检索结果-内蒙古大学图书馆

Enhancing medical vision-language contrastive learning via inter-matching relation modelling

IEEE Transactions on Medical Imaging 2025年 PP卷 PP页

作者： Li, Mingjian Meng, Mingyuan Fulham, Michael Feng, David Dagan Bi, Lei Kim, Jinman University of Sydney Biomedical data analysis and visualization lab School of Computer Science NSW Australia Royal Prince Alfred Hospital Department of Molecular Imaging NSW Australia Shanghai Jiao Tong University Institute of Translational Medicine Shanghai China

Medical image representations can be learned through medical vision-language contrastive learning (mVLCL) where medical imaging reports are used as weak supervision through image-text alignment. These learned image representations can be transferred to and benefit various downstream medical vision tasks such as disease classification and segmentation. Recent mVLCL methods attempt to align image sub-regions and the report keywords as local-matchings. However, these methods aggregate all local-matchings via simple pooling operations while ignoring the inherent relations between them. These methods therefore fail to reason between local-matchings that are semantically related, e.g., local-matchings that correspond to the disease word and the location word (semantic-relations), and also fail to differentiate such clinically important local-matchings from others that correspond to less meaningful words, e.g., conjunction words (importance-relations). Hence, we propose a mVLCL method that models the inter-matching relations between local-matchings via a relation-enhanced contrastive learning framework (RECLF). In RECLF, we introduce a semantic-relation reasoning module (SRM) and an importance-relation reasoning module (IRM) to enable more fine-grained report supervision for image representation learning. We evaluated our method using six public benchmark datasets on four downstream tasks, including segmentation, zero-shot classification, linear classification, and cross-modal retrieval. Our results demonstrated the superiority of our RECLF over the state-of-the-art mVLCL methods with consistent improvements across single-modal and cross-modal tasks. These results suggest that our RECLF, by modelling the inter-matching relations, can learn improved medical image representations with better generalization capabilities. © 2025 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Discovering parametric clusters in social small-world graphs 05

Discovering parametric clusters in social small-world graphs

引用

20th Annual ACM Symposium on Applied Computing

作者： McPherson, Jonathan Ma, Kwan-Liu Ogawa, Michael Institute for Data Analysis and Visualization University of California Davis

ISBN: (纸本)9781581139648

We present a strategy for analyzing large, social small-world graphs, such as those formed by human networks. Our approach brings together ideas from a number of different research areas, including graph layout, graph clustering and partitioning, machine learning, and user interface design. It helps users explore the networks and develop insights concerning their members and structure that may be difficult or impossible to discover via traditional means, including existing graph visualization and/or statistical methods. Copyright 2005 ACM.

关键词： Graph theory

来源：评论

学校读者我要写书评

暂无评论

Massive data visualization: A survey

Massive data visualization: A survey

引用

Workshop on Mathematical Foundations of Scientific visualization, Computer Graphics, and Massive data Exploration, 2004

作者： Joy, Kenneth I. Institute for Data Analysis and Visualization University of California Davis United States

ISBN: (纸本)9783642543005

Today’s scientific and engineering problems require a different approach to address the massive data problems in organization, storage, transmission, visualization, exploration, and analysis. Visual techniques for data exploration are now common in many scientific, engineering, and business applications. However, the massive amount of data collected through simulation, collection and logging is inhibiting the use of conventional visualization methods. We need to discover new visualization methods that allow us to explore the massive multidimensional time-varying information streams and turn overwhelming tasks into opportunities for discovery and analysis. © 2009 Springer-Verlag Berlin Heidelberg.

关键词： data visualization

来源：评论

学校读者我要写书评

暂无评论

Temporal sample reuse for next event estimation and path guiding for real-time path tracing

Temporal sample reuse for next event estimation and path gui...

引用

31st Eurographics Symposium on Rendering, EGSR 2020 - Digital Library Only Track

作者： Dittebrandt, Addis Hanika, Johannes Dachsbacher, Carsten Karlsruhe Institute of Technology Institute for Visualization and Data Analysis Germany

ISBN: (纸本)9783038681175

Good importance sampling is crucial for real-time path tracing where only low sample budgets are possible. We present two efficient sampling techniques tailored for massively-parallel GPU path tracing which improve next event estimation (NEE) for rendering with many light sources and sampling of indirect illumination. As sampling densities need to vary spatially, we use an octree structure in world space and introduce algorithms to continuously adapt the partitioning and distribution of the sampling budget. Both sampling techniques exploit temporal coherence by reusing samples from the previous frame: For NEE we collect sampled, unoccluded light sources and show how to deduplicate, but also diffuse this information to efficiently sample light sources in the subsequent frame. For sampling indirect illumination, we present a compressed directional quadtree structure which is iteratively adapted towards high-energy directions using samples from the previous frame. The updates and rebuilding of all data structures takes about 1ms in our test scenes, and adds about 6ms at 1080p to the path tracing time compared to using state-of-the-art light hierarchies and BRDF sampling. We show that this additional effort reduces noise in terms of mean squared error by at least one order of magnitude in many situations. © 2020 The Author(s) Eurographics Proceedings © 2020 The Eurographics Association.

关键词： Light sources

来源：评论

学校读者我要写书评

暂无评论

Marching diamonds for unstructured meshes

Marching diamonds for unstructured meshes

引用

VIS 05: IEEE visualization 2005, Proceedings

作者： Anderson, John C. Bennett, Janine C. Joy, Kenneth I. Institute for Data Analysis and Visualization Computer Science Department University of California Davis

ISBN: (纸本)0780394623

We present a higher-order approach to the extraction of isosurfaces from unstructured meshes. Existing methods use linear interpolation along each mesh edge to find isosurface intersections. In contrast, our method determines intersections by performing barycentric interpolation over diamonds formed by the tetrahedra incident to each edge. Our method produces smoother, more accurate isosurfaces. Additionally, interpolating over diamonds, rather than linearly interpolating edge endpoints, enables us to identify up to two isosurface intersections per edge. This paper details how our new technique extracts isopoints, and presents a simple connection strategy for forming a triangle mesh isosurface. © 2005 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Isosurface extraction using fixed-sized buckets

Isosurface extraction using fixed-sized buckets

引用

7th Joint Eurographics - IEEE VGTC Symposium on visualization, VisSym 2005

作者： Waters, Kenneth W. Co, Christopher S. Joy, Kenneth I. Institute for Data Analysis and Visualization University of California Davis United States

ISBN: (纸本)3905673193

We present a simple and output optimal algorithm for accelerated isosurface extraction from volumetric data sets. Output optimal extraction algorithms perform an amount of work dominated by the size of the (output) isosurface rather than the size of the (input) data set. While several optimal methods have been proposed to accelerate isosurface extraction, these algorithms are relatively complicated to implement or require quantized values as input. Our method is based on a straightforward array data structure that only requires an auxiliary sorting routine for construction. The method works equally well for floating point data as it does for quantized data sets. We demonstrate how the data structure can exploit coherence between isosurfaces by performing searches incrementally. We show results for real application data validating the method’s optimality. © The Eurographics Association 2005.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficient dense reconstruction using geometry and image consistency constraints

Efficient dense reconstruction using geometry and image cons...

引用

IEEE Applied Imagery Pattern Recognition Workshop, AIPR 2015

作者： Shashkov, Mikhail M. Mak, Jason Recker, Shawn Nguyen, Connie Owens, John Joy, Kenneth I. Institute for Data Analysis and Visualization University of California - Davis DavisCA95616 United States

ISBN: (纸本)9781467395588

We introduce a method for creating very dense reconstructions of datasets, particularly turn-table varieties. The method takes in initial reconstructions (of any origin) and makes them denser by interpolating depth values in two-dimensional image space within a superpixel region and then optimizing the interpolated value via image consistency analysis across neighboring images in the dataset. One of the core assumptions in this method is that depth values per pixel will vary gradually along a gradient for a given object. As such, turntable datasets, such as the dinosaur dataset, are particularly easy for our method. Our method modernizes some existing techniques and parallelizes them on a GPU, which produces results faster than other densification methods. © 2015 IEEE.

关键词： Phonographs

来源：评论

学校读者我要写书评

暂无评论

A foveal inset for large display environments

A foveal inset for large display environments

引用

ACM International Conference on Virtual Reality Continuum and its Applications, VRCIA 2006

作者： Staadt, Oliver G. Ahlborn, Benjamin A. Kreylos, Oliver Hamann, Bernd Institute for Data Analysis and Visualization Department of Computer Science University of California Davis United States

ISBN: (纸本)1595933247

We introduce a system that adds a foveal inset to large-scale projection displays. The effective resolution of the foveal inset projection is higher than the original display resolution, allowing the user to see more details and finer features in large data sets. The foveal inset is generated by projecting a high-resolution image onto a mirror mounted on a pan - tilt unit that is controlled by the user with a laser pointer. Our implementation is based on Chromium and supports many OpenGL applications without modifications. We present experimental results using high-resolution image data from medical imaging and areal photography. © 2006 ACM.

关键词： Calibration

来源：评论

学校读者我要写书评

暂无评论

Interactive assembly of molecular structures

Interactive assembly of molecular structures

引用

5th Conference on Foundations of Nanoscience: Self-Assembled Architectures and Devices, FNANO 2008

作者： Kreylos, Oliver W. M. Keck Center for Active Visualization in Earth Sciences Institute for Data Analysis and Visualization University of California Davis United States

来源：评论

学校读者我要写书评

暂无评论

A new sampling scheme for slice based volume rendering

A new sampling scheme for slice based volume rendering

引用

作者： Krüger, J. Interactive Visualization and Data Analysis Group SCI Institute University Of Utah United States

ISBN: (纸本)9783905674217

In this paper we present a novel approach to generate proxy geometry for slice based volume rendering. The basic idea is derived from the behavior of a ray-caster and is a simple extension of the well known 2D object-aligned texture stack based technique. From this our novel scheme inherits the advantage that it enables hardware-based volume rendering for devices that do not support 3D textures. On these devices previous object-aligned 2D texture based approaches suffered from disturbing view angle dependent stack-switching artifacts which are avoided by our novel method. Our approach also shows benefits compared to the widely used view aligned slicing algorithm as it avoids jagged boundary artifacts and increases performance. © The Eurographics Association 2010.

关键词： Volume rendering

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：