Transnasal frontal intersinus septum takedown for frontal sinus pyocele.

INTRODUCTION: The open frontal intersinus septum takedown (FISST) technique was first described in 1976. We describe our experience with an endoscopic transnasal approach to manage a frontal sinus pyocele arising from an obstructed frontal sinus outflow tract due to anterolateral thigh flap reconstruction of a maxillectomy defect. CASE REPORT: A 40-year-old lady experienced upper eyelid swelling and purulent nasal discharge 3 weeks after undergoing a left extended medial maxillectomy with free anterolateral thigh flap reconstruction. A computed tomography (CT) scan revealed total opacification of the left frontal sinus. There was no improvement with intravenous antibiotics and she underwent a surgery, whenshe was found intraoperatively to have a frontal sinus pyocele, which was then drained. She then underwent an endoscopic transnasal FISST to ventilate the left frontal sinus via the contralateral frontal recess with good results. A CT scan performed 3 months postoperatively showed a widely patent interfrontal sinus septal window and right frontal outflow tract with no disease recurrence. DISCUSSION: The FISST is a useful technique to manage unilateral frontal sinus disease by taking advantage of the contralateral outflow tract when the ipsilateral frontal recess is obstructed.

Fast methods for training Gaussian processes on large datasets.

Gaussian process regression (GPR) is a non-parametric Bayesian technique for interpolating or fitting data. The main barrier to further uptake of this powerful tool rests in the computational costs associated with the matrices which arise when dealing with large datasets. Here, we derive some simple results which we have found useful for speeding up the learning stage in the GPR algorithm, and especially for performing Bayesian model comparison between different covariance functions. We apply our techniques to both synthetic and real data and quantify the speed-up relative to using nested sampling to numerically evaluate model evidences.