Improving Segmentation of Breast Ultrasound Images Semi Automatic Two Pointers Histogram Splitting Technique

2025-05-08 0 0 2.54MB 5 页 10玖币

侵权投诉

Improving Segmentation of Breast Ultrasound

Images: Semi Automatic Two Pointers Histogram

Splitting Technique

Rasheed Abid

Biomedical engineering

Illinois Institute of Technology

Chicago, IL, USA

r.abid94.bogra@gmail.com

S. Kaisar Alam

Rutgers University

Newark, NJ,USA

kaisar.alam@ieee.org

Abstract—Automatically segmenting lesion area in breast ul-

trasound (BUS) images is a challenging one due to its noise,

speckle and artifacts. Edge-map of BUS images also does not

help because in most cases the edge-map gives no information

whatsoever. Almost all segmentation technique takes the edge-

map of the image as its ﬁrst step, though there are a few

algorithms that try to avoid edge-maps as well. Improving the

edge-map of breast ultrasound images theoretically improves

the chances of automatic segmentation to be more precise. In

this paper, we propose a semi-automatic technique of histogram

splitting using two pointers. Here the user only has to select

two initially guessed points denoting a circle on the region of

interest (ROI). The method will automatically study the internal

histogram and split it using two pointers. The output BUS image

has improved edge-map and ultimately the segmentation on it is

better compared to regular segmentation using same algorithm

and same initialization. Also, we further processed the edge-map

to have less edge-pixels to area ratio, improving the homogeneity

and the chances of easy segmentation in the future.

Keywords—segmentation, elastography, ultrasound, B-mode,

histogram

I. INTRODUCTION

Breast cancer is the second most deadly cancer for women.

In US only, it was estimated that in 2018-2019, every 1 of 8

US women will develop invasive breast cancer, over 268,600

new cases [1]. The statistics of the previous year 2017-2018,

the number of death was 40,610 [2]. This makes breast

cancer study for ultrasound imaging a pressing need for the

researches. Irteza et. al. [3], [4] described getting strain images

from breast RF data, which can be noisy due to motion of

the patient. Getting segmentation maps along with the strain

images will help the radiologist to diagnose the tumor.

Segmentation is a challenge when applied on ultrasound

images. [5] surveys and describes about segmentation al-

gorithms used on Breast Ultrasound images. [6] mentioned

methods to ﬁnd out seed point in breast ultrasound images.

[7], [8] mentions that due to poor quality images and noise,

speckle and artifacts, it is relatively harder to ﬁnd. Many

characteristics dependent segmentation algorithms like the

automatic [9] or semi-automatic [10] or neural networks

such as described in [ref6] have tried to segment the Region

of interest (ROI). Still very high accuracy is not achieved

because the properties of breast tissues [11] plays a big

role in making shadows and low-resolution imaging make it

harder for the algorithm to understand the image. As a ﬁrst

step towards segmentation, most algorithms take the edge-map

of the image to get an approximate idea about the gradient

change. For ultrasound breast image, the edge image provides

almost no clear information at all. This is due to the fact that

our modern edge determining algorithms ﬁnd edges that are

very sensitive in nature. Even with slightest of sharp gradient

change, these algorithms such as Canny edge detector [12],

improved Sobel edge detector [13] can ﬁnd edges. In an

ultrasound image this sensitive property of the edge ﬁnding

algorithm works as a double edged sword. It ﬁgures out the

ROI but due to unwanted noise, which is many times more

in ultrasound image than in regular image, the edge ﬁnding

algorithms also considers the noise and artifacts to be edges.

Regular edge ﬁnding algorithms fail to distinguish between

ROI and other artifacts because of low signal to noise ratio

of ultrasound images. Improving the edge image around the

lesion area i.e. clearing out unnecessary noise edges, should

help us do better segmentation with traditional algorithms, at

least theoretically. In this paper, we propose a semi-automatic

technique, that improves the edge-map around the lesion area

of a BUS image, improving the segmentation precision with

regular algorithms.

Improving the edge images require manually initializing two

points on the original image, which will cover a major region

of the ROI and from there the method will initialize a simple

shape automatically. This shape is used to determine estimated

pixel information of the ROI and then two separating pointers

are used on its histogram to ﬁlter the main image. This ﬁltered

image, which we call histogram split and stretched image

(HSSI) has a better edge-map with respect to the original edge-

map. General segmentation algorithms such as active contour

[14], level set [15] and MetaMorph [16] segmentation

has yielded better results on these intermediate images. To

measure the improvement, we needed the ground truth of the

original BUS images, which were manually segmented by

arXiv:2210.10975v1 [eess.IV] 20 Oct 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ImprovingSegmentationofBreastUltrasoundImages:SemiAutomaticTwoPointersHistogramSplittingTechniqueRasheedAbidBiomedicalengineeringIllinoisInstituteofTechnologyChicago,IL,USAr.abid94.bogra@gmail.comS.KaisarAlamRutgersUniversityNewark,NJ,USAkaisar.alam@ieee.orgAbstractAutomaticallysegmentinglesionarea...

展开>> 收起<<

Improving Segmentation of Breast Ultrasound Images Semi Automatic Two Pointers Histogram Splitting Technique.pdf

共5页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Improving Segmentation of Breast Ultrasound Images Semi Automatic Two Pointers Histogram Splitting Technique

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: