Abstract

Malignant lesions have a high tendency to invade their surrounding environment compared to benign ones. This paper proposes a dynamic lesion model and explores the 2nd order derivatives at each image voxel, which reflect the rate of change of image intensity, as a quantitative measure of the tendency. The 2nd order derivatives at each image voxel are usually represented by the Hessian matrix, but it is difficult to quantify a matrix field (or image) through the lesion space as a measure of the tendency. We conjecture that the three eigenvalues contain important information of the Hessian matrix and are chosen as the surrogate representation of the Hessian matrix. By treating the three eigenvalues as a vector, called Hessian vector, which is defined in a local coordinate formed by three orthogonal Hessian eigenvectors and further adapting the gray level occurrence computing method to extract the vector texture descriptors (or measures) from the Hessian vector, a quantitative presentation for the dynamic lesion model is completed. The vector texture descriptors were applied to differentiate malignant from benign lesions from two pathologically proven datasets: colon polyps and lung nodules. The classification results not only outperform four state-of-the-art methods but also three radiologist experts.

Highlights

  • Malignant lesions have a high tendency to invade their surrounding environment compared to benign ones, the lesion growth rate has been widely used as a figure of merit (FOM) for computer-aided diagnosis (CADx) of the lesion, for example, of colon p­ olyps[1,2] and lung ­nodules[3,4]

  • VCM and vector texture features (VTFs) calculations of the pulmonary nodules share the same steps as the VCM and VTF calculations of the colon polyps, i.e. going through the 13 independent offsets or directions and performing the digitalization scheme to obtain the 235 different combinations of Qa, Qb, and Qc for the corresponding 235 features

  • A dynamic lesion model was proposed to consider the clinical observation that malignant lesions have a high tendency to invade their surrounding environment compared to benign ones

Read more

Summary

Introduction

Malignant lesions have a high tendency to invade their surrounding environment compared to benign ones. The efforts have been further devoted to expanding the CNN architectures to learn the medical images and the textures and ­more[16,17] To our knowledge, both the above research endowers for sophisticated feature extraction and classification operations and the recent CNN-based deep machine learning architectures have not explicitly considered the clinical observations that malignant lesions have high tendency to invade their surrounding environment compared to benign o­ nes[18,19,20,21,22,23]. This paper proposes a dynamic lesion model and explores the feasibility of explicitly considering the high invading tendency for the task of CADx of lesions

Methods
Results
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call