"Integration of Medical Imaging in Physicians' Reporting Workflow" by Sharmili Roy
Medical doctors in virtually all fields of medicine now rely on imaging technology to make diagnoses and clinical decisions for treatment. The workflow generally involves an ordering physician who requests an imaging study to be performed on a patient. A radiologist interprets the exam using a dedicated image workstation which allows the radiologist to make visual annotations on the images to denote regions of interest, or to make quantitative measurements, or simply to select key images in the study that are of clinical importance. The radiologist also prepares a textual report that refers back to the visual information prepared during the exam. Surprisingly, however, the ordering physician often relies solely on the text-based report. This lack of access to the image-based annotations is due to variety of reasons ranging from software incompatibilities to cumbersome workflows.
The aim of this work is to address the limitations in current medical imaging and reporting workflow, in particular the out-dated reliance of ordering physicians on text only reports. Our focus is to develop a software framework that allows radiologist to produce visual summaries to augment or integrate into their text reports in a format that is not only easily accessible, but is concise and visually informative to the ordering physician.
"2D-3D Neural Stem Cell Neurosphere Modelisation" by Stéphane Rigaud
The study of stem cells is one of the current most important biomedical research field. Understanding their development could allow multiple applications in regenerative medicine. For this purpose, we need automated methods for the segmentation and the modeling of neural stem cell development process into a neurosphere colony from phase contrast mi- croscopy. We use such methods to extract relevant structural and textural features like cell division dynamism and cell behavior patterns for biological interpretation. The combination of phase contrast imaging, high fragility and complex evolution of neural stem cells pose many challenges in image processing and image analysis. We introduces an on-line analysis method for the modeling of neurosphere evolution during the first three days of their development. From the corresponding time-lapse sequences, we extract information from the neurosphere using a combination of fast level set and curve detection for segmenting the cells. Then, based on prior biological knowledge, we generate possible and optimal 3-dimensional configuration using registration and evolutionary optimisation algorithm.
"Incorporation of Prior-Knowledge into Learning with SVMs" by Antoine Veillard
The Knowledge-Enhanced RBF (KE-RBF) framework is a set of original kernel methods for the incorporation of prior-knowledge into SVMs. Based on transformations of the RBF kernel, it comprises 3 new types of kernels allowing for the incorporation of a wide array of commonly available problem-specific prior-knowledge. An extensive and thorough empirical validation based on 5 different applications of these new kernels shows that the KE-RBF framework is highly usable in practice, has the potential to largely improve learning performances over the RBF kernel, and sharply reduces the requirements in training data. This framework paves the way for several interesting new possibilities with SVMs such as learning with very small or strongly biased datasets.