The H-scan approach to classifying ultrasound echoes

对超声回波进行分类的 H 扫描方法

基本信息

批准号：
9754139
负责人：
KEVIN J PARKER
金额：
$ 22.97万
依托单位：
UNIVERSITY OF ROCHESTER
依托单位国家：
美国
项目类别：
财政年份：
2018
资助国家：
美国
起止时间：
2018-08-01 至 2021-05-31
项目状态：
已结题

来源：
https://reporter.nih.gov/project-details/9754139
关键词：
Acoustics Address Animal Model Animals Area Biological Biological Assay Cell Size Cellular Structures Color Data Dependence Diagnosis Diagnostic Procedure Disease Ensure Fatty Liver Fatty acid glycerol esters Fibrosis Frequencies Gaussian model Gel Glucose Goals Hepatectomy High Fat Diet Histology Human Image Image Analysis Inflammation Inflammatory Link Liver Liver diseases Lobular Malignant Neoplasms Mathematics Modeling Nature Organ Output Pathology Performance Placenta Polynomial Models Probability Radiology Specialty Rattus Resolution Scanning Series Serum Shapes Site Steatohepatitis Structure Systems Analysis Techniques Testing Time Tissues Transducers Triglycerides Ultrasonography Vacuole Variant Vesicle Weight Wistar Rats attenuation clinical practice clinically significant electric impedance experimental study imaging platform improved in vivo innovation interest statistics

项目摘要

Traditional ultrasound B-Scan images show the envelope of received echoes as a grey scale image. The echoes are produced from specular reflections and scattering sites where changes in acoustic impedance occur. A long- standing area of interest concerns the frequency dependence of backscattered ultrasound from within different tissues. Some advanced backscatter analyses estimate the frequency dependence and angular dependence of backscattered waves. However, most statistical averaging techniques require some region of interest over which to calculate the expected value of scattering parameters. Unfortunately, the unfavorable statistics of ultrasound echoes can limit the spatial resolution or accuracy of these estimators. Specific changes in pathology can cause changes in the underlying size, structure, and composition of tissue, and so the goal of somehow capturing these changes remains. We hypothesize that a new matched filter approach using matched Hermite functions can classify and visualize major scattering classes at high resolution. This enables clinicians to distinguish subtle cellular and parenchymal changes that would otherwise appear similar, thereby adding new and relevant information to diagnostic ultrasound. The recently derived H-scan is a fresh approach where the received echoes can be linked to three major classes of echoes from tissues. Echoes are linked to the mathematics of Gaussian Weighted Hermite Polynomials so that the overall identification task can be simplified. The resulting images are denoted as H-scans, where ‘H’ represents Hermite or hue, since the identification by hue is distinct from the traditional B-scan. The framework was given an initial test in biological tissues – liver and placenta – where changes in tissue H-scan images are plausibly linked to changes in the concentration of small scatterers. However, in order to establish H-Scan as a viable diagnostic technique, two issues must be proven. First, the H-scan must be shown to give consistent results within tissues over a range of depths and despite attenuation. Second, the H-scan must be shown to be sensitive to cellular and sub-cellular changes in tissue scatterers, relevant to a clinically significant condition. This project will address both these issues. The depth and attenuation dependence will be studied and corrected in a series of phantom and tissue experiments. The sensitivity and accuracy will be tested in a liver steatosis model in rats. The results should establish the key performance issues for H-scan, and thereby characterize its ability to advance diagnostic ultrasound imaging for assessing pathology in humans.

传统的超声 B 扫描图像将接收回波的包络显示为灰度图像。由发生声阻抗变化的镜面反射和散射点产生。感兴趣的站立区域涉及来自不同内部的反向散射超声波的频率依赖性一些先进的反向散射分析估计了频率依赖性和角度依赖性。然而，大多数统计平均技术需要一些感兴趣的区域不幸的是，超声的统计数据不利。回声会限制这些估计器的空间分辨率或准确性，从而导致病理学的特定变化。组织的基本尺寸、结构和成分发生变化，因此以某种方式捕获这些变化的目标我们发现，使用匹配 Hermite 函数的新匹配过滤器方法仍然存在。以高分辨率对主要散射类别进行分类，这使得能够区分细微的差别。细胞和实质变化，否则会显得相似，从而增加新的和相关的最近推出的 H 扫描是一种新方法，其中接收到的回声。可以与来自组织的三大类回声联系起来。回声与高斯数学相关。加权 Hermite 多项式可以简化整体识别任务生成的图像。表示为 H 扫描，其中“H”代表 Hermite 或色调，因为色调的识别与该框架在生物组织（肝脏和胎盘）中进行了初步测试。组织 H 扫描图像的变化似乎与小散射体浓度的变化有关。然而，为了将 H-Scan 确立为一种可行的诊断技术，必须证明两个问题：首先。尽管有衰减，H 扫描必须能够在一定深度范围内的组织内提供一致的结果。其次，必须证明 H 扫描对组织散射体中的细胞和亚细胞变化敏感，该项目将解决这些问题的深度和重要性。衰减依赖性将在一系列模型和组织实验中进行研究和纠正。将在大鼠肝脏脂肪变性模型中测试敏感性和准确性，结果应确定关键。 H 扫描的性能问题，并由此表征其推进超声诊断成像的能力评估人类的病理学。

项目成果

期刊论文数量（16）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

Scattering Signatures of Normal versus Abnormal Livers with Support Vector Machine Classification.

DOI：
10.1016/j.ultrasmedbio.2020.08.009
发表时间：
2020-12
期刊：
ULTRASOUND IN MEDICINE AND BIOLOGY
影响因子：
2.9
作者：
Baek, Jihye;Poul, Sedigheh S.;Swanson, Terri A.;Tuthill, Theresa;Parker, Kevin J.
通讯作者：
Parker, Kevin J.

Fine-tuning the H-scan for discriminating changes in tissue scatterers.