Search

Ultrasound imaging plays an important role in breast cancer screening for which early and accurate lesion detection is crucial for clinical practice. Many researches were performed on supporting the breast lesion detection based on ultrasound data. In the paper, a novel hierarchical model is proposed to automatically detect breast lesion from ultrasound 3D data. The model simultaneously considers the data information from low-level to high-level for the detection by processing with a joint probability. For each layer of the model, the corresponding algorithm is performed to denote the certain level image information. A dynamic programming approach is applied to efficiently obtain the optimal solution. With a preliminary dataset, the superior performance of the proposed model has been demonstrated for the automated detection of breast lesion with 0.375 false positive per case at 91.7% sensitivity.

Ultrasound lesion segmentation is an important and challenging task. Comparing with other methods, region-based level set has many advantages, but still requires considerable improvement to deal with the characteristic of lesions in the ultrasound modality such as shadowing, speckle and heterogeneity. In the clinical workflow, the physician would usually denote long and short axes of a lesion for measurement purpose yielding four markers in an image. Inspired by this workflow, a constrained level set method is proposed to fully utilize these four markers as prior knowledge and global constraint for the segmentation. First, the markers are detected using template-matching algorithm and B-Spline is applied to fit four markers as the initial contour. Then four-marker constrained energy is added to the region-based local level set to make sure that the contour evolves without deviation from the four markers. Finally the algorithm is implemented in a multi-resolution scheme to achieve sufficient computational efficiency. The performance of the proposed segmentation algorithm was evaluated by comparing our results with manually segmented boundaries on 308 ultrasound images with breast lesions. The proposed method achieves Dice similarity coefficient 89.49 ± 4.76% and could be run in real-time.

Polycystic Ovary Syndrome (PCOS) is a complex endocrine disorder which seriously impacts women's health. The disorder is characterized by the formation of many follicular cysts in the ovary. Nowadays the diagnosis performed by doctors is to manually count the number of follicular cysts, which may lead to problems of the variability, reproducibility and low efficiency. To overcome these problems, an automated scheme is proposed to detect the PCOS. Firstly the input ovary ultrasound image is filtered by an adaptive morphological filter. Then a modified labeled watershed algorithm is used to extract contours of targets. Finally a clustering method is applied to identify expected follicular cysts. The experimental application verifies the effectivity of this proposed scheme, which achieves the accuracy rate of 84%.