Analysis of shape features for lesion classification in breast ultrasound images
Dina Arifatul Khusna, Hanung Adi Nugroho, Indah Soesanti · AIP conference proceedings · 2016
Classification accuracy in image processing is highly related to previous steps such as feature extraction and feature selection.In breast ultrasound imaging, lesion classification is performed based on several criteria, including edge regularity.This research aims at implementing and evaluating some shape features for edge regularity classification.Breast lesion is divided into two classes based on edge regularity : regular and non-regular.Several shape features is implemented and evaluated by means of average value and standard deviation.The distance of average value from corresponding feature in each class is evaluated and concluded the best feature ability to distinguish breast ultrasound lesion's edge regularity.Aspect ratio is defined as ratio of the object height to its own width.Aspect Ratio is not rotation invariant.In [4], Aspect Ratio is defined as ratio of Major Axis to Minor Axis of object.