EN FR
EN FR


Section: New Results

Image Guided Intervention

Classification of Surgical Process using Dynamic Time Warping

Participants : Pierre Jannin, Germain Forestier, Florent Lalys, Brivael Trelhu.

Toward the creation of new computer-assisted intervention systems, Surgical Process Models (SPMs) are more and more used as a tool for analyzing and assessing surgical interventions. SPMs represent Surgical Process (SPs) which are defined as symbolic structured descriptions of surgical interventions, using a pre-defined level of granularity and a dedicated terminology. In this context, an important challenge is the creation of new metrics for the comparison and the evaluation of SPs. Thus, correlations between these metrics and pre-operative data allow to classify surgeries and highlight specific information on the surgery itself and on the surgeon, such as its level of expertise. In this study, we explored the automatic classification of a set of SPs based on the Dynamic Time Warping (DTW) algorithm. DTW allows to compute a distance between two SPs that focuses on the different types of activities performed during the surgery and their sequencing, by minimizing the time differences. Indeed, it turns out to be a complementary approach to classical methods focusing only on the time and the number of activities differences. Experiments were carried out on 24 lumbar disc herniation surgeries to discriminate the level of expertise of surgeons according to prior classification of SPs. Supervised and unsupervised classification experiments have shown that this approach was able to automatically identify groups of surgeons according to their level of expertise (senior and junior), and opens many perspectives for the creation of new metrics for surgeries comparison and evaluation. This work was was performed in collaboration with Dr. Laurent Riffaud, and xas published in the International Journal of Biomedical Informatics [14] .

Surgical phases detection from microscope videos by machine learning

Participants : Pierre Jannin, Florent Lalys, Xavier Morandi.

Surgical process analysis and modeling is a recent and important topic aiming at introducing a new generation of computer-assisted surgical systems. Among all of the techniques already in use for extracting data from the Operating Room, the use of image videos allows automating the surgeons' assistance without altering the surgical routine. In collaboration with Carl Zeiss Medical Systems (Oberkochen, Germany), we proposed an application-dependent framework able to automatically extract the phases of the surgery only by using microscope videos as input data and that can be adaptable to different surgical specialties. First, four distinct types of classifiers based on image processing were implemented to extract visual cues from video frames. Each of these classifiers was related to one kind of visual cue: visual cues recognizable through color were detected with a color histogram approach, for shape-oriented visual cues we trained a Haar classifier, for texture-oriented visual cues we used a bag-of-word approach with SIFT descriptors, and for all other visual cues we used a classical image classification approach including a feature extraction, selection, and a supervised classification. The extraction of this semantic vector for each video frame then permitted to classify time series using either Hidden Markov Model or Dynamic Time Warping algorithms. The framework was validated on cataract surgeries, obtaining accuracies of 95%. This work was performed in collaboration with Laurent Riffaud and was published at the ORASIS and MICCAI conferences.

Surgical tools recognition and pupil segmentation for cataract surgery modeling

Participants : Pierre Jannin, Florent Lalys.

In the above project work performed through the MS intership of David Bouget, we focus on developping an application-dependant framework able to extract surgical phases from microscope videos. The aim of this study was to enhance results of this framework by adding new visual cues extraction modules. We studied two modules: one to segment the pupil and one to extract and recognize surgical tools. Validation studies, performed with cataract surgery videos, show an increase of the framework accuracy to detect eight surgical phases. This work has been accepted at the MMVR 2012 international conference.

Automatic computation of electrode trajectories for Deep Brain Stimulation: a hybrid symbolic and numerical approach

Participants : Pierre Jannin, Florent Lalys, Camille Maumet, Claire Haegelen.

The optimal electrode trajectory is needed to assist surgeons in planning Deep Brain Stimulation (DBS). We developed and tested a method for image-based trajectory planning. Rules governing the DBS surgical procedure were defined with geometric constraints. A formal geometric solver using multimodal brain images and a template built from 15 brain MRI scans were used to identify a space of possible solutions and select the optimal one. For validation, a retrospective study of 30 DBS electrode implantations from 18 patients was performed. A trajectory was computed in each case and compared with the trajectories of the electrodes that were actually implanted. Computed trajectories had an average difference of 6.45 degrees compared with reference trajectories and achieved a better overall score based on satisfaction of geometric constraints. Trajectories were computed in 2min for each case. We demonstrated that a rule-based solver using pre-operative MR brain images can automatically compute relevant and accurate patient-specific DBS electrode trajectories. This work was published in the International Journal of Computer Assisted Radiology and Surgery.

Analysis of electrodes' placement and deformation in deep brain stimulation from medical images

Participants : Pierre Jannin, Florent Lalys, Alexandre Abadie, Xavier Morandi, Claire Haegelen.

This work was performed during the intership of Maroua Mehri. Deep brain stimulation (DBS) is used to reduce the motor symptoms such as rigidity or bradykinesia, in patients with Parkinson's disease (PD). The Subthalamic Nucleus (STN) has emerged as prime target of DBS in idiopathic PD. However, DBS surgery is a difficult procedure requiring the exact positioning of electrodes in the pre-operative selected targets. This positioning is usually planned using patients' pre-operative images, along with digital atlases, assuming that electrode's trajectory is linear. However, it has been demonstrated that anatomical brain deformations induce electrode's deformations resulting in errors in the intra-operative targeting stage. In order to meet the need of a higher degree of placement accuracy and to help constructing a computer-aided-placement tool, we studied the electrodes' deformation in regards to patients' clinical data (i.e., sex, mean PD duration and brain atrophy index). Firstly, we presented an automatic algorithm for the segmentation of electrode's axis from post-operative CT images, which aims to localize the electrodes' stimulated contacts. To assess our method, we applied our algorithm on 25 patients who had undergone bilateral STNDBS. We found a placement error of 0.91+/-0.38 mm. Then, from the segmented axis, we quantitatively analyzed the electrodes' curvature and correlated it with patients' clinical data. We found a positive significant correlation between mean curvature index of the electrode and brain atrophy index for male patients and between mean curvature index of the electrode and mean PD duration for female patients. These results help understanding DBS electrode' deformations and would help ensuring better anticipation of electrodes' placement. This work has been accepted at the SPIE Medical Imaging 2012 conference.