Automated Biomarker Extraction from Medical Images

Medical images are widely available to support clinical decision making. This book describes how computer-aided feature extraction techniques can help to obtain biomarkers from medical images that relate to a certain clinical state or the disease progression of individual patients. Different techniques are presented for the example of biomarkers for Alzheimer's disease extracted from brain magnetic resonance imaging (MRI). The first part of the book deals with the automated measurement of traditional morphometric biomarkers. In particular, methods to determine hippocampal volume and hippocampal atrophy are described. In the second part, more data-driven approaches for biomarker extraction based on novel machine-learning techniques are discussed. One application area for the described biomarkers is the selection of subjects in large-scale clinical trials. Other areas include the incorporation of biomarkers into computer aided diagnosis (CAD) systems and the measurement of effects in clinical trials. All described methods are evaluated on a large MRI study on Alzheimer's disease.