Search

Quantitative imaging biomarkers identification has become a powerful tool for predictive diagnosis given increasingly available clinical imaging data. In parallel, molecular profiles have been well documented in non-small cell lung cancers (NSCLCs). However, there has been limited studies on leveraging the two major sources for improving lung cancer computer-aided diagnosis. In this paper, we investigate the problem of predicting molecular profiles with CT imaging arrays in NSCLC. In particular, we formulate a discriminative convolutional neural network to learn deep features for predicting epidermal growth factor receptor (EGFR) mutation states that are associated with cancer cell growth. We evaluated our approach on two independent datasets including a discovery set with 595 patients (Datset1) and a validation set with 89 patients (Dataset2). Extensive experimental results demonstrated that the learned CNN-based features are effective in predicting EGFR mutation states (AUC=0.828, ACC=76.16%) on Dataset1, and it further demonstrated generalized predictive performance (AUC=0.668, ACC=67.55%) on Dataset2.

Lossy conductors are not perfectly electric conductors and their finite conductivity needs to be carefully accounted for in the accurate solution of electromagnetic problems. Traditionally, surface integral equations (SIEs) are used to approximately describe the problems, but we use volume integral equations (VIEs) to exactly formulate the problems by treating the lossy conductors as dielectric-like objects. The VIEs are solved by a point-matching scheme which allows the use of JM-formulation, yielding some desirable merits. Typical numerical examples are presented to demonstrate the scheme and its robustness has been verified.

Tunability analysis based on electromagnetic modeling and simulation is performed for a miniaturized patch antenna loaded with self-biased magnetic films. The antenna includes a lossy dielectric substrate and layered thin-ferrite-film superstrate above microstrip. The structure is of multiscale and multimaterial feature and the system matrix may not be well-conditioned if the problem is described by surface integral equations (SIEs) in the integral equation approach. We use volume-surface integral equations (VSIEs) to formulate the problem and the VSIEs are solved by the method of moments (MoM) with a different implementation scheme. A numerical example is presented to illustrate the approach.