To correlate computed tomography findings and endoscopic localisation of the anterior ethmoidal artery during surgery, and to analyse the intranasal landmarks and abnormalities of the artery.

The anterior ethmoidal artery was studied with high-resolution computed tomography and endoscopic surgery in 30 patients undergoing functional endoscopic sinus surgery (group A), and with endoscopic dissection on 30 human cadavers (group B).

The anterior ethmoidal artery was demonstrated on computed tomography in 25 patients and intra-operatively in 12 (group A). It was identified in 26 cadavers (group B). Dehiscence of bony canal and branching was noted in 10.53 per cent of cases. The mean (± standard deviation) intranasal length of the anterior ethmoidal artery was 7.29 (± 1.21) mm, the distance of the artery from the axilla of the middle turbinate was 16.24 (± 2.75) mm, and the mean distance from the ground lamella was 8.97 (± 1.46) mm.

High-resolution computed tomography scanning prior to functional endoscopic sinus surgery is mandatory to identify the anterior ethmoidal artery. Endoscopically, the axilla of the middle turbinate and the ground lamella can serve as dependable reference points to identify the artery. Cadaver dissection improves understanding of anatomy.

Deep learning using convolutional neural networks represents a form of artificial intelligence where computers recognise patterns and make predictions based upon provided datasets. This study aimed to determine if a convolutional neural network could be trained to differentiate the location of the anterior ethmoidal artery as either adhered to the skull base or within a bone ‘mesentery’ on sinus computed tomography scans.

Coronal sinus computed tomography scans were reviewed by two otolaryngology residents for anterior ethmoidal artery location and used as data for the Google Inception-V3 convolutional neural network base. The classification layer of Inception-V3 was retrained in Python (programming language software) using a transfer learning method to interpret the computed tomography images.

A total of 675 images from 388 patients were used to train the convolutional neural network. A further 197 unique images were used to test the algorithm; this yielded a total accuracy of 82.7 per cent (95 per cent confidence interval = 77.7–87.8), kappa statistic of 0.62 and area under the curve of 0.86.

Convolutional neural networks demonstrate promise in identifying clinically important structures in functional endoscopic sinus surgery, such as anterior ethmoidal artery location on pre-operative sinus computed tomography.