Objectives Disorders with chronically elevated ICP have salient imaging findings associated with the sella turcica and optic nerves. We aim to quantify the degree of correlation between imaging features and ICP.

Design Prospective case-cohort study.

Subjects One-hundred and twenty-six patients (35M:91F) underwent ICPM with recent MR imaging.

Methods T1-saggital views for sella volume, optic nerve vertical tortuosity, then T2-axial views for optic nerve sheath distension were blindly reviewed against respective median ICP and pulse amplitudes (PA). Imaging was triple reviewed for discordant values.

Results The mean ICP of four sella morphologies (full/flat/concave/empty) were 1.2, 4.8, 8.4 and 16.7 mmHg respectively (p<0.01). AUROC for sella morphology predicting ICP was 0.81. This measurement was able to detect minimum ICP of 5.3 mmHg with 73.0% sensitivity and specificity, 73.0% PPV and 69.8% NPV. The mean PA values were 4.0, 5.2, 6.1 and 9.6 mmHg respectively (p<0.01). AUROC for sella morphology predicting PA was 0.78. This measurement was able to detect minimum PA of 5.47 mmHg with 76.3% sensitivity, 79.5% specificity, 63.5% PPV and 81.0% NPV. Mean PA values for vertical tortuosity (nil/uni/bi) were 5.2, 7.1 and 7.0 mmHg respectively (p<0.05). Mean ICP values for rail tracking (nil/uni/bi) were 4.5, 7.5 and 15.7 mmHg respectively (p<0.01). Mean PA values were 5.2, 5.8 and 8.0 mmHg respectively (p<0.0001).

Conclusions Combined radiological features of ICP are promising non-invasive markers for raised ICP.