A Novel Hybrid Image Segmentation Method for Detection of Suspicious Regions in Mammograms Based on Adaptive Multi-Thresholding (HCOW)

Abstract

Suspicious region segmentation is one of the most important parts of CAD systems that are used for breast cancer detection in mammograms. In a CAD system, there can be so many suspicious regions determined for a mammogram because of the complex structure of the breast. This study proposes a hybrid thresholding method to use in the CAD systems for efficient segmentation of the mammograms and reducing the number of the suspicious regions. The proposed method provides fully-automatic segmentation of the suspicious regions. This method is based on determining an adaptive multi-threshold value by using three different techniques together. These techniques are Otsu multilevel thresholding, Havrda & Charvat entropy, and w-BSAFCM algorithm that was developed by the authors of this paper for image clustering applications. In the proposed method, segmentation of a mammogram is performed on two sub-images obtained from that mammogram, the pectoral muscle and the breast region to prevent any information loss. The method was tested on 55 mass-mammograms and 210 non-mass mammograms of the mini-MIAS database, and it was compared with Shannon, Renyi, and Kapur entropy methods and with some of the related studies from the literature. The segmentation results of the tests were evaluated in terms of the number of suspicious regions, the number of correctly detected masses, and the performance measure parameters, accuracy, false-positive rate, specificity, volumetric overlap, and dice similarity coefficient. According to the evaluations, it was shown that the proposed method can both successfully locate the mass regions and significantly reduce the number of the non-mass suspicious regions on the mammograms.

Spatial Attention Mechanism and Cascade Feature Extraction in a U-Net Model for Enhancing Breast Tumor Segmentation

“…Table 1 presents a comparison of the sensitivity, specificity, and accuracy results obtained from five distinct methods applied to the Mini-MIAS dataset for breast tumor segmentation. The five techniques are Deep Supervision [16], Capsule Neural Network [17], Deep Features [18], HCOW [19], and the proposed model. We trained each of the five methods on the Mini-MIAS dataset and evaluated their performance in terms of sensitivity, specificity, and accuracy.…”

“…The ground truth segmentation was represented in green, while the segmented output was shown in yellow. (a) Deep Supervision [16], (b) Capsule Neural Network [17], (c) Deep Features [18], (d) HCOW [19], and (e) Our framework.…”

A New Fat-Removal-Based Preprocessing Pipeline for MLO View in Digital Mammograms

“…However, global or local thresholding techniques are frequently used for breast delineation [14]. The difficulty of the binarization techniques lies in selecting a strong threshold value [20] which can differentiate among regions (e.g., the area of the breast, pectoral muscle, and background [19]).…”

Lung nodule pre-diagnosis and insertion path planning for chest CT images

“…Later, in 2019, Khan [15] used a combination of fuzzy logic and a color featurebased fuzzy C-means clustering method to segment the lung parenchyma. For the boundary repair method, there are the rolling ball method, mathematical morphology method, convex hull, and snake algorithm [16][17][18][19][20]. When the boundary defect is large, the rolling ball method to repair the boundary easily fails, and the convex hull algorithm is also ineffective.…”