tsta

Home / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Multi-transform diagnostic analysis based on gradient-based features for breast cancer detection in thermal imaging

Authors :

Ainaz Daneshdoust¹ Sedigheh Ghofrani² Mahdi Eslami³ Iman Ahanian⁴

1- دانشگاه آزاد اسلامی واحد تهران جنوب 2- دانشگاه آزاد اسلامی واحد تهران جنوب 3- دانشگاه آزاد اسلامی واحد علوم و تحقیقات 4- دانشگاه آزاد اسلامی واحد تهران جنوب

Keywords :

Breast Cancer،GradEn،2D-FFT،2D-EWT،2D-DOST،SIFT،HOG،GLTP

Abstract :

Breast cancer is among the deadliest cancers affecting women worldwide. Early detection plays a critical role in effective treatment and so survival rate. This study presents a framework for the classification of breast thermograms using three different transforms as the spatial-frequecy representation, and four gradient-based descriptors for feature extraction. The transforms are two-dimensional Fast Fourier Transform (2D-FFT), two-dimensional Empirical Wavelet Transform (2D-EWT), two-dimensional Discrete Orthonormal Stockwell Transform (2D-DOST) and the descriptors are Scale-Invariant Feature Transform (SIFT), Histogram of Oriented Gradients (HOG), Gradient Local Ternary Pattern (GLTP), Gradient Entropy (GradEn). Furthoremore, random forest (RF) is used as the classifier. Experimental results recommend using 2D-DOST as the spartial-frequency transform, and either SIFT or HOG as the feature extraction.