MRI was used to investigate 100 patients with hemifacial spasm, using 3D-FT T2-weighted (CISS) and contrast-enhanced 3D-FT T1-weighted (turbo-FLASH) sequences in all cases. MR angiography was performed in 54 patients, using 3D-MT FISP images. Decompression of the facial nerve through a retromastoid craniotomy was performed in all patients. Hemifacial spasm caused by tumours in the cerebellopontine angle was not included. Vascular contact with the facial nerve root-exit zone or at the internal auditory canal was present in 96 of 100 patients with hemifacial spasm. The vessel responsible was the vertebral artery (VA) in 18 cases, the posterior inferior cerebellar artery (PICA) in 23, the anterior inferior cerebellar artery (AICA) in 22, the VA and PICA in 24, VA and AICA in 3, PICA and AICA in 1, VA, PICA and AICA in 4, and a vein in 1 case. CISS images showed compressive vascular loops better than contrast-enhanced turbo-FLASH images alone. The sensitivity of MRI was high, since only one false-negative case was found among the 100 patients who underwent surgery.

Sixty patients with primitive hemifacial spasm were treated by means of a minimally invasive retrosigmoid approach in which endoscopic and microsurgical procedures were combined. Intraoperative endoscopic examination of the cerebellopontine angle showed that for 56 of the patients vessel-nerve conflict was the cause of hemifacial spasm. The most common offending vessel was the posterior inferior cerebellar artery (39 patients), next was the vertebral artery (23 patients), and last was the anterior inferior cerebellar artery (16 patients). Nineteen of the patients had multiple offending vascular loops. In one patient, another cause of hemifacial spasm was an epidermoid tumor of the cerebellopontine angle. For three patients, it was not possible to determine the exact cause of the facial disorder. Follow-up information was reviewed for 54 of 60 patients; the mean follow-up period was 14 months. Fifty of the patients were in the vessel-nerve conflict group. Forty of the 50 were free of symptoms, and four had marked improvement. The overall success rate was 88%, and there was minimal morbidity (no facial palsy, two cases of severe hearing loss).