The task of segmenting breast tumours in mammograms is very difficult, as its difficulty lies in the lack of contrast between the tumour and the surrounding breast tissue, especially when dealing with small tumours that are not clear boundaries and hidden under the tissues. Segmentation algorithms often lose the path of tumor boundaries in an attempt to determine the position of them. Active contours are used widely for segmentation as a high-level technology for boundary recognition. The main aim to create a clear contrast between the tumour and the normal breast region. In this study, two approaches to active contour are applied: snakes and level sets. The proposed methods were applied to all abnormal mammogram images taken from mini-MIAS database. The first approach showed a weakness in the segmentation of this type of image, while the other approach was able to segment all the mammogram's tumours. The Chan-Vese method was the most superior of all the active contour segmentation methods. The proposed models were tested in two ways, the first is statistical the best result was for the Chan-Vese method and it came as follows (90%,95% 98%, 97%, 97%) for Jaccard, Dice, PF-Score, Precision, and Sensitivity respectively. And the other is based on the segmented region's characteristics, Chan-Vese was able to accurately determine the location and shape of the tumor. The proposed Chan-Vese approach is appropriate in adopting computer assisted detection systems to predict tumor boundaries and locations in mammography for its reliability and superior performance over other algorithms.

The accurate segmentation of tumours is a crucial stage of diagnosis and treatment, reducing the damage that breast cancer causes, which is the most common type of cancer among women, especially after the age of forty. The task of segmenting breast tumours in mammograms is very difficult, as its difficulty lies in the lack of contrast between the tumour and the surrounding breast tissue, especially when dealing with small tumours that are not clear boundaries and hidden under the tissues. As algorithms often lose an automatic path toward the boundaries of the tumour at try to determine the site of this type of tumour. The study aims to create a clear contrast between the tumour and the healthy breast area. For this purpose, we used a Gaussian filter as a pre-processing as it works to intensify the low-frequency components while reducing the high-frequency components as the breast structure is enhanced and noise suppression. Then, CLAHE was used to improve the contrast of the image, which increases the contrast between the tumour and the surrounding tissue and sharpens the edges of the tumour. Next, the tumour was segmented by using the Chan-Vese method with appropriate parameters defined. The proposed method was applied to all abnormal mammogram images taken from a publicly available mini-MIAS database. The proposed model was tested in two ways, the first is statistical that got results (90.1, 94.8, 95.5, 92.1, 99.5) for Jaccard, Dice, PF-Score, precision, and sensitivity respectively. And the other is based on the segmented region's characteristics that results showed the algorithm could identify the tumour with high efficiency.

Medical imaging, like ultrasound, gives a good visual picture of how an organ works. However, a radiologist has a hard time and takes a long time to process these images, which delays the diagnosis. Several automated methods for detecting and segmenting breast lesions have been developed. Nevertheless, due to ultrasonic artifacts and the intricacy of lesion forms and locations, the segmentation of lesions or tumors from breast ultrasonography remains an open issue. Medical image segmentation has seen a breakthrough thanks to deep learning. U-Net is the most noteworthy deep network in this regard. The traditional U-Net design lacks precision when dealing with complex data sets, despite its exceptional performance in segmenting multimedia medical images. To reduce texture detail redundancy and avoid overfitting, we suggest developing the U-Net architecture by including dropout layers after each max pooling layer. Batchnormalization layers and a binary cross-entropy loss function were used to preserve breast tumor texture features and edge attributes while decreasing computational costs. We used the breast ultrasound dataset of 780 images with normal, benign, or malignant tumors. Our model showed superior segmentation results for breast ultrasound pictures compared to previous deep neural networks. Quantitative measures, accuracy, and IoU values were utilized to evaluate the suggested model?s effectiveness. The results were 99.34% and 99.60% for accuracy and IoU. The results imply that the augmented U-Net model that has been suggested has high diagnostic potential in the clinic since it can correctly segment breast lesions.

No abstract