Semantic Segmentation of Earth Observation Data Using Multimodal and Multi-scale Deep Networks (1609.06846v1)

Abstract: This work investigates the use of deep fully convolutional neural networks (DFCNN) for pixel-wise scene labeling of Earth Observation images. Especially, we train a variant of the SegNet architecture on remote sensing data over an urban area and study different strategies for performing accurate semantic segmentation. Our contributions are the following: 1) we transfer efficiently a DFCNN from generic everyday images to remote sensing images; 2) we introduce a multi-kernel convolutional layer for fast aggregation of predictions at multiple scales; 3) we perform data fusion from heterogeneous sensors (optical and laser) using residual correction. Our framework improves state-of-the-art accuracy on the ISPRS Vaihingen 2D Semantic Labeling dataset.

• The paper establishes that transmembrane proteins increase intermonolayer friction, thereby reducing protein mobility and altering membrane dynamics.
• It demonstrates that protein asymmetry couples with membrane curvature, making protein density the slowest relaxing variable under sufficient tension.
• The study reveals non-self-similar diffusion profiles with a decrease in collective diffusion at small wavelengths due to friction and curvature effects.

Hydrodynamics of Bilayer Membranes with Diffusing Transmembrane Proteins

Andrew Callan-Jones, Marc Durand, and Jean-Baptiste Fournier present a detailed exploration of the hydrodynamic behavior of lipid bilayers containing transmembrane proteins, with a focus on the relaxation dynamics arising from the interplay of membrane shape, mass densities of bilayer leaflets, and the concentration of diffusing proteins. This paper operates within the framework of Onsager's variational principle to derive a comprehensive set of equations that describe the dynamics of such complex systems.

Key Insights and Findings

1. Intermonolayer Friction and Protein Mobility: The authors establish that the presence of transmembrane proteins in the lipid bilayer increases the intermonolayer friction coefficient. This enhancement is inversely proportional to the protein's mobility, which implies that slower-moving proteins generate more substantial resistance to the sliding motion between the monolayers.
2. Asymmetry and Membrane Curvature: Asymmetric proteins exhibit coupling with the membrane curvature and monolayer density differences. For membrane tensions exceeding $\sigma > 10^{-8}$ N/m, the protein density emerges as the slowest relaxing variable, largely due to the coupling with curvature at short wavelengths.
3. Diffusion Dynamics: A significant result is the non-self-similar diffusion profile for a concentrated protein patch. This occurs due to the wavevector-dependent effective diffusion coefficient, illustrating a complex behavior that deviates from classical diffusion models.
4. Collective Diffusion Coefficient: The collective diffusion coefficient $D_{\mathrm{eff}}(q)$ is found to decrease at smaller wavelengths due to intermonolayer friction and curvature coupling effects. This behavior presents a nuanced understanding of protein dynamics on curved, lipid surfaces.

Implications and Future Directions

The implications of this paper span both the practical and theoretical aspects of biophysics and cellular biology. Practically, understanding membrane dynamics can inform the development of bio-compatible materials and drug delivery mechanisms where membrane permeability and protein transport play critical roles. Theoretically, the paper enriches the fundamental understanding of membrane physics, particularly in how proteins, curvature, and friction interplay to influence membrane behavior.

Future Research Directions:

• Experiments at High Protein Concentrations: The anomalous diffusion behaviors predicted at high protein concentrations warrant experimental validation. Techniques such as fluorescence recovery after photobleaching (FRAP) and single-particle tracking could elucidate these dynamics.
• Multi-Wavelength Studies: Further investigations could explore how different wavelength-dependent behaviors correlate with particular biological processes, such as endocytosis or exocytosis, where membrane deformation plays a crucial role.
• Extensions to Multi-Component Membranes: Extending this framework to multi-component systems, including various lipid species and membrane-bound proteins, may provide insights into more realistic cellular environments.

The paper's meticulous derivation of equations and exploration of multi-phase flow dynamics offers a robust platform for further exploration into the hydrodynamics of complex biological membranes. Researchers engaged in cellular biophysics and synthetic biology may find valuable insights in this well-structured treatment of membrane dynamics involving diffusing proteins.