SEMI-SUPERVISED ANALYSIS OF HUMAN BRAIN TUMOURS FROM PARTIALLY LABELED MRS INFORMATION,USING MANIFOLD LEARNING MODELS

摘要

Medical diagnosis can often be understood as a classification problem. In oncology, this typically involvesndifferentiating between tumour types and grades, or some type of discrete outcome prediction. From thenviewpoint of computer-based medical decision support, this classification requires the availability of accuratendiagnoses of past cases as training target examples. The availability of such labeled databases isnscarce in most areas of oncology, and especially so in neuro-oncology. In such context, semi-supervisednlearning oriented towards classification can be a sensible data modeling choice. In this study, semisupervisednvariants of Generative Topographic Mapping, a model of the manifold learning family, arenapplied to two neuro-oncology problems: the diagnostic discrimination between different brain tumournpathologies, and the prediction of outcomes for a specific type of aggressive brain tumours. Their performancencompared favorably with those of the alternative Laplacian Eigenmaps and Semi-SupervisednSVM for Manifold Learning models in most of the experiments