arXiv subjects

Yin Guo

Publications and source records attributed to Yin Guo.

4 recordsLinked to original sources

arXiv2025

Unified and Semantically Grounded Domain Adaptation for Medical Image Segmentation

Most prior unsupervised domain adaptation approaches for medical image segmentation are narrowly tailored to either the source-accessible setting, where adaptation is guided by source-target alignment, or the source-free setting, which typically resorts to implicit adaptation mechanisms such as pseudo-labeling and network distillation. This substantial divergence in methodological designs between the two settings reveals an inherent flaw: the lack of an explicit, structured construction of anatomical knowledge that naturally generalizes across domains and settings. To bridge this longstanding divide, we introduce a unified, semantically grounded framework that supports both source-accessible and source-free adaptation. Fundamentally distinct from all prior works, our framework's adaptability emerges naturally as a direct consequence of the model architecture, without relying on explicit cross-domain alignment strategies. Specifically, our model learns a domain-agnostic probabilistic manifold as a global space of anatomical regularities, mirroring how humans establish visual understanding. Thus, the structural content in each image can be interpreted as a canonical anatomy retrieved from the manifold and a spatial transformation capturing individual-specific geometry. This disentangled, interpretable formulation enables semantically meaningful prediction with intrinsic adaptability. Extensive experiments on challenging cardiac and abdominal datasets show that our framework achieves state-of-the-art results in both settings, with source-free performance closely approaching its source-accessible counterpart, a level of consistency rarely observed in prior works. The results provide a principled foundation for anatomically informed, interpretable, and unified solutions for domain adaptation in medical imaging. The code is available at https://github.com/wxdrizzle/remind

cs.CV↗

arXiv2025

RemInD: Remembering Anatomical Variations for Interpretable Domain Adaptive Medical Image Segmentation

This work presents a novel Bayesian framework for unsupervised domain adaptation (UDA) in medical image segmentation. While prior works have explored this clinically significant task using various strategies of domain alignment, they often lack an explicit and explainable mechanism to ensure that target image features capture meaningful structural information. Besides, these methods are prone to the curse of dimensionality, inevitably leading to challenges in interpretability and computational efficiency. To address these limitations, we propose RemInD, a framework inspired by human adaptation. RemInD learns a domain-agnostic latent manifold, characterized by several anchors, to memorize anatomical variations. By mapping images onto this manifold as weighted anchor averages, our approach ensures realistic and reliable predictions. This design mirrors how humans develop representative components to understand images and then retrieve component combinations from memory to guide segmentation. Notably, model prediction is determined by two explainable factors: a low-dimensional anchor weight vector, and a spatial deformation. This design facilitates computationally efficient and geometry-adherent adaptation by aligning weight vectors between domains on a probability simplex. Experiments on two public datasets, encompassing cardiac and abdominal imaging, demonstrate the superiority of RemInD, which achieves state-of-the-art performance using a single alignment approach, outperforming existing methods that often rely on multiple complex alignment strategies.

cs.CV↗

arXiv2019

Gas modulation refractometry (GAMOR) - On its ability to eliminate the influence of drifts

Gas modulation refractometry (GAMOR) is a technique based on a dual-Fabry-Perot (FP) cavity (DFPC) for assessment of gas refractivity, density, and pressure that can alleviate significant limitations of conventional refractometry systems, predominantly those related to drifts. Repeated assessments of the beat frequency when the measurement cavity is evacuated provide conditions under which the methodology is immune to the linear parts of the drifts in the system, both those from length changes of the cavities and those from gas leaks and outgassing. This implies that the technique is solely influenced by the non-linear parts of the drifts. This work provides a description of the principle behind the GAMOR methodology and explicates the background to its unique property. Based on simple models of the drifts of the temperature in the cavity spacer and the residual gas in the reference cavity, this work predicts that a GAMOR system, when used for assessment of refractivity, can sustain significant temperature drifts and leakage rates without being affected by noticeable errors or uncertainties. The cavity spacer can be exposed to temperature fluctuations of 100 mK over 103 s, and the reference cavity can have a leakage that fills it up with gas on a timescale of days, without providing errors or uncertainties in the assessment of refractivity that are 3 x 10^(-12), which, for N2, corresponds to 0.01 ppm (parts per million) of the value under atmospheric pressure conditions, and thereby 1 mPa. Since well-designed systems often have temperature fluctuations and leakage rates that are smaller than these, it is concluded that there will, in practice, not be any appreciable influence from cavity length drifts, gas leaks, and outgassing in the GAMOR methodology.

physics.ins-det↗

arXiv2015

Uncovering low-dimensional, miR-based signatures of acute myeloid and lymphoblastic leukemias with a machine-learning-driven network approach

Complex phenotypic differences among different acute leukemias cannot be fully captured by analyzing the expression levels of one single molecule, such as a miR, at a time, but requires systematic analysis of large sets of miRs. While a popular approach for analysis of such datasets is principal component analysis (PCA), this method is not designed to optimally discriminate different phenotypes. Moreover, PCA and other low-dimensional representation methods yield linear or non-linear combinations of all measured miRs. Global human miR expression was measured in AML, B-ALL, and T-ALL cell lines and patient RNA samples. By systematically applying support vector machines to all measured miRs taken in dyad and triad groups, we built miR networks using cell line data and validated our findings with primary patient samples. All the coordinately transcribed members of the miR-23a cluster (which includes also miR-24 and miR-27a), known to function as tumor suppressors of acute leukemias, appeared in the AML, B-ALL and T-ALL centric networks. Subsequent qRT-PCR analysis showed that the most connected miR in the B-ALL-centric network, miR-708, is highly and specifically expressed in B-ALLs, suggesting that miR-708 might serve as a biomarker for B-ALL. This approach is systematic, quantitative, scalable, and unbiased. Rather than a single signature, our approach yields a network of signatures reflecting the redundant nature of biological signaling pathways. The network representation allows for visual analysis of all signatures by an expert and for future integration of additional information. Furthermore, each signature involves only small sets of miRs, such as dyads and triads, which are well suited for in depth validation through laboratory experiments such as loss- and gain-of-function assays designed to drive changes in leukemia cell survival, proliferation and differentiation.

q-bio.QM↗