Juergen Backhaus

Infrared spectroscopy: a new diagnostic tool in Alzheimer disease.

Biological markers play an evolving role in the diagnosis of Alzheimer disease (AD). We compare conventional measurements of cerebrospinal fluid (CSF) tau and beta-amyloid(1-42) proteins to a novel approach - Fourier transformed infrared (FT-IR) spectroscopy - a simple technique derived from chemical and physical sciences that characterizes intramolecular bonds. For automatic diagnostic analysis, we developed an artificial neural network (ANN). We examined 71 patients with a clinical diagnosis of AD and 66 controls. beta-Amyloid(1-42) was decreased (sensitivity 80% and specificity 78%); tau was elevated (sensitivity 76% and specificity 88%) in CSF of AD patients. The combined tau/beta-amyloid(1-42) quotient was able to distinguish healthy from diseased subjects with 99% sensitivity and 86% specificity. The ANN could separate FT-IR spectroscopy data with 88.5% sensitivity and 80% specificity. FT-IR spectroscopy proved to be cost-effective and simple to perform. Diagnostic sensitivity and specificity is in the range of CSF tau and beta-amyloid(1-42) protein analysis. Larger sample numbers for ANN training and validation could increase diagnostic accuracy and thus prove to be a useful screening tool.

Near-infrared fiber optic spectroscopy as a novel diagnostic tool for the detection of pancreatic cancer

We have investigated the application of near-infrared (NIR) fiber-optic spectroscopy for the diagnosis of pancreatic cancer. Cluster analysis of the Fourier transformed near-infrared (FTNIR) fiber-optic spectra of surgically resected pancreatic tumor tissues allowed discrimination of tumor from normal tissue with high sensitivity and specificity. The sensitivity of the method using spectral information of the CH stretching first overtone region (5951-5608 cm(-1)) was 83.3% with a specificity of 83.3%. Based on the CH stretching second overtone region (8605-7938 cm(-1)) we could achieve a sensitivity of 88.9% and specificity of 72.2%. These findings suggest that NIR spectroscopy offers the potential for minimally invasive in-vivo diagnosis of pancreatic cancer.

A factorization method for the classification of infrared spectra

BackgroundBioinformatics data analysis often deals with additive mixtures of signals for which only class labels are known. Then, the overall goal is to estimate class related signals for data mining purposes. A convenient application is metabolic monitoring of patients using infrared spectroscopy. Within an infrared spectrum each single compound contributes quantitatively to the measurement.ResultsIn this work, we propose a novel factorization technique for additive signal factorization that allows learning from classified samples. We define a composed loss function for this task and analytically derive a closed form equation such that training a model reduces to searching for an optimal threshold vector. Our experiments, carried out on synthetic and clinical data, show a sensitivity of up to 0.958 and specificity of up to 0.841 for a 15-class problem of disease classification. Using class and regression information in parallel, our algorithm outperforms linear SVM for training cases having many classes and few data.ConclusionsThe presented factorization method provides a simple and generative model and, therefore, represents a first step towards predictive factorization methods.