Seventy-two patients with juvenile angiofibroma have been investigated by computerized tomography (CT) and/or magnetic resonance imaging (MRI) over a period of 20 years. The evidence from these studies indicates that angiofibroma takes origin in the pterygo-palatine fossa at the aperture of the pterygoid (vidian) canal. An important extension of the tumour is posteriorly along the pterygoid canal with invasion of the cancellous bone of the pterygoid base, and greater wing of the sphenoid (60 per cent of patients). Distinctive features of angiofibroma are the high recurrence rate, and the rapidity with which many tumours recur. It is postulated that the principle determinant of recurrence is a high tumour growth rate at the time of surgery coupled with incomplete surgical excision. The inability to remove the tumour in toto is principally due to deep invasion of the sphenoid, as described above. In this series 93 per cent of recurrences occurred withthis type of tumour extension. A contributory cause in these patients is the use of pre-operative embolization. The treatment implications of these findings are examined.

A control series of biplane CT scans of the paranasal sinuses, derived from patients examined for orbital tumours, is described. The scans were assessed for the presence of anatomical variants in the middle meatus, said to contribute to meatal stenosis, and for signs of asymptomatic infection revealed by the presence of clouding or mucosal thickening in the sinuses.Of the anatomical variants, only concha bullosa (pneumatisation of the middle turbinate) was associated with a high incidence of infection in the sinuses (85 per cent). Evidence of asymptomatic sinus infection was as high as 39 per cent overall, the highest incidence occurring in the ethmoid cells (28 per cent). Isolated ethmoid clouding on CT was observed in 15 per cent and is likely to be found in as many as one in seven of the adult population in the UK. It is concluded that in the majority of patients clouding confined to a few ethmoid cells shown on CT is without clinical significance.The evidence from the control series did not support the concept that most sinus infection starts in the middle meatus. The presence of a large reservoir of quiescent or chronic sinus infection in the control group suggests that in most instances sinusitis derives from a recrudescence of this pre-existing infection.

CT scans of 100 patients from the Rhinology Clinic at the Royal National Throat, Nose and Ear Hospital were reviewed in order to test the precepts forming the basis of functional endoscopic sinus surgery, especially as they relate to the radiological investigation. These were: (a) the site of origin of sinus infection. (b) The relevance of certain anatomical variants in the middle meatus to sinus infection and (c) The use of CT as the radiological method of diagnosis in all cases.Obstruction in the middle meatus and ostiomeatal complex was associated with an increased incidence of opacity in the sinuses but the primary site of disease was not established: the concept that sinus disease takes origin in the middle meatus was not proven.Anatomical variants in the middle meatus were not associated with an increase in sinus opacity and there was no evidence that these anomalies have any effect on sinus disease by causing middle meatal stenosis.The radiological assessment of patients with inflammatory naso-sinus disease should start with plain X-ray. CT is unnecessary as a routine examination. It should be reserved for the pre-operative assessment of patients for endoscopic surgery, its main function being to show important anatomical landmarks.

Meticulous removal of angiofibroma infiltrating the pterygoid canal and basisphenoid is paramount to avoid "recurrence."

Imaging is an important component in the investigation of unilateral watery rhinorrhoea suspicious of cerebrospinal uid (CSF). Whilst the demonstration of the presence of beta 2 transferrin con rms that CSF is present it may prove dif cult to demonstrate the exact site of origin. Fine detail coronal computed tomography (CT) with sections of 1-2.mm thickness through the anterior skull base may show small dehiscences and fractures. The commonest site for congenital dehiscences is the cribriform niche adjacent to the vertical attachment of the middle turbinate anteriorly and the superior and lateral walls of the sphenoid posteriorly. In the presence of frequent or constant CSF rhinorrhoea a CT cisternogram can be helpful in de ning the exact site of the leak. Magnetic resonance imaging (MRI) is reserved for de ning the nature of soft tissue i.e. in ammatory tissue, meningoencephalocele or tumour. Finally, per-operative intrathecal uorescein is helpful when imaging does not prove positive. A management algorithm for CSF rhinorrhoea is presented.