Biomedical image segmentation plays a pivotal role in medical imaging, facilitating precise identification and delineation of anatomical structures and abnormalities. This review explores the application of the Denoising Diffusion Probabilistic Model (DDPM) in the realm of biomedical image segmentation. DDPM, a probabilistic generative model, has demonstrated promise in capturing complex data distributions and reducing noise in various domains. In this context, the review provides an in-depth examination of the present status, obstacles, and future prospects in the application of biomedical image segmentation techniques. It addresses challenges associated with the uncertainty and variability in imaging data analyzing commonalities based on probabilistic methods. The paper concludes with insights into the potential impact of DDPM on advancing medical imaging techniques and fostering reliable segmentation results in clinical applications. This comprehensive review aims to provide researchers, practitioners, and healthcare professionals with a nuanced understanding of the current state, challenges, and future prospects of utilizing DDPM in the context of biomedical image segmentation.

Medical image segmentation remains particularly challenging for complex and low‐contrast anatomical structures, especially in brain MRI glioma segmentation. Gliomas appear with extensive heterogeneity in appearance and location on brain MR images, making robust tumour segmentation extremely challenging and leads to highly variable even in manual segmentation. U‐Net has become the de facto standard in medical image segmentation tasks with great success. Previous researches have proposed various U‐Net‐based 2D Convolutional Neural Networks (2D‐CNN) and their 3D variants, called 3D‐CNN‐based architectures, for capturing contextual information. However, U‐Net often has limitations in explicitly modelling long‐term dependencies due to the inherent locality of convolution operations. Inspired by the recent success of natural language processing transformers in long‐range sequence learning, a multi‐view 2D U‐Nets with transformer (TransMVU) method is proposed, which combines the advantages of transformer and 2D U‐Net. On the one hand, the transformer encodes the tokenized image patches in the CNN feature map into an input sequence for extracting global context for global feature modelling. On the other hand, multi‐view 2D U‐Nets can provide accurate segmentation with fewer parameters than 3D networks. Experimental results on the BraTS20 dataset demonstrate that our model outperforms state‐of‐the‐art 2D models and classic 3D model.

Electrostatic MEMS micromirrors provide a compact and low-power beam-steering solution for miniaturized laser communication terminals. However, when they are used for quasi-static beam pointing rather than resonant scanning, the nonlinear voltage–angle relationship, bidirectional actuation asymmetry, and terminal-level installation errors can significantly degrade pointing accuracy. In this paper, a nonlinear modeling and differential-voltage control method is investigated for a two-axis electrostatic MEMS micromirror used in a miniaturized laser communication terminal. The device under test is a bonded aluminum MEMS micromirror with a 5.0 mm aperture. Static and dynamic characterization results show that the micromirror achieves maximum mechanical deflection angles of 5.215° and 5.161° along the X and Y axes, respectively, with resonant frequencies of 317 Hz and 319 Hz. To improve the accuracy of quasi-static pointing, the differential-voltage actuation principle is analyzed, and a nonlinear voltage–angle model is established based on measured deflection data. Compared with a first-order linear model, the cubic nonlinear model reduces the root-mean-square fitting error from 0.142° to 0.0127° for the X axis and from 0.132° to 0.0109° for the Y axis. Furthermore, a terminal-level calibration architecture based on a quadrant detector is introduced to map the MEMS angular deflection to the received spot position. The proposed modeling and calibration approach provides an actuator-level basis for accurate beam pointing and closed-loop acquisition in miniaturized laser communication terminals.

A reusable angular disturbance input is essential for the design and verification of precision pointing systems in space laser communication. This study develops a configurable telemetry-based framework for reconstructing and modeling elevation and azimuth disturbances at a calibrated terminal interface. On-orbit centroid residuals and actual actuator feedback are combined, and a stable bivariate vector autoregressive (VAR) model preserves spectral structure and cross-axis dependence. Four operating conditions from one intersatellite experiment are represented by a common structure with condition-specific coefficients. Multiresolution analysis identifies sub-hertz behavior, components near 3.3 and 36 Hz, and measured content up to 450 Hz. A chronological validation over 1–100 Hz yields trace-spectrum errors of 0.926–1.975 dB and absolute band-limited root-mean-square (RMS) errors within 3.39%; an additional complete-segment holdout results in RMS errors within 3.68%. Across 54 numerical terminal configurations per condition, model-based pointing RMS agrees with time-domain telemetry replay within 4.75%. A numerical frequency-selective input-attenuation scenario reduces nominal pointing RMS by up to 39.60%. The transferable elements are the reconstruction procedure, spectral-matrix interface, and terminal-response mapping; new conditions require calibrated parameter identification. Digitally generated replay sequences specify a proposed optical ground-verification interface for subsequent terminal development.