Search

International audience; Purpose - Annotation of surgical videos is a time-consuming task which requires specific knowledge. In this paper, we present and evaluate a deep learning-based method that includes pre-annotation of the phases and steps in surgical videos and user assistance in the annotation process. Methods - We propose a classification function that automatically detects errors and infers temporal coherence in predictions made by a convolutional neural network. First, we trained three different architectures of neural networks to assess the method on two surgical procedures: cholecystectomy and cataract surgery. The proposed method was then implemented in an annotation software to test its ability to assist surgical video annotation. A user study was conducted to validate our approach, in which participants had to annotate the phases and the steps of a cataract surgery video. The annotation and the completion time were recorded. Results - The participants who used the assistance system were 7% more accurate on the step annotation and 10 min faster than the participants who used the manual system. The results of the questionnaire showed that the assistance system did not disturb the participants and did not complicate the task. Conclusion - The annotation process is a difficult and time-consuming task essential to train deep learning algorithms. In this publication, we propose a method to assist the annotation of surgical workflows which was validated through a user study. The proposed assistance system significantly improved annotation duration and accuracy.

In minimally-invasive procedures like biopsy, the physician has to insert a needle into the tissues of a patient to reach a target. Currently, this task is mostly performed manually and under visual guidance. However, manual needle insertion can result in a large final positioning error of the tip that might lead to misdiagnosis and inadequate treatment. A way to solve this limitation is to use shared control; a gesture-assistance paradigm that combines the cognitive skills of the operator with the precision, stamina and repeatability of a robotic or haptic device. In this paper, we propose to assist the physician with a haptic device that holds the needle and generates mechanical guides during the phase of manual needle pre-positioning. In the latter, the physician has to place the tip of the needle on a planned entry point, with a pre-defined angle of incidence. From this pre-operative information and also from intra-operative measurements, we propose to generate haptic cues, known as virtual fixtures, to guide the physician towards the desired position and orientation of the needle. It takes the form of five haptic guides, each one implementing virtual fixtures. We conducted a user study where those guides were compared to the unassisted reference gesture. The most constraining guide, in terms of assisted degrees of freedom, was highlighted as the one that provides the best results in terms of performance and user experience.

Article dans revue scientifique avec comité de lecture.; PURPOSE: To evaluate three-dimensional (3D) digital subtraction angiography (DSA) as a supplement to two-dimensional (2D) DSA in the endovascular treatment (EVT) of intracranial aneurysms. MATERIALS AND METHODS: In 22 ruptured aneurysms, neck visualization, aneurysm shape, and EVT feasibility were analyzed at 2D DSA (anteroposterior, lateral, rotational views) and at maximum intensity projection (MIP) and surface shaded display (SSD) 3D DSA. The possibility of obtaining a working view for EVT at 3D DSA and the relevance of measurements in choosing the first coil also were assessed. RESULTS: Two-dimensional DSA images clearly depicted the aneurysm neck in four of 22 aneurysms; MIP images, in 10; and SSD images, in 21, but SSD led to overestimation of the neck size in one aneurysm. Aneurysm shape was precisely demonstrated in five of 22 aneurysms, EVT seemed to be nonfeasible at 2D DSA; however, SSD demonstrated feasibility and EVT was successfully performed. In one aneurysm, only SSD demonstrated the extension of the neck to a parent vessel, which was proved at surgery. Working views for EVT were deduced from 3D DSA findings in 20 of 21 aneurysms. The choice of the first coil was correct in 19 of 21 aneurysms. CONCLUSION: Three-dimensional DSA is valuable for evaluating the potential for EVT, finding a working view, and performing accurate measurements.

Colloque avec actes et comité de lecture.; Today, 3-D angiography volumes are routinely generated from rotational angiography sequences. In previous work \cite{Kerrien98}, we have studied the precision reached by registering such volumes with classical 2-D angiography images, inferring this matching only from the sensors of the angiography machine. The error led by such a registration can be described as a 3-D rigid motion composed of a large translation and a small rotation. This paper describes the strategy we followed to correct this error. The angiography image is compared in a two-step process to the Maximum Intensity Projection (MIP) of the angiography volume. The first step provides most of the translation by maximizing the cross-correlation. The second step recovers the residual rigid-body motion, thanks to a modified optical flow technique. A fine analysis of the equations encountered in both steps allows for a speed-up of the calculations. This algorithm was validated on 17 images of a phantom, and 5 patients. The residual error was determined by manually indicating points of interest and was found to be around 1 mm.