Brion Dolenko

Near-optimal region selection for feature space reduction novel preprocessing methods for classifying MR spectra

We introduce a global feature extraction method specifically designed to preprocess magnetic resonance spectra of biomedical origin. Such preprocessing is essential for the accurate and reliable classification of diseases or disease stages manifest in the spectra. The new method is genetic algorithm‐guided. It is compared with our enhanced version of the standard forward selection algorithm. Both seek and select optimal spectral subregions. These subregions necessarily retain spectral information, thus aiding the eventual identification of the biochemistry of disease presence and progression. The power of the methods is demonstrated on two biomedical examples: the discrimination between meningioma and astrocytoma in brain tissue biopsies, and the classification of colorectal biopsies into normal and tumour classes. Both preprocessing methods lead to classification accuracies over 97% for the two examples.

Tolerance to analog hardware of on-chip learning in backpropagation networks

In this paper we present results of simulations performed assuming both forward and backward computation are done on-chip using analog components. Aspects of analog hardware studied are component variability, limited voltage ranges, components (multipliers) that only approximate the computations in the backpropagation algorithm, and capacitive weight decay. It is shown that backpropagation networks can learn to compensate for all these shortcomings of analog circuits except for zero offsets, and the latter are correctable with minor circuit complications. Variability in multiplier gains is not a problem, and learning is still possible despite limited voltage ranges and function approximations. Fixed component variation from fabrication is shown to be less detrimental to learning than component variation due to noise. Weight decay is tolerable provided it is sufficiently small, which implies frequent refreshing by rehearsal on the training data or modest cooling of the circuits. The former approach allows for learning nonstationary problem sets.

Rapid Identification of Candida Species by Using Nuclear Magnetic Resonance Spectroscopy and a Statistical Classification Strategy

ABSTRACT Nuclear magnetic resonance (NMR) spectra were acquired from suspensions of clinically important yeast species of the genus Candida to characterize the relationship between metabolite profiles and species identification. Major metabolites were identified by using two-dimensional correlation NMR spectroscopy. One-dimensional proton NMR spectra were analyzed by using a staged statistical classification strategy. Analysis of NMR spectra from 442 isolates of Candida albicans, C. glabrata, C. krusei, C. parapsilosis, and C. tropicalis resulted in rapid, accurate identification when compared with conventional and DNA-based identification. Spectral regions used for the classification of the five yeast species revealed species-specific differences in relative amounts of lipids, trehalose, polyols, and other metabolites. Isolates of C. parapsilosis and C. glabrata with unusual PCR fingerprinting patterns also generated atypical NMR spectra, suggesting the possibility of intraspecies discontinuity. We conclude that NMR spectroscopy combined with a statistical classification strategy is a rapid, nondestructive, and potentially valuable method for identification and chemotaxonomic characterization that may be broadly applicable to fungi and other microorganisms.

Metabolomic biomarkers in a model of asthma exacerbation: urine nuclear magnetic resonance.

RATIONALE Airway obstruction in patients with asthma is associated with airway dysfunction and inflammation. Objective measurements including sputum analysis can guide therapy, but this is often not possible in typical clinical settings. Metabolomics is the study of molecules generated by metabolic pathways. We hypothesize that airway dysfunction and inflammation in an animal model of asthma would produce unique patterns of urine metabolites measured by multivariate statistical analysis of high-resolution proton nuclear magnetic resonance ((1)H NMR) spectroscopy data. OBJECTIVES To develop a noninvasive means of monitoring asthma status by metabolomics and urine sampling. METHODS Five groups of guinea pigs were studied: control, control treated with dexamethasone, sensitized (ovalbumin, administered intraperitoneally), sensitized and challenged (ovalbumin, administered intraperitoneally, plus ovalbumin aerosol), and sensitized-challenged with dexamethasone. Airway hyperreactivity (AHR) to histamine (administered intravenously) and inflammation were measured. Multivariate statistical analysis of NMR spectra based on a library of known urine metabolites was performed by partial least-squares discriminant analysis. In addition, the raw NMR spectra exported as xy-trace data underwent linear discriminant analysis. MEASUREMENTS AND MAIN RESULTS Challenged guinea pigs developed AHR and increased inflammation compared with sensitized or control animals. Dexamethasone significantly improved AHR. Using concentration differences in metabolites, partial least-squares discriminant analysis could discriminate challenged animals with 90% accuracy. Using only three or four regions of the NMR spectra, linear discriminant analysis-based classification demonstrated 80-90% separation of the animal groups. CONCLUSIONS Urine metabolites correlate with airway dysfunction in an asthma model. Urine NMR analysis is a promising, noninvasive technique for monitoring asthma in humans.

Magnetic resonance spectroscopy detects biochemical changes in the brain associated with chronic low back pain: a preliminary report

Magnetic resonance (MR) spectroscopy is a noninvasive technique that can be used to detect and measure the concentration of metabolites and neurotransmitters in the brain and other organs. We used in vivo 1H MR spectroscopy in subjects with low back pain compared with control subjects to detect alterations in biochemistry in three brain regions associated with pain processing. A pattern recognition approach was used to determine whether it was possible to discriminate accurately subjects with low back pain from control subjects based on MR spectroscopy. MR spectra were obtained from the prefrontal cortex, anterior cingulate cortex, and thalamus of 32 subjects with low back pain and 33 control subjects without pain. Spectra were analyzed and compared between groups using a pattern recognition method (Statistical Classification Strategy). Using this approach, it was possible to discriminate between subjects with low back pain and control subjects with accuracies of 100%, 99%, and 97% using spectra obtained from the anterior cingulate cortex, thalamus, and prefrontal cortex, respectively. These results demonstrate that MR spectroscopy, in combination with an appropriate pattern recognition approach, is able to detect brain biochemical changes associated with chronic pain with a high degree of accuracy.

Disease pattern recognition testing for rheumatoid arthritis using infrared spectra of human serum

BACKGROUND In view of the importance of the diagnosis of rheumatoid arthritis, a novel diagnostic method based on spectroscopic pattern recognition in combination with laboratory parameters such as the rheumatoid factor is described in the paper. Results of a diagnostic study of rheumatoid arthritis employing this method are presented. METHOD The method uses classification of infrared (IR) spectra of serum samples by means of discriminant analysis. The spectroscopic pattern yielding the highest discriminatory power is found through a complex optimization procedure. In the study, IR spectra of 384 serum samples have been analyzed in this fashion with the objective of differentiating between rheumatoid arthritis and healthy subjects. In addition, the method integrates results from the classification with levels of the rheumatoid factor in the sample by optimized classifier weighting, in order to enhance classification accuracy, i.e. sensitivity and specificity. RESULTS In independent validation, sensitivity and specificity of 84% and 88%, respectively, have been obtained purely on the basis of spectra classification employing a classifier designed specifically to provide robustness. Sensitivity and specificity are improved by 1% and 6%, respectively, upon inclusion of rheumatoid factor levels. Results for less robust methods are also presented and compared to the above numbers. CONCLUSION The discrimination between RA and healthy by means of the pattern recognition approach presented here is feasible for IR spectra of serum samples. The method is sufficiently robust to be used in a clinical setting. A particular advantage of the method is its potential use in RA diagnosis at early stages of the disease.

Detecting colorectal cancer by 1H magnetic resonance spectroscopy of fecal extracts

Colorectal cancer is one of the most common cancers in the western world. Its early detection has been found to improve the prognosis of the patient, providing a wide window of opportunity for successful therapeutic interventions. However, current diagnostic techniques all have some limitations; there is a need to develop a better technique for routine screening purposes. We present a new methodology based on magnetic resonance spectroscopy of fecal extracts for the non‐invasive detection of colorectal cancer. Five hundred twenty‐three human subjects (412 with no colonic neoplasia and 111 with colorectal cancer, who were scheduled for colonoscopy or surgery) were recruited to donate a single sample of stool. One‐dimensional 1H magnetic resonance spectroscopy (MRS) experiments were performed on the supernatant of aqueous dispersions of the stool samples. Using a statistical classification strategy, several multivariate classifiers were developed. Applying the preprocessing, feature selection and classifier development stages of the Statistical Classification Strategy led to ∼87% average balanced sensitivity and specificity for both training and monitoring sets, improving to ∼92% when only crisp results, i.e. class assignment probabilities ≥75%, are considered. These results indicate that 1H magnetic resonance spectroscopy of human fecal extracts, combined with appropriate data analysis methodology, has the potential to detect colorectal neoplasia accurately and reliably, and could be a useful addition to the current screening tools. Copyright

Mapping high-dimensional data onto a relative distance plane: an exact method for visualizing and characterizing high-dimensional patterns

We introduce a distance (similarity)-based mapping for the visualization of high-dimensional patterns and their relative relationships. The mapping preserves exactly the original distances between points with respect to any two reference patterns in a special two-dimensional coordinate system, the relative distance plane (RDP). As only a single calculation of a distance matrix is required, this method is computationally efficient, an essential requirement for any exploratory data analysis. The data visualization afforded by this representation permits a rapid assessment of class pattern distributions. In particular, we can determine with a simple statistical test whether both training and validation sets of a 2-class, high-dimensional dataset derive from the same class distributions. We can explore any dataset in detail by identifying the subset of reference pairs whose members belong to different classes, cycling through this subset, and for each pair, mapping the remaining patterns. These multiple viewpoints facilitate the identification and confirmation of outliers. We demonstrate the effectiveness of this method on several complex biomedical datasets. Because of its efficiency, effectiveness, and versatility, one may use the RDP representation as an initial, data mining exploration that precedes classification by some classifier. Once final enhancements to the RDP mapping software are completed, we plan to make it freely available to researchers.

Determination of Grade and Receptor Status from the Primary Breast Lesion by Magnetic Resonance Spectroscopy

Magnetic resonance spectra (MRS) from fine needle aspiration biopsies (FNAB) from primary breast lesions were analysed using a pattern recognition method, Statistical Classification Strategy, to assess tumor grade and oestrogen receptor (ER) and progesterone receptor (PgR) status. Grade 1 and 2 breast cancers were separated from grade 3 cancers with a sensitivity and specificity of 96% and 95%, respectively. The ER status was predicted with a sensitivity of 91% and a specificity of 90%, and the PgR status with a sensitivity of 91% and a specificity of 86%. These classifiers provide rapid and reliable, computerized information and may offer an objective method for determining these prognostic indicators simultaneously with the diagnosis of primary pathology and lymph node involvement.

Cancer diagnosis by infrared spectroscopy: methodological aspects

IR spectroscopy is proving to be a powerful tool for the study and diagnosis of cancer. The application of IR spectroscopy to the analysis of cultured tumor cells and grading of breast cancer sections is outlined. Potential sources of error in spectral interpretation due to variations in sample histology and artifacts associated with sample storage and preparation are discussed. The application of statistical techniques to assess differences between spectra and to non-subjectively classify spectra is demonstrated.