Lesion Segmentation in Breast Ultrasound Images using Optimized Marked Watershed Method

Abstract Background: Breast cancer is one of the most serious diseases threatening women’s health. Early screening based on ultrasound can help to detect and treat tumors in early stage. However, due to the lack of radiologists with professional skills, ultrasound based breast cancer screening has not been widely used in rural area. Computer-aided diagnosis (CAD) technology can eﬀectively alleviates this problem. Since Breast Ultrasound (BUS) images have low resolution and speckle noise, lesion segmentation, which is an important step in CAD system, is challenging.Results: Two datasets were used for evaluation. Dataset A comprises 500 BUS images from local hospitals, while dataset B comprises 205 BUS images from open source. The experimental results show that the proposed method outperformed its related classic segmentation methods and the state-of-the-art deep learning model, RDAU–NET. And its’ Accuracy(Acc), Dice eﬃcient(DSC) and Jaccard Index(JI) reached 96.25%, 78.4% and 65.34% on dataset A, and ACC, DC and Sen reached 97.96%, 86.25% and 88.79% on dataset B.Conclusions: We proposed an adaptive morphology snake based on marked watershed(AMSMW) algorithm for BUS images segmentation. It was proven to be robust, eﬃcient and eﬀective. In addition, it was found to be more sensitive to malignant lesions than benign lesions. What’s more, since the Rectangular Region of Interest(RROI) in the proposed method is drawn manually, we will consider adding deep learning network to automatically identify RROI, and completely liberate the hands of radiologists.Methods: The proposed method consists of two main steps. In the ﬁrst step, we used Contrast Limited Adaptive Histogram Equalization(CLAHE) and Side Window Filter(SWF) to preprocess BUS images. Therefore, lesion contours can be eﬀectively highlighted and the inﬂuence of noise can be eliminated to a great extent. In the second step, we proposed adaptative morphology snake(AMS) as an embedded segmentation function of AMSMW. It can adjust working parameters adaptively, according to diﬀerent lesions’ size. By combining segmentation results of AMS with marker region obtained by morphological method, we got the marker region of marked watershed (MW). Finally, we obtained candidate contours by MW. And the best lesion contour was chosen by maximum Average Radial Derivative(ARD).

Download Full-text

Lesion segmentation in breast ultrasound images using the optimized marked watershed method

BioMedical Engineering OnLine ◽

10.1186/s12938-021-00891-7 ◽

2021 ◽

Vol 20 (1) ◽

Author(s):

Xiaoyan Shen ◽

He Ma ◽

Ruibo Liu ◽

Hong Li ◽

Jiachuan He ◽

...

Keyword(s):

Breast Cancer ◽

Rural Areas ◽

Early Stage ◽

Speckle Noise ◽

Breast Ultrasound ◽

Dice Similarity Coefficient ◽

Ultrasound Images ◽

Lesion Segmentation ◽

Early Screening ◽

Side Window

Abstract Background Breast cancer is one of the most serious diseases threatening women’s health. Early screening based on ultrasound can help to detect and treat tumours in the early stage. However, due to the lack of radiologists with professional skills, ultrasound-based breast cancer screening has not been widely used in rural areas. Computer-aided diagnosis (CAD) technology can effectively alleviate this problem. Since breast ultrasound (BUS) images have low resolution and speckle noise, lesion segmentation, which is an important step in CAD systems, is challenging. Results Two datasets were used for evaluation. Dataset A comprises 500 BUS images from local hospitals, while dataset B comprises 205 open-source BUS images. The experimental results show that the proposed method outperformed its related classic segmentation methods and the state-of-the-art deep learning model RDAU-NET. Its accuracy (Acc), Dice similarity coefficient (DSC) and Jaccard index (JI) reached 96.25%, 78.4% and 65.34% on dataset A, and its Acc, DSC and sensitivity reached 97.96%, 86.25% and 88.79% on dataset B, respectively. Conclusions We proposed an adaptive morphological snake based on marked watershed (AMSMW) algorithm for BUS image segmentation. It was proven to be robust, efficient and effective. In addition, it was found to be more sensitive to malignant lesions than benign lesions. Methods The proposed method consists of two steps. In the first step, contrast limited adaptive histogram equalization (CLAHE) and a side window filter (SWF) are used to preprocess BUS images. Lesion contours can be effectively highlighted, and the influence of noise can be eliminated to a great extent. In the second step, we propose adaptive morphological snake (AMS). It can adjust the working parameters adaptively according to the size of the lesion. Its segmentation results are combined with those of the morphological method. Then, we determine the marked area and obtain candidate contours with a marked watershed (MW). Finally, the best lesion contour is chosen by the maximum average radial derivative (ARD).

Download Full-text

Evaluating the Accuracy of Breast Cancer and Molecular Subtype Diagnosis by Ultrasound Image Deep Learning Model

Frontiers in Oncology ◽

10.3389/fonc.2021.623506 ◽

2021 ◽

Vol 11 ◽

Author(s):

Xianyu Zhang ◽

Hui Li ◽

Chaoyun Wang ◽

Wen Cheng ◽

Yuntao Zhu ◽

...

Keyword(s):

Breast Cancer ◽

Deep Learning ◽

Sensitivity And Specificity ◽

Molecular Subtypes ◽

Learning Model ◽

Breast Ultrasound ◽

Ultrasound Images ◽

Test Set ◽

External Test ◽

Deep Learning Model

Background: Breast ultrasound is the first choice for breast tumor diagnosis in China, but the Breast Imaging Reporting and Data System (BI-RADS) categorization routinely used in the clinic often leads to unnecessary biopsy. Radiologists have no ability to predict molecular subtypes with important pathological information that can guide clinical treatment.Materials and Methods: This retrospective study collected breast ultrasound images from two hospitals and formed training, test and external test sets after strict selection, which included 2,822, 707, and 210 ultrasound images, respectively. An optimized deep learning model (DLM) was constructed with the training set, and the performance was verified in both the test set and the external test set. Diagnostic results were compared with the BI-RADS categorization determined by radiologists. We divided breast cancer into different molecular subtypes according to hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2) expression. The ability to predict molecular subtypes using the DLM was confirmed in the test set.Results: In the test set, with pathological results as the gold standard, the accuracy, sensitivity and specificity were 85.6, 98.7, and 63.1%, respectively, according to the BI-RADS categorization. The same set achieved an accuracy, sensitivity, and specificity of 89.7, 91.3, and 86.9%, respectively, when using the DLM. For the test set, the area under the curve (AUC) was 0.96. For the external test set, the AUC was 0.90. The diagnostic accuracy was 92.86% with the DLM in BI-RADS 4a patients. Approximately 70.76% of the cases were judged as benign tumors. Unnecessary biopsy was theoretically reduced by 67.86%. However, the false negative rate was 10.4%. A good prediction effect was shown for the molecular subtypes of breast cancer with the DLM. The AUC were 0.864, 0.811, and 0.837 for the triple-negative subtype, HER2 (+) subtype and HR (+) subtype predictions, respectively.Conclusion: This study showed that the DLM was highly accurate in recognizing breast tumors from ultrasound images. Thus, the DLM can greatly reduce the incidence of unnecessary biopsy, especially for patients with BI-RADS 4a. In addition, the predictive ability of this model for molecular subtypes was satisfactory,which has specific clinical application value.

Download Full-text

BGM-Net: Boundary-Guided Multiscale Network for Breast Lesion Segmentation in Ultrasound

Frontiers in Molecular Biosciences ◽

10.3389/fmolb.2021.698334 ◽

2021 ◽

Vol 8 ◽

Author(s):

Yunzhu Wu ◽

Ruoxin Zhang ◽

Lei Zhu ◽

Weiming Wang ◽

Shengwen Wang ◽

...

Keyword(s):

Breast Lesion ◽

Speckle Noise ◽

Breast Ultrasound ◽

Ultrasound Images ◽

Lesion Segmentation ◽

Low Contrast ◽

Feature Enhancement ◽

Boundary Map ◽

Weak Boundaries ◽

Feature Pyramid

Automatic and accurate segmentation of breast lesion regions from ultrasonography is an essential step for ultrasound-guided diagnosis and treatment. However, developing a desirable segmentation method is very difficult due to strong imaging artifacts e.g., speckle noise, low contrast and intensity inhomogeneity, in breast ultrasound images. To solve this problem, this paper proposes a novel boundary-guided multiscale network (BGM-Net) to boost the performance of breast lesion segmentation from ultrasound images based on the feature pyramid network (FPN). First, we develop a boundary-guided feature enhancement (BGFE) module to enhance the feature map for each FPN layer by learning a boundary map of breast lesion regions. The BGFE module improves the boundary detection capability of the FPN framework so that weak boundaries in ambiguous regions can be correctly identified. Second, we design a multiscale scheme to leverage the information from different image scales in order to tackle ultrasound artifacts. Specifically, we downsample each testing image into a coarse counterpart, and both the testing image and its coarse counterpart are input into BGM-Net to predict a fine and a coarse segmentation maps, respectively. The segmentation result is then produced by fusing the fine and the coarse segmentation maps so that breast lesion regions are accurately segmented from ultrasound images and false detections are effectively removed attributing to boundary feature enhancement and multiscale image information. We validate the performance of the proposed approach on two challenging breast ultrasound datasets, and experimental results demonstrate that our approach outperforms state-of-the-art methods.

Download Full-text

Perbandingan Teknik Reduksi Derau Speckle Pada Citra Ultrasonograpi Payudara

Jurnal Elektro dan Mesin Terapan - JURNAL ELEMENTER ◽

10.35143/elementer.v4i2.2409 ◽

2018 ◽

Vol 4 (2) ◽

pp. 27-36

Author(s):

Yuli Triyani

Keyword(s):

Breast Cancer ◽

Ultrasound Image ◽

Speckle Noise ◽

Breast Ultrasound ◽

Ultrasound Images ◽

Reduction Techniques ◽

Noise Filter ◽

Incidence And Mortality ◽

Structural Content ◽

Aided Diagnosis

Breast cancer is the most commonly diagnosed cancer with the highest prevalence, incidence, and mortality rate for females in Indonesia and worldwide. Ultrasonography is a recommended modality for breast cancer, because it is comfortable, radiation free and it can be widely used. However, ultrasound images often occur in quality degradation caused by speckle noise that appears during image acquisition. It causes difficulty for radiologists or Computer Aided Diagnosis (CAD) systems to diagnose these images. Some techniques are proposed for reducing the speckle noise. This journal aims to compare the performance of 14 noise reduction techniques in breast ultrasound images. Quantitative testing was carried out on 58 breast ultrasound images and 3 artificial breast ultrasound image. The quantitative parameters are used include texture analysis (Mean, Variant, skewness, kurtosis, contrast and entropy) and evaluation of image quality (MSE, RMSE, SNR, SSIM, Structural content and Maximum Difference). The qualitative testing was also carried out with the assessment of 3 radiology specialists on 3 samples of each reduction technique. Based on test results, the 3 best performance filters are DsFsrad, DsFamedian dan DsFhmedian. Keywords: Ultrasound, speckle noise, filter

Download Full-text

A q-Extension of Sigmoid Functions and the Application for Enhancement of Ultrasound Images

Entropy ◽

10.3390/e21040430 ◽

2019 ◽

Vol 21 (4) ◽

pp. 430 ◽

Cited By ~ 2

Author(s):

Paulo Sergio Rodrigues ◽

Guilherme Wachs-Lopes ◽

Ricardo Morello Santos ◽

Eduardo Coltri ◽

Gilson Antonio Giraldi

Keyword(s):

Image Processing ◽

Asymptotic Behavior ◽

Deep Learning ◽

Speckle Noise ◽

Histogram Equalization ◽

Regions Of Interest ◽

Ultrasound Images ◽

Learning Approaches ◽

Exponential Functions ◽

Sigmoid Functions

This paper proposes the q-sigmoid functions, which are variations of the sigmoid expressions and an analysis of their application to the process of enhancing regions of interest in digital images. These new functions are based on the non-extensive Tsallis statistics, arising in the field of statistical mechanics through the use of q-exponential functions. The potential of q-sigmoids for image processing is demonstrated in tasks of region enhancement in ultrasound images which are highly affected by speckle noise. Before demonstrating the results in real images, we study the asymptotic behavior of these functions and the effect of the obtained expressions when processing synthetic images. In both experiments, the q-sigmoids overcame the original sigmoid functions, as well as two other well-known methods for the enhancement of regions of interest: slicing and histogram equalization. These results show that q-sigmoids can be used as a preprocessing step in pipelines including segmentation as demonstrated for the Otsu algorithm and deep learning approaches for further feature extractions and analyses.

Download Full-text

Using Deep Learning Methods to Analyze Automated Breast Ultrasound Images, to Establish a Diagnosis, Therapy Assessment and Prognosis Prediction Model of Breast Cancer.

Case Medical Research ◽

10.31525/ct1-nct04270032 ◽

2020 ◽

Author(s):

Keyword(s):

Breast Cancer ◽

Deep Learning ◽

Prediction Model ◽

Breast Ultrasound ◽

Ultrasound Images ◽

Learning Methods ◽

Therapy Assessment ◽

Prognosis Prediction Model ◽

Diagnosis Therapy ◽

Automated Breast Ultrasound

Download Full-text

Lesion Segmentation in Breast Ultrasound Images using Optimized Marked Watershed Method

10.21203/rs.3.rs-137797/v1 ◽

2020 ◽

Author(s):

Xiaoyan Shen ◽

He Ma ◽

Ruibo Liu ◽

Hong Li ◽

Jiachuan He ◽

...

Keyword(s):

Malignant Tumors ◽

Classical Method ◽

Ultrasound Image ◽

Breast Ultrasound ◽

Experimental Result ◽

Benign Tumors ◽

Ultrasound Images ◽

Lesion Segmentation ◽

Obvious Effect ◽

Side Window

Abstract Background: Ultrasound is the most popular tool for early detection of breast cancer because of its non radiation and low cost. However, breast ultrasound(BUS) images have low resolution and speckle noise, which make lesion segmentation become a challenge. Most of deep learning(DL) models applied on images segmentation don't have good generalization ability for BUS images. Therefore, it is time to go back to the classical method and consider combining it with DL to achieve more accurate and efficient effect in a semi-automatic way.Methods: This paper mainly proposed an effective and efficient semi-automatic BUS images segmentation method, Adaptive morphological snake and marked watershed( AMSMW). It includes two parts: preprocessing and segmentation. In the first part, we combine contrast limited adaptive histogram equalization(CLAHE) and side window filtering(SWF) methods for the first time. In the second part, We use the proposed adaptive morphological snake algorithm (AMS) to provide a mark for marked watershed(MW) method. Results: we tested on 500 BUS images, whose ratio of benign and malignant is 1:1. After quantitative and qualitative analysis, AMSMW is proven to outperform existing classical methods on the effectiveness and efficiency. Furthermore, we compared with Zhuang’s RDAU-NET on both our dataset and theirs. Experimental result showes AMSMW achieved better performance on most of indicators, including loss, accuracy, sensitivity, dice and F1-score. Conlusions: The new image preprocessing method proposed by us has obvious effect on segmentation of breast ultrasound image. In addition, the proposed adaptive morphology snake method and optimized marked watershed turn out to be more efficient and effective than some relative classical method and the advanced DL method at present. Moreover, by studying on the algorithm’s sensitivity in segmenting benign and malignant tumors, we found that AMSMW is more sensitivity to malignant tumors, and more stable to benign tumors, which is significant for further research of precision medicine.

Download Full-text

Diagnosing breast cancer tumors using stacked ensemble model

Journal of Intelligent & Fuzzy Systems ◽

10.3233/jifs-219176 ◽

2021 ◽

pp. 1-9

Author(s):

Ahmet Haşim Yurttakal ◽

Hasan Erbay ◽

Türkan İkizceli ◽

Seyhan Karaçavuş ◽

Cenker Biçer

Keyword(s):

Breast Cancer ◽

Deep Learning ◽

Medical Imaging ◽

Early Stage ◽

False Negative ◽

Gradient Boosting ◽

Physical Sign ◽

Ensemble Model ◽

Learning Methods ◽

Dce Mri

Breast cancer is the most common cancer that progresses from cells in the breast tissue among women. Early-stage detection could reduce death rates significantly, and the detection-stage determines the treatment process. Mammography is utilized to discover breast cancer at an early stage prior to any physical sign. However, mammography might return false-negative, in which case, if it is suspected that lesions might have cancer of chance greater than two percent, a biopsy is recommended. About 30 percent of biopsies result in malignancy that means the rate of unnecessary biopsies is high. So to reduce unnecessary biopsies, recently, due to its excellent capability in soft tissue imaging, Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) has been utilized to detect breast cancer. Nowadays, DCE-MRI is a highly recommended method not only to identify breast cancer but also to monitor its development, and to interpret tumorous regions. However, in addition to being a time-consuming process, the accuracy depends on radiologists’ experience. Radiomic data, on the other hand, are used in medical imaging and have the potential to extract disease characteristics that can not be seen by the naked eye. Radiomics are hard-coded features and provide crucial information about the disease where it is imaged. Conversely, deep learning methods like convolutional neural networks(CNNs) learn features automatically from the dataset. Especially in medical imaging, CNNs’ performance is better than compared to hard-coded features-based methods. However, combining the power of these two types of features increases accuracy significantly, which is especially critical in medicine. Herein, a stacked ensemble of gradient boosting and deep learning models were developed to classify breast tumors using DCE-MRI images. The model makes use of radiomics acquired from pixel information in breast DCE-MRI images. Prior to train the model, radiomics had been applied to the factor analysis to refine the feature set and eliminate unuseful features. The performance metrics, as well as the comparisons to some well-known machine learning methods, state the ensemble model outperforms its counterparts. The ensembled model’s accuracy is 94.87% and its AUC value is 0.9728. The recall and precision are 1.0 and 0.9130, respectively, whereas F1-score is 0.9545.

Download Full-text