Distortion and instability compensation with deep learning for rotational scanning endoscopic optical coherence tomography.

Optical Coherence Tomography (OCT) is increasingly used in endoluminal procedures since it provides high-speed and high resolution imaging. Distortion and instability of images obtained with a proximal scanning endoscopic OCT system are significant due to the motor rotation irregularity, the friction between the rotating probe and outer sheath and synchronization issues. On-line compensation of artefacts is essential to ensure image quality suitable for real-time assistance during diagnosis or minimally invasive treatment. In this paper, we propose a new online correction method to tackle both B-scan distortion, video stream shaking and drift problem of endoscopic OCT linked to A-line level image shifting. The proposed computational approach for OCT scanning video correction integrates a Convolutional Neural Network (CNN) to improve the estimation of azimuthal shifting of each A-line. To suppress the accumulative error of integral estimation we also introduce another CNN branch to estimate a dynamic overall orientation angle. We train the network with semi-synthetic OCT videos by intentionally adding rotational distortion into real OCT scanning images. The results show that networks trained on this semi-synthetic data generalize to stabilize real OCT videos, and the algorithm efficacy is demonstrated on both ex vivo and in vivo data, where strong scanning artifacts are successfully corrected.

Geometric and mechanical evaluation of 3D-printing materials for skull base anatomical education and endoscopic surgery simulation - A first step to create reliable customized simulators.

INTRODUCTION: Endoscopic skull base surgery allows minimal invasive therapy through the nostrils to treat infectious or tumorous diseases. Surgical and anatomical education in this field is limited by the lack of validated training models in terms of geometric and mechanical accuracy. We choose to evaluate several consumer-grade materials to create a patient-specific 3D-printed skull base model for anatomical learning and surgical training. METHODS: Four 3D-printed consumer-grade materials were compared to human cadaver bone: calcium sulfate hemihydrate (named Multicolor), polyamide, resin and polycarbonate. We compared the geometric accuracy, forces required to break thin walls of materials and forces required during drilling. RESULTS: All materials had an acceptable global geometric accuracy (from 0.083mm to 0.203mm of global error). Local accuracy was better in polycarbonate (0.09mm) and polyamide (0.15mm) than in Multicolor (0.90mm) and resin (0.86mm). Resin and polyamide thin walls were not broken at 200N. Forces needed to break Multicolor thin walls were 1.6-3.5 times higher than in bone. For polycarbonate, forces applied were 1.6-2.5 times higher. Polycarbonate had a mode of fracture similar to the cadaver bone. Forces applied on materials during drilling followed a normal distribution except for the polyamide which was melted. Energy spent during drilling was respectively 1.6 and 2.6 times higher on bone than on PC and Multicolor. CONCLUSION: Polycarbonate is a good substitute of human cadaver bone for skull base surgery simulation. Thanks to short lead times and reasonable production costs, patient-specific 3D printed models can be used in clinical practice for pre-operative training, improving patient safety.