کد مقاله کد نشریه سال انتشار مقاله انگلیسی نسخه تمام متن
6920794 864433 2016 17 صفحه PDF دانلود رایگان
عنوان انگلیسی مقاله ISI
Multi-scale texture-based level-set segmentation of breast B-mode images
موضوعات مرتبط
مهندسی و علوم پایه مهندسی کامپیوتر نرم افزارهای علوم کامپیوتر
پیش نمایش صفحه اول مقاله
Multi-scale texture-based level-set segmentation of breast B-mode images
چکیده انگلیسی
Automatic segmentation of ultrasonographic breast lesions is very challenging, due to the lesions′ spiculated nature and the variance in shape and texture of the B-mode ultrasound images. Many studies have tried to answer this challenge by applying a variety of computational methods including: Markov random field, artificial neural networks, and active contours and level-set techniques. These studies focused on creating an automatic contour, with maximal resemblance to a manual contour, delineated by a trained radiologist. In this study, we have developed an algorithm, designed to capture the spiculated boundary of the lesion by using the properties from the corresponding ultrasonic image. This is primarily achieved through a unique multi-scale texture identifier (inspired by visual system models) integrated in a level-set framework. The algorithm׳s performance has been evaluated quantitatively via contour-based and region-based error metrics. We compared the algorithm-generated contour to a manual contour delineated by an expert radiologist. In addition, we suggest here a new method for performance evaluation where corrections made by the radiologist replace the algorithm-generated (original) result in the correction zones. The resulting corrected contour is then compared to the original version. The evaluation showed: (1) Mean absolute error of 0.5 pixels between the original and the corrected contour; (2) Overlapping area of 99.2% between the lesion regions, obtained by the algorithm and the corrected contour. These results are significantly better than those previously reported. In addition, we have examined the potential of our segmentation results to contribute to the discrimination between malignant and benign lesions.
ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: Computers in Biology and Medicine - Volume 72, 1 May 2016, Pages 30-42
نویسندگان
, , ,