摘要
Since uncontrolled growth of malignant masses introduces uneven shape irregularities and spiculations in the boundary, shape representing shift invariant features are essential to resolve the problem of discrimination. However, ambiguous nature of shape, size, margin, orientation of masses produces imprecise feature values. In this view, a new concept of extrema based feature characterization scheme is proposed for capturing radiating nature of mass morphology. Computation of extrema patterns needs only few algorithmic steps. Beside this, present study employs an automated enhancement procedure to improve the classification accuracy. Experimental results show that extrema characterization reduces the feature redundancy to produce high efficiency in reasonably low time.
学科与关键词
- An automatic BI-RADS description of mammographic masses by fusing multiresolution featuresNarvaez, Fabian; Diaz, Gloria; Poveda, Cesar; Romero, Eduardo*EXPERT SYSTEMS WITH APPLICATIONS, 2017, 74: 82-95.被引 11
- Mammogram classification using sparse-ROI: A novel representation to arbitrary shaped massesKanadam, Karteeka Pavan*; Chereddy, Srinivasa RaoEXPERT SYSTEMS WITH APPLICATIONS, 2016, 57: 204-213.被引 16
- Mammogram retrieval on similar mass lesionsWei, Chia-Hung*; Chen, Sherry Y.; Liu, XiaohuiComputer Methods and Programs in Biomedicine, 2012, 106(3): 234-248.被引 45
