基于模糊数的乳腺肿瘤超声图像边缘快速提取方法

乳腺肿瘤边缘的准确提取在临床上对肿瘤良恶性的判别有重要的意义。本文利用三角模糊数的概念，采用重叠式窗口从图像中得到与不同隶属度对应的模糊数，从而建立以步进方格（marching square）为基本单元的模糊数平面；通过区间阈值得到步进方格上的映射区间，根据步进方格算法将对应映射区间着色绘制出肿瘤的边界。分别对恶性和良性肿瘤超声图像进行边缘提取。结果显示，本文方法相比一般提取边缘的算法具有快速准确提取乳腺肿瘤边缘的特点。实验证明本方法可以有效用于乳腺肿瘤超声图像边缘提取。     

乳腺肿瘤超声图像特征参数量化研究进展

乳腺肿瘤超声图像的特征量化分析对判别肿瘤的良、恶性具有重要价值。本文总结了良性和恶性乳腺肿瘤在超声图像上的特点，将乳腺良性肿瘤和恶性肿瘤鉴别特征在形状、边缘、边界、朝向、回声特点几个方面的量化方法和量化参数进行了较为全面的梳理，并对量化特征与肿瘤良、恶性之间的关系进行了分析。     

用于乳腺肿瘤分级诊断超声图像增强的算法

为提高乳腺肿瘤分级诊断的能力,提出一种基于超声信号用于乳腺肿瘤分级诊断的图像增强算法。通过分析良性和不同恶性程度肿瘤的超声图像的特征差异,提出了一种将灰度的动态变换方法和利用局部标准差及熵特征相结合的办法,对图像对比度进行增强处理,增强了乳腺超声图像的细节,提高了图像质量。该算法可对良性、恶性肿瘤等不同超声图像进行增强处理,使得图像之间差异更加明显,为临床医生分级诊断提供更加清晰的图像,具有一定的实际应用价值。     

基于超声图像的乳腺钙化点自动检测技术

当前乳腺钙化点检测方法多基于X光片,难以应用于超声图像,本研究提出基于超声图像的乳腺钙化点自动检测技术,首先将乳腺超声图像中的肿瘤区域通过勾画模板提取出来,基于简单线性迭代聚类算法进行超像素分割;然后提取表征各超像素的特征量来计算显著性图,基于钙化区域显著性进行粗钙化点分割;最终对分割后的粗钙化点进行形态学检测,达到对超声图像中的细钙化点自动检测。该方法取得了较好的分割效果,具有较强的鲁棒性,为形成具有普适性的肿瘤自动诊断方案奠定了研究基础。     

基于Bandlet的压缩感知算法及其在智能乳腺全容积超声系统中的应用

目的在智能乳腺全容积超声系统中需扫描很多个切面同时进行成像和保存,数据量庞大。为此,本文提出基于Bandlet变换的压缩感知方法并应用于该系统,以降低存储和传输的数据量。方法首先利用超声图像的Bandlet变换域能够根据图像的"几何正则性"来自适应改变得到稀疏表示的特点,将所得图像进行Bandlet变换。然后选择与Bandlet基矩阵不相干的随机测量来降低图像压缩的数据量,之后利用匹配追踪算法由压缩数据重建超声图像。最后以智能乳腺全容积超声系统的图像数据为例进行压缩效率和重建有效性的验证。结果压缩后的数据大小为原数据的30%,降低了传输和存储的数据量,同时可得到高质量的重建图像。结论基于Bandlet的压缩感知算法可降低智能乳腺全容积超声系统图像的传输带宽和数据量,并保证了图像重建的质量,适用于智能乳腺全容积超声系统。     

以肺痨病为例探讨中医证候本体的构建

本体是语义web的关键技术。中医证候本体的构建对于中医理论探讨、数据整理、知识共享是一件极具现实和学术意义的工作。本文探讨了中医哲学对中医证候本体的影响，阐明了中医证候本体与中医顶层本体的关系；并以protege-owl plugin为工具．对中医证候本体的构建思路和过程进行简单的介绍。     

一种基于语义模型的乳腺钙化病灶标注方法

为了实现对乳腺X线影像的医学语义标注,提出一种利用贝叶斯网络(BN)的多层乳腺影像钙化点语义建模方法。该方法首先用支持向量机(SVM)得到从图像底层视觉特征到中层特征语义的映射,然后再利用BN融合特征语义,最终提取出高层病症语义即恶性程度的概率表达,完成语义模型。将模型应用于乳腺图像的语义标注,本实验选用142幅图像作为训练集,50幅图像作为测试集,结果表明,样本标注诊断语义的准确率:恶性为81.48%,良性为73.91%。     

应用MPEG-7建立乳腺X线影像数据库

乳腺肿瘤是妇女的多发疾病。建立有效的数字化乳腺X线影像存储与检索技术，可以充分利用医疗资源，并且帮助用户进行个人健康管理。MPEG-7技术可以实现对多媒体内容低级特征和高级语义信息的注释。将MPEG-7技术引入数字化乳腺X线影像存储，详细分析对医学影像MPEG-7注释以及将MPEG-7文件与医学图像一起存储到数据库的技术实现，在Linux系统下构建包含MPEG-7信息的数字化乳腺X线影像MySQL数据库和基于网络的查询系统。结果说明使用MPEG-7对医学图像低级特征和高级语义注释是可行的，数据库中存储的注释文件有利于医学图像高级语义信息的检索。     

基于局部调整动态轮廓模型提取超声图像乳腺肿瘤边缘

提出一种基于局部调整动态轮廓模型提取超声图像乳腺肿瘤边缘的算法.该算法在Chan-Vese (CV)模型基础上,定义了一个局部调整项,采用基于水平集的动态轮廓模型提取超声图像乳腺肿瘤边缘.将该算法应用于89例临床超声图像乳腺肿瘤的边缘提取实验,结果表明:该算法比CV模型更适用于具有区域非同质性的超声图像的分割,可有效实现超声图像乳腺肿瘤边缘的提取.     

一种基于小波变换的乳腺X线图肿块分割方法

目的：乳腺癌的早期诊断和治疗是能够降低乳腺癌患者死亡率的有效途径。通过乳腺X线图像观察乳腺状况是目前乳腺癌普查的首选影像方法。随着图像处理技术的高速发展,计算机辅助检测技术在乳腺癌的检测方面起到越来越重要的作用。方法：本文首先利用图像处理领域的形态学处理、区域增长等相关知识,对乳腺X线图像进行预处理操作,去除图像中所包含的干扰信息。之后提出一种对图像的灰度直方图进行小波变换,并根据其小波变换的模极大值点确定图像分割阈值的方法对乳腺X线图像中的疑似肿块区域进行粗分割。在通过粗分割过程获得乳腺肿块的大致位置信息之后,再利用区域增长的方法获得肿块的边缘信息。结果：本文选取MIAS乳腺图像数据库中的65幅图像作为测试图像,保证每幅图像至少包含一个乳腺肿块。利用本文所提方法对这65幅图像进行实验,并将实验结果与该数据库中的专家标注信息作对比,实验结果为当采用db40的小波系数时的检出率为95．5％。结论：本文所述方法能够有效地分割出乳腺X线图中的肿块区域,并且有较高的检出率,具有进一步研究和应用的价值。     

医学图像语义概念识别方法研究

医学图像语义概念识别是医学图像知识表示的重要技术环节。研究医学图像语义概念识别方法,有助于机器理解和学习医学图像中的潜在医学知识,在影像辅助诊断和智能读片等应用中发挥重要作用。将医学图像的高频概念识别问题转化为多标签分类任务,利用基于卷积神经网络的深度迁移学习方法,识别有限数量的高频医学概念;同时利用基于图像检索的主题建模方法,从给定医学图像的相似图像中提取语义相关概念。国际跨语言图像检索论坛ImageCLEF于2018年5月组织ImageCLEFcaption 2018评测,其子任务“概念检测”的目标是给定222 314张训练图片和9 938张测试图片,识别111 156个语义概念。上述两种方法的实验结果已被提交。实验结果表明,利用基于卷积神经网络的深度迁移学习方法识别医学图像高频概念,F1值为0.092 8,在提交团队中排名第二;基于图像检索的主题模型可召回部分低频相关概念,F1值为0.090 7,然而其性能依赖于图像检索结果的质量。基于卷积神经网络的深度迁移学习方法识别医学图像高频概念的鲁棒性优于基于图像检索方法的鲁棒性,但在大规模开放语义概念的识别技术研究上仍需进一步完善。     

Using semantic dependencies for consistency management of an ontology of brain-cortex anatomy

OBJECTIVE: In the context of the Semantic Web, ontologies have to be usable by software agents as well as by humans. Therefore, they must meet explicit representation and consistency requirements. This article describes a method for managing the semantic consistency of an ontology of brain-cortex anatomy. METHOD: The methodology relies on the explicit identification of the relationship properties and of the dependencies that might exist among concepts or relationships. These dependencies have to be respected for insuring the semantic consistency of the model. We propose a method for automatically generating all the dependent items. As a consequence, knowledge base updates are easier and safer. RESULT: Our approach is composed of three main steps: (1) providing a realistic representation, (2) ensuring the intrinsic consistency of the model and (3) checking its incremental consistency. The corner stone of ontological modeling lies in the expressiveness of the model and in the sound principles that structure it. This part defines the ideal possibilities of the ontology and is called realism of representation. Regardless of how well a model represents reality, the intrinsic consistency of a model corresponds to its lack of contradiction. This step is particularly important as soon as dependencies between relationships or concepts have to be fulfilled. Eventually, the incremental consistency encompasses the respect of the two previous criteria during the successive updates of the ontology. CONCLUSION: The explicit representation of dependencies among concepts and relationships in an ontology can be helpfully used to assist in the management of the knowledge base and to ensure the model's semantic consistency.     

GOClonto: An ontological clustering approach for conceptualizing PubMed abstracts

Concurrent with progress in biomedical sciences, an overwhelming of textual knowledge is accumulating in the biomedical literature. PubMed is the most comprehensive database collecting and managing biomedical literature. To help researchers easily understand collections of PubMed abstracts, numerous clustering methods have been proposed to group similar abstracts based on their shared features. However, most of these methods do not explore the semantic relationships among groupings of documents, which could help better illuminate the groupings of PubMed abstracts. To address this issue, we proposed an ontological clustering method called GOClonto for conceptualizing PubMed abstracts. GOClonto uses latent semantic analysis (LSA) and gene ontology (GO) to identify key gene-related concepts and their relationships as well as allocate PubMed abstracts based on these key gene-related concepts. Based on two PubMed abstract collections, the experimental results show that GOClonto is able to identify key gene-related concepts and outperforms the STC (suffix tree clustering) algorithm, the Lingo algorithm, the Fuzzy Ants algorithm, and the clustering based TRS (tolerance rough set) algorithm. Moreover, the two ontologies generated by GOClonto show significant informative conceptual structures.     

Computation of semantic similarity within an ontology of breast pathology to assist inter-observer consensus

Computer-assisted consensus in medical imaging involves automatic comparison of morphological abnormalities observed by physicians in images. We built an ontology of morphological abnormalities in breast pathology to assist inter-observer consensus. Concepts of morphological abnormalities extracted from existing terminologies, published grading systems and medical reports were organized in an taxonomic hierarchy and furthermore linked by the relation "is a diagnostic criterion of" according to diagnostic meaning. We implemented position-based, content-based and mixed semantic similarity measures between concepts in this ontology and compared the results with experts' judgment. The position-based similarity measure using both taxonomic and non-taxonomic relations performed as well as the other measures and was used for automatic comparison of morphological abnormalities within the IDEM computer-assisted consensus platform.     

Measures of semantic similarity and relatedness in the biomedical domain

Measures of semantic similarity between concepts are widely used in Natural Language Processing. In this article, we show how six existing domain-independent measures can be adapted to the biomedical domain. These measures were originally based on WordNet, an English lexical database of concepts and relations. In this research, we adapt these measures to the SNOMED-CT ontology of medical concepts. The measures include two path-based measures, and three measures that augment path-based measures with information content statistics from corpora. We also derive a context vector measure based on medical corpora that can be used as a measure of semantic relatedness. These six measures are evaluated against a newly created test bed of 30 medical concept pairs scored by three physicians and nine medical coders. We find that the medical coders and physicians differ in their ratings, and that the context vector measure correlates most closely with the physicians, while the path-based measures and one of the information content measures correlates most closely with the medical coders. We conclude that there is a role both for more flexible measures of relatedness based on information derived from corpora, as well as for measures that rely on existing ontological structures.     

Semantic enrichment for medical ontologies

The Unified Medical Language System (UMLS) contains two separate but interconnected knowledge structures, the Semantic Network (upper level) and the Metathesaurus (lower level). In this paper, we have attempted to work out better how the use of such a two-level structure in the medical field has led to notable advances in terminologies and ontologies. However, most ontologies and terminologies do not have such a two-level structure. Therefore, we present a method, called semantic enrichment, which generates a two-level ontology from a given one-level terminology and an auxiliary two-level ontology. During semantic enrichment, concepts of the one-level terminology are assigned to semantic types, which are the building blocks of the upper level of the auxiliary two-level ontology. The result of this process is the desired new two-level ontology. We discuss semantic enrichment of two example terminologies and how we approach the implementation of semantic enrichment in the medical domain. This implementation performs a major part of the semantic enrichment process with the medical terminologies, with difficult cases left to a human expert.     

Building an ontology of pulmonary diseases with natural language processing tools using textual corpora

Pathologies and acts are classified in thesauri to help physicians to code their activity. In practice, the use of thesauri is not sufficient to reduce variability in coding and thesauri are not suitable for computer processing. We think the automation of the coding task requires a conceptual modeling of medical items: an ontology. Our task is to help lung specialists code acts and diagnoses with software that represents medical knowledge of this concerned specialty by an ontology. The objective of the reported work was to build an ontology of pulmonary diseases dedicated to the coding process. To carry out this objective, we develop a precise methodological process for the knowledge engineer in order to build various types of medical ontologies. This process is based on the need to express precisely in natural language the meaning of each concept using differential semantics principles. A differential ontology is a hierarchy of concepts and relationships organized according to their similarities and differences. Our main research hypothesis is to apply natural language processing tools to corpora to develop the resources needed to build the ontology. We consider two corpora, one composed of patient discharge summaries and the other being a teaching book. We propose to combine two approaches to enrich the ontology building: (i) a method which consists of building terminological resources through distributional analysis and (ii) a method based on the observation of corpus sequences in order to reveal semantic relationships. Our ontology currently includes 1550 concepts and the software implementing the coding process is still under development. Results show that the proposed approach is operational and indicates that the combination of these methods and the comparison of the resulting terminological structures give interesting clues to a knowledge engineer for the building of an ontology.     

Enabling semantic similarity estimation across multiple ontologies: an evaluation in the biomedical domain

The estimation of the semantic similarity between terms provides a valuable tool to enable the understanding of textual resources. Many semantic similarity computation paradigms have been proposed both as general-purpose solutions or framed in concrete fields such as biomedicine. In particular, ontology-based approaches have been very successful due to their efficiency, scalability, lack of constraints and thanks to the availability of large and consensus ontologies (like WordNet or those in the UMLS). These measures, however, are hampered by the fact that only one ontology is exploited and, hence, their recall depends on the ontological detail and coverage. In recent years, some authors have extended some of the existing methodologies to support multiple ontologies. The problem of integrating heterogeneous knowledge sources is tackled by means of simple terminological matchings between ontological concepts. In this paper, we aim to improve these methods by analysing the similarity between the modelled taxonomical knowledge and the structure of different ontologies. As a result, we are able to better discover the commonalities between different ontologies and hence, improve the accuracy of the similarity estimation. Two methods are proposed to tackle this task. They have been evaluated and compared with related works by means of several widely-used benchmarks of biomedical terms using two standard ontologies (WordNet and MeSH). Results show that our methods correlate better, compared to related works, with the similarity assessments provided by experts in biomedicine.