Meningioma classification using an adaptive discriminant wavelet packet transform

Meningioma subtypes classification is a real world problem from the domain of histological image analysis that requires new methods for its resolution. Computerised histopathology presents a whole new set of problems and introduces new challenges in image classification. High intra-class variation and low inter-class differences in textures is often an issue in histological image analysis problems such as Meningioma subtypes classification. In this thesis, we present an adaptive wavelets based technique that adapts to the variation in the texture of meningioma samples and provides high classification accuracy results. The technique provides a mechanism for attaining an image representation consisting of various spatial frequency resolutions that represent the image and are referred to as subbands. Each subband provides different information pertaining to the texture in the image sample. Our novel method, the Adaptive Discriminant Wavelet Packet Transform (ADWPT), provides a means for selecting the most useful subbands and hence, achieves feature selection. It also provides a mechanism for ranking features based upon the discrimination power of a subband. The more discriminant a subband, the better it is for classification. The results show that high classification accuracies are obtained by selecting subbands with high discrimination power. Moreover, subbands that are more stable i.e. have a higher probability of being selected provide better classification accuracies. Stability and discrimination power have been shown to have a direct relationship with classification accuracy. Hence, ADWPT acquires a subset of subbands that provide a highly discriminant and robust set of features for Meningioma subtype classification. Classification accuracies obtained are greater than 90% for most Meningioma subtypes. Consequently, ADWPT is a robust and adaptive technique which enables it to overcome the issue of high intra-class variation by statistically selecting the most useful subbands for meningioma subtype classification. It overcomes the issue of low inter-class variation by adapting to texture samples and extracting the subbands that are best for differentiating between the various meningioma subtype textures