Multi-Dimensional Tumor Detection in Automated Whole Breast Ultrasound Using Topographic Watershed

Abstract

Automated whole breast ultrasound (ABUS) is becoming a popular screening modality for whole breast examination. Compared to conventional handheld ultrasound, ABUS achieves operator-independent and is feasible for mass screening. However, reviewing hundreds of slices in an ABUS image volume is time-consuming. A computer-aided detection (CADe) system based on watershed transform was proposed in this study to accelerate the reviewing. The watershed transform was applied to gather similar tissues around local minima to be homogeneous regions. The likelihoods of being tumors of the regions were estimated using the quantitative morphology, intensity, and texture features in the 2-D/3-D false positive reduction (FPR). The collected database comprised 68 benign and 65 malignant tumors. As a result, the proposed system achieved sensitivities of 100% (133/133), 90% (121/133), and 80% (107/133) with FPs/pass of 9.44, 5.42, and 3.33, respectively. The figure of merit of the combination of three feature sets is 0.46 which is significantly better than that of other feature sets ( [Formula: see text]). In summary, the proposed CADe system based on the multi-dimensional FPR using the integrated feature set is promising in detecting tumors in ABUS images.

Quantitative breast mass classification based on the integration of B-mode features and strain features in elastography

“…According to prior study [47], round shape was more likely to be a malignant finding for some malignant tumors. Also, some invasive cancers larger than 1 cm are round [48]. In such situation, the value of SpiculationNum for round malignant tumors would be close to zero.…”

Automatic breast ultrasound image segmentation: A survey

“…255 groups of markers were selected by thresholding (th = 1, 2, ⋯, 255) the image [124,125]; the external and the internal markers were defined by using the morphological dilation and erosion. Watershed method was applied to generate 255 potential lesion boundaries by using the markers on different thresholds; the average Used empirical rules to refine the results [128] 2014 Watershed Pre-segmentation Local intensity minima Used empirical rules to refine the results radial derivative (ARD) function [132,133] was applied to determine the final tumor boundary. Zhang et al [126,127] applied watershed to determine the boundaries of gray level images.…”

“…[175] is the first survey paper for BUS image segmentation published recently; however, this paper is much more comprehensive in terms of major issue discussions, future direction prediction, theory fundamentals, application theme and the number of references. In this paper, we classify breast cancer segmentation approaches into six main categories: (1) graph-based approaches [7, 9-13, 22-25, 27, 28, 32-34], (2) deformable models [42, 50, 52, 54-59, 61-65, 67, 69, 71, 73, 75, 76, 78-82, 85, 124, 139, 148, 166], (3) learning-based approaches [7,9,10,87,89,91,94,95,98,100,[103][104][105][106][107]120], (4) thresholding [22,[109][110][111][112][113][114][115], (5) region growing [54,55,113,117,118], and (6) watershed [109,122,123,[126][127][128]. As shown in Fig.…”

Detection of Breast Tumour and Speckle Noise Removal using Bilateral Filter and Bivariate Shrinkage

“…The input given to the segmentation process is the medical images and the output will be the attributes that are extracted from those images by splitting the images into various constituent parts. A topography watershed transform [10] model is implemented in ABUS images for segmentation that are based on the mathematical morphology operation. This model is applied to the gray level ABUS images for fast analysis, but is cannot manage the process of over segmentation.…”