Search

In order to better understand and describe surgical procedures by surgical process models, the field of workflow segmentation has recently emerged. It aims to recognize high-level surgical tasks in the Operating Room, with the help of sensors or human-based systems. Our approach focused on the automatic recognition of surgical phases by microscope images analysis. We used a hybrid method that combined Support Vector Machine and discrete Hidden Markov Model. We first performed features extraction and selection on surgical microscope frames to create an image database. SVMs were trained to extract surgical scene information, and then outputs were used as observations for training a discrete HMM. Our framework was tested on pituitary surgery, where six phases were identified by neurosurgeons. Cross-validation studies permitted to find a percentage of detected phases of 93% that will allow the use of the system in clinical applications such as post-operative videos indexation. [ABSTRACT FROM AUTHOR]

The segmentation of the surgical workflow might be helpful for providing context-sensitive user interfaces, or generating automatic report. Our approach focused on the automatic recognition of surgical phases by microscope image classification. Our workflow, including images features extraction, image database labelisation, Principal Component Analysis (PCA) transformation and 10-fold cross-validation studies was performed on a specific type of neurosurgical intervention, the pituitary surgery. Six phases were defined by an expert for this type of intervention. We thus assessed machine learning algorithms along with the data dimension reduction. We finally kept 40 features from the PCA and found a best correct classification rate of the surgical phases of 82% with the multiclass Support Vector Machine. [ABSTRACT FROM AUTHOR]

We propose a method for the creation of an anatomically and mechanically realistic brain phantom from polyvinyl alcohol cryogel (PVA-C) for validation of image processing methods for segmentation, reconstruction, registration, and denoising. PVA-C is material widely used in medical imaging phantoms for its mechanical similarities to soft tissues. The phantom was cast in a mold designed using the left hemiphere of the Colin27 brain dataset [1] and contains deep sulci, a complete insular region, and an anatomically accurate left ventricle. Marker spheres and inflatable catheters were also implanted to enable good registration and simulate tissue deformation, respectively. The phantom was designed for triple modality imaging, giving good contrast images in computed tomography, ultrasound, and magnetic resonance imaging. Multimodal data acquired from this phantom are made freely available to the image processing community (http://pvabrain.inria.fr) and will aid in the validation and further development of medical image processing techniques. [ABSTRACT FROM AUTHOR]

This paper presents an interactive system for the annotation of brain anatomical structures in Magnetic Resonance Images. The system is based on hybrid knowledge and techniques. First, it exploits both numerical knowledge from atlases and symbolic knowledge from a rule-extended ontology represented in OWL, the Web ontology language, and combines them with graphical data about cortical sulci, automatically extracted from the images. Second, the annotations of the parts of gyri and of sulci located in a region of interest are obtained with different reasoning techniques: Constraint Satisfaction Solving and Description Logics techniques. Preliminary experiments have been achieved on normal and also pathological data. The results obtained so far are very promising. [ABSTRACT FROM AUTHOR]

Ultrasound (US) imaging is often proposed as an interoperative imaging modality. This use nearly always requires that the collected data be registered to preoperative data of another modality. Existing intensity-based registration approaches all begin by reconstructing a 3D US volume from the collected 2D slices. We propose to directly register the set of 2D slices to the preoperative images. We argue this has a number of advantages, including the omission of the potentially complex reconstruction step, greater adaptability of the similarity measures, and easier parallelization. We describe a system for performing this task and present results on phantom data that show that our slice based method consistently outperforms a reconstruction based method in both speed and accuracy. [ABSTRACT FROM AUTHOR]