valve prolapse is no longer considered an important cause of cryptogenic stroke.
PFO management for secondary stroke prevention remains controversial.
High level studies do not support any special approach (systemic anticoagulation, closure) for most patients.