HYHuimin Yan

Huimin Yan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Huimin Yan has a total of 13 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Local Alignment for Medical Vision-Language Pre-Training

    by Huimin Yan, Xian Yang, Liang Bai, Jiye Liang

    Establishing local semantic correspondences between medical images and their corresponding reports is crucial for effective medical vision-language pre-training. However, existing methods encounter two major challenges: (1) lesion regions in radiological images are often small, blurry, or lack clear boundaries, complicating accurate localization; and (2) medical reports typically contain redundant or non-diagnostic words, hindering precise semantic alignment. To overcome these issues, we propose MedAligner, a specialized local alignment network for medical vision-language pre-training. MedAligner employs dual encoders to extract both global and local representations and uses global contrastive learning to maintain coarse semantic consistency. To enhance local alignment, we introduce a Word-Region Alignment, which generates a learnable word-pixel similarity matrix that is sparsified to identify salient lesion regions accurately. Additionally, our Diagnostic Term Filtering dynamically samples high-importance diagnostic terms from reports, aligning them with identified lesion areas via a local contrastive loss. Importantly, we adopt a progressive training strategy that gradually refines both the input text and semantic alignment. This is achieved by reconstructing concise diagnostic reports and progressively updating word-pixel similarity, generating increasingly accurate image-text pairs. Extensive experiments demonstrate that MedAligner significantly surpasses existing approaches on tasks such as phrase grounding, image-text retrieval, and zero-shot classification, setting new benchmarks in medical vision-language pre-training.


  • Multi-Grained Vision-and-Language Model for Medical Image and Text Alignment

    by Huimin Yan, Xian Yang, Liang Bai, Jiamin Li, Jiye Liang

    The increasing interest in learning from paired medical images and textual reports highlights the need for methods that can achieve multi-grained alignment between these two modalities. However, most existing approaches overlook finegrained semantic alignment, which can constrain the quality of the generated representations. To tackle this problem, we propose the Multi-Grained Vision-and-Language Alignment (MGVLA) model, which effectively leverages multi-grained correspondences between medical images and texts at different levels, including disease, instance, and token levels. For disease-level alignment, our approach adopts the concept of contrastive learning and uses medical terminologies detected from textual reports as soft labels to guide the alignment process. At the instance level, we propose a strategy for sampling hard negatives, where images and texts with the same disease type but differing in details such as disease locations and severity are considered as hard negatives. This strategy helps our approach to better distinguish between positive and negative image-text pairs, ultimately enhancing the quality of our learned representations. For token-level alignment, we employ a masking and recovery technique to achieve finegrained semantic alignment between patches and sub-words. This approach effectively aligns the different levels of granularity between the image and language modalities. To assess the efficacy of our MGVLA model, we conduct comprehensive experiments on the image-text retrieval and phrase grounding tasks.


  • Combination of dot-matrix lighting and floodlighting for multipath interference suppression in ToF imaging

    by Xiaoyue Liu, Jiayi Huo, Huimin Yan

    Time-of-flight (ToF) cameras obtain the depth information of the whole scene simultaneously by floodlighting it. In the presence of mutual reflection between the targets, the measurement would suffer from the multipath interference (MPI), because the signal received by the sensor is a mixture of direct component and multipath (or global) component. MPI would lead to significant measurement errors. In this paper is introduced a method of separating the direct and global components by once dot-matrix lighting and twice floodlighting with different frequencies. With dot-matrix lighting, the depth information at the dot matrix position is almost only determined by the direct component. The phase value of the direct component is readily calculated. The global component at the dot position with floodlighting is separated by minimizing the separation error through solving the nonlinear least square problem. Then the global component of the whole scene can be obtained by two-dimensional interpolation from data at dot matrix position. As a result, the direct component can be calculated by subtracting the interpolation result from the floodlighting signal, and the depth were calculated only from direct component. Semi-physical experiments were made for three kinds of scenes, blank corner with uniform reflectivity, the corner with different reflectivity areas and the corner with an object placed in front of it. The results demonstrate that the MPI has been significantly suppressed in all scenes. Especially in the area with strong MPI in the first two kinds of scenes, the measurement errors can decrease to about 10%~20%.

  • The analysis on fluorescence signals of the multi-channel DNA analyzer

    by Yi Fang, Liqiang Wang, Jin Xu, Pengpeng Wang, Huimin Yan

    The original data generated by multi-channel DNA analyzer is a mixture in time domain of data of multiple channels and that of multiple spectra. Thus, it is necessary to preprocess the original data before the DNA sequence can be identified. In this paper, the processing procedure is proposed, including choosing data segment, deburring, generating four-color information, choosing channels, data filtering and removing baseline. The experiment results state that useful information can be well separated from original data. The noise is greatly reduced and signal peaks are clearer. And that will provide the basis for determining the type and position of bases.