arXiv subjects

Stephen Cross

Publications and source records attributed to Stephen Cross.

3 recordsLinked to original sources

arXiv2024

DD_RoTIR: Dual-Domain Image Registration via Image Translation and Hierarchical Feature-matching

Microscopy images obtained from multiple camera lenses or sensors in biological experiments provide a comprehensive understanding of objects from diverse perspectives. However, using multiple microscope setups increases the risk of misalignment of identical target features across different modalities, making multimodal image registration crucial. In this work, we build upon previous successes in biological image translation (XAcGAN) and mono-modal image registration (RoTIR) to develop a deep learning model, Dual-Domain RoTIR (DD_RoTIR), specifically designed to address these challenges. While GAN-based translation models are often considered inadequate for multimodal image registration, we enhance registration accuracy by employing a feature-matching algorithm based on Transformers and rotation equivariant networks. Additionally, hierarchical feature matching is utilized to tackle the complexities of multimodal image registration. Our results demonstrate that the DD_RoTIR model exhibits strong applicability and robustness across multiple microscopy image datasets.

eess.IV↗

arXiv2024

RoTIR: Rotation-Equivariant Network and Transformers for Fish Scale Image Registration

Image registration is an essential process for aligning features of interest from multiple images. With the recent development of deep learning techniques, image registration approaches have advanced to a new level. In this work, we present 'Rotation-Equivariant network and Transformers for Image Registration' (RoTIR), a deep-learning-based method for the alignment of fish scale images captured by light microscopy. This approach overcomes the challenge of arbitrary rotation and translation detection, as well as the absence of ground truth data. We employ feature-matching approaches based on Transformers and general E(2)-equivariant steerable CNNs for model creation. Besides, an artificial training dataset is employed for semi-supervised learning. Results show RoTIR successfully achieves the goal of fish scale image registration.

eess.IV↗

arXiv2022

Fluctuations of cell geometry and their non-equilibrium thermodynamics in living epithelial tissue

We measure different contributions to entropy production in a living functional epithelial tissue. We do this by extracting the functional dynamics of development while at the same time quantifying fluctuations. Using the translucent Drosophila melanogaster pupal epithelium as an ideal tissue for high resolution live imaging [1], we measure the entropy associated with the stochastic geometry of cells in the epithelium. This is done using a detailed analysis of the dynamics of the shape and orientation of individual cells which enables separation of local and global aspects of the tissue behaviour. We find intriguingly that we can observe irreversible dynamics in the cell geometries but without a change in the entropy associated with those degrees of freedom, showing that there is a flow of energy into those degrees of freedom. Hence the living system is controlling how the entropy is being produced and partitioned into its different parts.

cond-mat.soft↗