Anne Rosemary Tate

Towards a method for automated classification of 1H MRS spectra from brain tumours

Recent studies have shown that MRS can substantially improve the non‐invasive categorization of human brain tumours. However, in order for MRS to be used routinely by clinicians, it will be necessary to develop reliable automated classification methods that can be fully validated. This paper is in two parts: the first part reviews the progress that has been made towards this goal, together with the problems that are involved in the design of automated methods to process and classify the spectra. The second part describes the development of a simple prototype system for classifying 1H single voxel spectra, obtained at an echo time (TE) of 135 ms, of the four most common types of brain tumour (meningioma (MM), astrocytic (AST), oligodendroglioma (OD) and metastasis (ME)) and cysts. This system was developed in two stages: firstly, an initial database of spectra was used to develop a prototype classifier, based on a linear discriminant analysis (LDA) of selected data points. Secondly, this classifier was tested on an independent test set of 15 newly acquired spectra, and the system was refined on the basis of these results. The system correctly classified all the non‐astrocytic tumours. However, the results for the the astrocytic group were poorer (between 55 and 100%, depending on the binary comparison). Approximately 50% of high grade astrocytoma (glioblastoma) spectra in our data base showed very little lipid signal, which may account for thepoorer results for this class. Consequently, for the refined system, the astrocytomas were subdivided into two subgroups for comparison against other tumour classes: those with high lipid content and those without.

Developing quality scores for electronic health records for clinical research: a study using the general practice research database

The General Practice Research Database (GPRD) is a collection of anonymised patient records obtained from UK general practices. Data are representative of approximately 8% of the UK population and are collected mainly for research purposes, which include assessing risk factors for disease, evaluating the side effects of drugs and comparing the effectiveness of different drugs. The data are used internationally by academics, governments and the pharmaceutical industry. As research findings arising from GPRD data may have potential public health and safety implications it is crucial importance that the data collected is of high quality. Data quality may vary within and between practices and may depend on the time of data collection. Although the GPRDs established framework for assessing data quality is comprehensive, it does not allow a systematic review of individual practice data quality markers. We are developing a framework for further improvement of existing methods of data quality assessment. We shall extend a set of current quality measures for each practice and, using statistical pattern recognition techniques, shall develop algorithms that will combine these measures into a smaller number of meaningful quality scores which will reflect different aspects of data quality and can be measured over time. We report the aims and rationale of the study and preliminary results.

Unsupervised Feature Extraction of in vivo Magnetic Resonance Spectra of Brain Tumours Using Independent Component Analysis

We present a method for automatically decomposing magnetic resonance (MR) spectra of different types of human brain tumours into components which directly reflect their different chemical compositions. The automatic analysis of in vivo MR spectra can be problematic due to their large dimensionality and the low signal to noise ratio. Principal Component Analysis allows an economic representation of the data but the extracted components themselves may bear little relationship to the underlying metabolites represented by the spectra. The Principal Components can be rotated in order to make them more meaningful but this requires expertise to decide on the transformation. In this study, we use Independent Component Analysis and show that this technique can overcome these two drawbacks and provide meaningful and representative components without requiring prior knowledge.