Characterization of mammographic masses using a gradient-based segmentation algorithm and a neural classifier

TL;DR

This study presents a gradient-based segmentation algorithm and neural classifier for mammographic mass characterization, demonstrating effective, dataset-independent segmentation and high classification accuracy using selected features.

Contribution

The paper introduces a parameter-free, gradient-based segmentation method and a feature selection process that enhances neural classifier performance for mammogram analysis.

Findings

Segmentation algorithm works efficiently on both benign and malignant masses.

12 features out of 16 are sufficient for optimal classification.

Achieved ROC area of approximately 0.78 to 0.81 depending on segmentation quality.

Abstract

The computer-aided diagnosis system we developed for the mass characterization is mainly based on a segmentation algorithm and on the neural classification of several features computed on the segmented mass. Mass segmentation plays a key role in most computerized systems. Our technique is a gradient-based one, showing the main characteristic that no free parameters have been evaluated on the dataset used in this analysis, thus it can directly be applied to datasets acquired in different conditions without any ad-hoc modification. A dataset of 226 masses (109 malignant and 117 benign) has been used in this study. The segmentation algorithm works with a comparable efficiency both on malignant and benign masses. Sixteen features based on shape, size and intensity of the segmented masses are analyzed by a multi-layered perceptron neural network. A feature selection procedure has been carried out on the basis of the feature discriminating power and of the linear correlations interplaying among them. The comparison of the areas under the ROC curves obtained by varying the number of features to be classified has shown that 12 selected features out of the 16 computed ones are powerful enough to achieve the best classifier performances. The radiologist assigned the segmented masses to three different categories: correctly-, acceptably- and non-acceptably-segmented masses. We initially estimated the area under ROC curve only on the first category of segmented masses (the 88.5% of the dataset), then extending the dataset to the second sub-class (reaching the 97.8% of the dataset) and finally to the whole dataset, obtaining Az = 0.805+-0.030, 0.787+-0.024 and 0.780+-0.023, respectively.