Training Bayesian networks for image segmentation

We are concerned with the problem of image segmentation in which each pixel is assigned to one of a predefined finite number of classes. In Bayesian image analysis, this requires fusing together local predictions for the class labels with a prior model of segmentations. Markov Random Fields (MRFs) have been used to incorporate some of this prior knowledge, but this not entirely satisfactory as inference in MRFs is NP-hard. The multiscale quadtree model of Bouman and Shapiro (1994) is an attractive alternative, as this is a tree-structured belief network in which inference can be carried out in linear time (Pearl 1988). It is an hierarchical model where the bottom-level nodes are pixels, and higher levels correspond to downsampled versions of the image. The conditional-probability tables (CPTs) in the belief network encode the knowledge of how the levels interact. In this paper we discuss two methods of learning the CPTs given training data, using (a) maximum likelihood and the EM algorithm and (b) emphconditional maximum likelihood (CML). Segmentations obtained using networks trained by CML show a statistically-significant improvement in performance on synthetic images. We also demonstrate the methods on a real-world outdoor-scene segmentation task.

Novel bioactive scaffolds incorporating nanogels as potential drug eluting devices

Big advances are being achieved in the design of new implantable devices with enhanced properties. For example, synthetic porous three-dimensional structures can mimic the architecture of the tissues, and serve as templates for cell seeding. In addition, polymeric nanoparticles are able to provide a programmable and sustained local delivery of different types of biomolecules. In this study novel alternative scaffolds with controlled bioactive properties and architectures are presented. Two complementary approaches are described. Firstly, scaffolds with nanogels as active controlled release devices incorporated inside the three-dimensional structure are obtained using the thermally induced phase separation (TIPS) method. Secondly, a novel coating method using the spraying technique to load these nanometric crosslinked hydrogels on the surface of two-dimensional (2D) and three-dimensional (3D) biodegradable scaffolds is described. The scanning electron microscopy (SEM) images show the distribution of the nanogels on the surface of different substrates and also inside the porous structure of poly-a-hydroxy ester derivative foams. Both of them are compared in terms of manufacturability, dispersion and other processing variables.

Reciprocating power generation in a chemically driven synthetic muscle

A scalable synthetic muscle has been constructed that transducts nanoscale molecular shape changes into macroscopic motion. The working material, which deforms affinely in response to a pH stimulus, is a self-assembled block copolymer comprising nanoscopic hydrophobic domains in a weak polyacid matrix. A device has been assembled where the muscle does work on a cantilever and the force generated has been measured. When coupled to a chemical oscillator this provides a free running chemical motor that generates a peak power of 20 mW kg 1 by the serial addition of 10 nm shape changes that scales over 5 orders of magnitude. It is the nanostructured nature of the gel that gives rise to the affine deformation and results in a robust working material for the construction of scalable muscle devices.

Surface waves in mesh synthetic photonic lattices

Eigenmodes and dispersion curves in different configurations of synthetic photonic lattices are studied numerically. Eigenmodes localized on borders between areas with different optical potential are found. Stability of these eigenmodes against potential disturbances of different type is studied.