Yrjö Hiltunen

Automated classification of human brain tumours by neural network analysis using in vivo 1H magnetic resonance spectroscopic metabolite phenotypes

We present a novel method to integrate in vivo nuclear magnetic resonance spectroscopy (MRS) information into the clinical diagnosis of brain tumours. Water-suppressed 1H MRS data were collected from 33 patients with brain tumours and 28 healthy controls in vivo. The data were treated in the time domain for removal of residual water and a region from the frequency domain (from 3.4 to 0.3 p.p.m.) together with the unsuppressed water signal were used as inputs for artificial neural network (ANN) analysis. The ANN distinguished tumour and normal tissue in each case and was able to classify benign and malignant gliomas as well as other brain tumours to match histology in a clinically useful manner with an accuracy of 82%. Thus the present data indicate existence of tumour tissue-specific metabolite phenotypes that can be detected by in vivo 1H MRS. We believe that a user-independent ANN analysis may provide an alternative method for tumour classification in clinical practice.

Diagnostic assessment of brain tumours and non-neoplastic brain disorders in vivo using proton nuclear magnetic resonance spectroscopy and artificial neural networks.

Purpose: Experiments were carried out to assess the potential of artificial neural network (ANN) analysis in the differential diagnosis of brain tumours (low- and high-grade gliomas) from non-neoplastic focal brain lesions (tuberculomas and abscesses), using proton magnetic resonance spectroscopy (1H MRS) as input data. Methods: Single-voxel stimulated echo acquisition mode (STEAM) (echo time of 20 ms) spectra were acquired from 138 subjects including 15 with low-grade gliomas, 47 with high-grade gliomas, 18 with tuberculomas, 18 with abscesses and 40 healthy controls. Two neural networks were constructed using the spectral points from 0.6 to 3.4 parts per million. In the first network construction, the ANN had to differentiate between tumours from infections, while the second network had to differentiate between all five histological classes. Results: ANN analysis gave a histologically correct diagnosis for low- and high-grade gliomas with an accuracy of 73% and 98% respectively. None of the 62 tumours was diagnosed as an infectious lesion. Among the non-neoplastic lesions, ANN classification was correct in 89% of tuberculomas and in 83% of brain abscesses. The specificity of ANN diagnosis was 98%, 92%, 99%, and 100% for low-grade gliomas, high-grade gliomas, tuberculomas and abscesses respectively. Conclusion: The present data show the clinical utility of non-invasive 1H MRS by automated ANN analysis in the diagnosis of tumour and non-tumour cerebral disorders.

Solute molecular structure determination by N.M.R.

The apparent variation of the HCH bond angle in methyl iodide and of the ratios of internuclear H-H distances in benzene was studied in several thermotropic liquid crystals and in their mixtures. In each case, 13C-methane was used as an internal ‘deformation reference’ and also as a 1H chemical shift reference for the determination of chemical shift anisotropies. The results show that structures in agreement with microwave studies are obtained in liquid crystal mixtures in which the dipolar C-H coupling constant of methane vanishes. The 1H chemical shift anisotropy, ΔσH, of methyl iodide was determined by the gradient method in the mixture of two thermotropic liquid crystals, ZLI 1167 and Phase IV. The variation of the relative concentrations of the components led to a wide range of ΔσH values: from 1·7 p.p.m. to 7·7 p.p.m. when going from Phase IV to ZLI 1167. Also a new means to determine 1H chemical shift anisotropies is proposed. For methyl iodide and benzene, this method predicts the ΔσH values of 8·...

The r ∞-structure of thiophene in various liquid crystals with 13C-methane as an internal reference

The r ∞-structure of thiophene was studied in thermotropic liquid crystals and in their mixtures. In each case, 13C-enriched methane was added to the sample to serve as an internal ‘deformation reference’. The results indicate that the structure of thiophene is very sensitive to the liquid crystal environment. The structure obtained for thiophene in such liquid crystal mixtures in which the D CH of methane is vanishingly small is in agreement with the microwave structure. The best agreement, however, is obtained in the mixture of MBBA and ZLI 1167 where the D CH of methane is c. 2 Hz.

Application of self‐organizing maps for the detection and classification of human blood plasma lipoprotein lipid profiles on the basis of 1H NMR spectroscopy data

Efficient and relevant classification of clinical findings, i.e. diagnostic decision making, poses a major challenge in medicine. In relation to biomedical NMR spectroscopy the problem of classification is often accompanied by complex, heavily overlapping information. Self‐organizing map (SOM) analysis has been successfully applied in many areas of research and was thus also considered as a potential tool for NMR data analysis. In this paper we demonstrate how SOM analysis can be used for automated NMR data classification. Our goal was analysis of plasma lipoprotein lipids, a complex but biochemically well understood and specified system. The results illustrate that clinically relevant lipid classifications can be obtained from the SOM analysis of 1H NMR spectral information alone. The resulting maps were calibrated using independent biochemical lipid analyses and were found to produce excellent clustering of the plasma samples into clinically useful groups: normal, type IIa, IIb and IV hyperlipidaemias. In addition to this traditional classification, we also present results from SOM analysis in which the reference vectors of the map were calibrated for plasma total cholesterol and triglycerides and high and low density lipoprotein C; the plasma lipid parameters that are currently considered as the most useful indicators of coronary heart disease risk. In all, the present results indicate that SOM analysis can cope well with complex NMR spectral information and is thus likely to have an independent role in the area of biomedical NMR data analysis.

Quality-oriented optimization of wave soldering process by using self-organizing maps

In this paper, the optimal process parameters of a wave soldering process were defined. The optimization was performed in respect to soldering quality by minimizing a cost function describing the total repairing cost of a wave-soldered printed circuit board (PCB). The data analysis stages were as follows. First, the process data were coded into inputs for a self-organizing map (SOM). Next, a function for the repairing cost was constructed and used to find the optimal map neurons. At the last phase, the optimal parameters were approximated on the basis of the reference vectors of the optimal neurons. The results showed clearly potential in the optimization of the wave soldering process, especially in the visualization of the optimal process conditions. Therefore, it would be useful to exploit the method more widely in the electronics industry.

Correlating tissue outcome with quantitative multiparametric MRI of acute cerebral ischemia in rats

Predicting tissue outcome remains a challenge for stroke magnetic resonance imaging (MRI). In this study, we have acquired multiparametric MRI data sets (including absolute T1, T2, diffusion, T1ρ using continuous wave and adiabatic pulse approaches, cerebral blood flow (CBF), and amide proton transfer ratio (APTR) images) during and after 65 mins of middle cerebral artery occlusion (MCAo) in rats. The MRI scans were repeated 24 h after MCAo, when the animals were killed for quantitative histology. Magnetic resonance imaging parameters acquired at three acute time points were correlated with regionally matching cell count at 24 h. The results emphasize differences in the temporal profile of individual MRI contrasts during MCAo and especially during early reperfusion, and suggest that complementary information from CBF and tissue damage can be obtained with appropriate MRI contrasts. The data show that by using three to four MRI parameters, sensitive to both hemodynamic changes and different aspects of parenchymal changes, the fate of the tissue can be predicted with increased correlation compared with single-parameter techniques. Combined multiparametric MRI data and multiparametric analysis may provide an excellent tool for preclinical testing of new treatments and also has the potential to facilitate decision-making in the management of acute stroke patients.

Application of self-organizing maps in analysis of wave soldering process

This paper presents an overview of a data analysis method based on self-organizing maps (SOM), a well-known unsupervised neural network learning algorithm, which was applied to a lead-free wave soldering process. The aim of the study was to determine whether the neural network modeling method could be a useful and time-saving way to analyze data from a discrete manufacturing process, such as wave soldering, which is a widely used technique in the electronics industry to solder components on printed circuit boards. The data variables were mostly various process parameters, but also some solder defect numbers were present in the data as a measure of the product quality. The data analysis procedure went as follows. At first, the process data were modeled using the SOM-algorithm. Next, the neuron reference vectors of the formed map were clustered to reveal the desired dominating elements of each territory of the map. At the final stage, the clusters were utilized as sub-models to indicate variable dependencies in these sub-models. The results show that the method presented here can be a good way to analyze this type of process data because interesting interactions between certain process parameters and solder defects were found by means of this data-driven modeling method.

An analysis of the NMR data for methyl iodide and methyl fluoride in liquid-crystal solvents allowing for the correlation between vibration and rotation

Abstract The dipolar coupling constants of methyl iodide and methyl fluoride, oriented in various types of liquid crystals, are analyzed by a method allowing for the correlation between the molecular reorientational and vibrational motion. Unlike the conventional technique, the new method is capable of explaining the data in very good agreement with the experimental values. The results suggest that the torques acting on the CH bonds of methane, methyl fluoride, and methyl iodide, dissolved into the same sample tube, are directly proportional to each other with ratios 1:2.0:2.4, respectively. When this information is incorporated in the analysis, the molecular geometries are determined very accurately, with the results α (CH 3 I) = 111.576 ± 0.010°, α (CH 3 F) = 110.443 ± 0.010°, and r CF / r CH = 1.2712 ± 0.0005 (α is the HCH bond angle and r ij is the bond length). These values compare excellently with the microwave results. Also, interesting information on the intermolecular forces is obtained because the torques acting on all the bonds of the molecules are determined.

Isotope and temperature effects on the 13C and 77Se nuclear shielding in carbon diselenide

A comprehensive theoretical and experimental study of the 13C and 77Se nuclear magnetic shieldings and their rovibrational corrections in carbon diselenide (CSe2) has been undertaken. The 13C and 77Se shielding tensors as well as all their first and second derivatives with respect to the internal displacement coordinates of the molecule have been calculated by several first principles gauge-including atomic orbital (GIAO) methods. Hartree-Fock (HF), multiconfiguration Hartree-Fock (MCHF), and density-functional (DFT) theories have been compared, the latter both in the local density approximation (LDA) and by using two gradient corrected exchange-correlation functionals. The shielding derivatives calculated with MCHF and DFT are very much smaller in magnitude than the derivatives obtained by using HF, being in reasonable mutual agreement. By using the theoretical shielding derivatives and the cubic anharmonic force constants calculated within LDA, together with an experimental harmonic force field, all the...