Lung Tissue Analysis Using Isotropic Polyharmonic B–Spline Wavelets

A texture classification system is described, based on isotropic polyharmonic B–spline wavelets that identify lung tissue patterns from high–resolution computed tomography (HRCT) images of patients affected with interstitial lung diseases (ILD). Along with several desirable properties for isotropic texture analysis, the nonseparable transform with a quincunx subsambling scheme allows a mean of 94.3% of correct matches among six lung tissue classes. A comparison with a classical dyadic transform suggests that the isotropic quincunx transform is preferable for lung tissue analysis. This is part of work on a tool for integrating visual and clinical features as diagnostic aid for emergency radiology.