A new non-invasive non-spectrophotometric method for measuring the blood concentration of analytes such as glucose has been developed. A plurality of broad spectrum filters (30A-30D) transmit distinguishably coded by choppers (20A-20D) beams of radiation in overlapping portions of the spectrum to the sample. Radiation reflected or transmitted by the sample is detected and decoded. LED's may be used instead of the broad spectrum radiation generating device and the filters. Further, a scanning interferometer can be used as the illuminating and coding device. In a preferred mode, congruent illumination is utilized. The coded signals are analyzed by analogy to colorimetry and visual processing and can be converted into concentration measurements.

Non-invasive non-spectrophotometric infrared measurement of blood analyte concentrations

Improvements in non-invasive detection methods for glucose and other constituents of interest in a sample have been developed. The apparatus and methods of the invention provide an analog of color perception of human vision, preferably in the near infrared region, replacing spectrophotometers and narrow band sources used in other non-invasive near infrared detection methods. A plurality of detector units are used, each covering a broad and overlapping region of the detected spectrum, paralleling color perception and colorimetry. The improvements are primarily concerned with improving the signal-to-background (or noise) ratio such that the data stream is improved. These improvements use congruent sampling, comparison of different data streams from different sample portions or filter sets, using an interrogation system with sufficient speed to allow testing of arterial blood, and using a filter with a spectral structure. In some circumstances, a neural net is used for analysis, allowing the system to learn. A novel method for background discrimination is also described.

Non-spectrophotometric measurement of analyte concentrations and optical properties of objects

The present invention provides a generally applicable apparatus and method for achieving measurements of a constituent in a sample. This is achieved by employing a detection means having a plurality of detectors responsive to radiation in a selected region of the spectrum, e.g., the infrared. In most embodiments, at least two of the detectors provide broad wavelength bandpass. If narrow bandpass sources or detectors are used, the information generated is processed in a manner similar to broadband information. The broad bandpass response of the detectors can be contrasted with the approach of classical spectrophotometry, in which the spectral response of the detectors is designed to be as narrow as feasible, and substantially narrower than the spectral features of the constituent or constituents of interest. The data is processed such that the contributions of known background constituents and scattering is eliminated prior to further processing, thereby yielding a better result in high background situations. The use of intensity processing rather than absorbance processing also allows the method and apparatus to be used in a variety of non-ideal situations.

Rapid non-invasive optical analysis using broad bandpass spectral processing

In this communication, we describe the application of Hadamard transform Raman microscopy to the imaging of edge plane microstructures in laser-modified highly ordered pyrolitic graphite (HOPG) electrodes. Hadamard transform Raman microscopy is a highly selective method of analysis which can image thermally sensitive materials at micron-scale spatial resolution. The technique is capable of imaging relatively weak Raman scatterers in only a few minutes. The current multichannel microscope generates images having 255 × 256 pixels. In addition, we characterize an imaging artifact present in our current system, as well as apply a first-order correction for the problem.

Hadamard transform Raman microscopy of laser-modified graphite electrodes

A movable two-dimensional (2D) Hadamard encoding mask is obtained and combined with conventional FT-IR spectrometers for use in both the mid- and near-infrared spectral regions. Chemical maps and spectra of individual pixels of the maps can be obtained from heterogeneous samples by using this combination of a moveable 2D Hadamard encoding mask and an FT-IR spectrometer. We call the procedure Hadamard transform/FT-IR spectrometry. Spectra of usable signal-to-noise ratio and reliable chemical maps are obtained in reasonable data acquisition and processing time.

Chemical Mapping in the Mid- and Near-IR Spectral Regions by Hadamard Transform/FT-IR Spectrometry

A computationally inexpensive method is presented for the recovery of spectra from measurements obtained with Hadamard transform spectrometers having nonideal masks. Normally, N measurements are required in order to recover an N-point spectrum; this method requires N + N0 measurements to be taken, where, typically, N0 ≤ 10. Once the additional measurements have been taken, only O(N[log2N + 2]) arithmetic operations—mostly additions or subtractions—are needed in order to recover the spectrum; a conventional procedure requires O(2N2) operations. Preliminary work for this method is minimal, requiring O(N) operations as opposed to O(N3) for a conventional procedure; this work needs to be done only once for a given spectrometer. The spectrum-estimate obtained is unbiased.

A Fast Spectrum-Recovery Method for Hadamard Transform Spectrometers Having Nonideal Masks

A spectrum-recovery method is presented which efficiently computes an optimal unbiased linear spectrum-estimate for measurements obtained with Hadamard transform (HT) spectrometers having nonideal masks. This method has the following advantages over other spectrum-recovery techniques: it is computationally efficient, it requires no additional measurements, and it computes an optimal spectrum-estimate. In the method presented, after the mask of the HT spectrometer has been characterized, approximately 3N preliminary arithmetic operations are performed once for a given spectrometer, where N is both the number of spectral resolution-elements desired and the number of measurements required. Each spectrum-estimate to be recovered then requires only an additional O[N(log2N + 4)] arithmetic operations. In contrast, conventional methods for obtaining an optimal unbiased linear spectrum-estimate require O(N3) preliminary operations, and O(2N2) operations during each spectrum-recovery.

An Efficient Method for Recovering the Optimal Unbiased Linear Spectrum-Estimate from Hadamard Transform Spectrometers Having Nonideal Masks

The authors consider a special class of problems in which the dominant noise source in the spectrum measurements is the noise generated in the detector itself. In this case, the signal-to-noise ratio of the spectrum estimates can be improved by a multiplexing technique known as Hadamard spectrometry. Specifically, if the MSE (mean-square error) of a single-slit spectrometer is sigma /sup 2/, the MSE of a Hadamard system will be approximately 4/ sigma /sup 2//N. In this expression, N is the number of spectral components to be estimated and the number of measurements to be taken. >

An introduction to Hadamard spectrometry and the multiplex advantage

The multiplex advantage offered by Hadamard transform spectrometry (HTS) can improve the signal-to-noise ratio (SNR) at the output of a spectrometer. However, additional processing of the spectrometer output is required to recover the individual spectral components. A block diagram description of HTS and the spectrum-recovery process is presented. A computer simulation of this model has been developed and can be used to examine the effects of certain nonidealities that may typically be encountered. Traditionally, the inverse Hadamard transform (IHT) has been used as the spectrum-recovery method, but the IHT does not take into account the nonidealities associated with the multiplexing process. Two spectrum-recovery methods that address the problems of nonidealities are presented. The relative performance of all three methods is compared with regard to mean square error (MSE) and the computational efficiency of the algorithms necessary to implement the schemes. An example application is described, and the performance of the three spectrum-recovery schemes is discussed. >

Implementation problems in Hadamard transform spectrometry

The multiplexing inherent in the Hadamard transform (HT) spectrometer can result in an improved spectrum-estimate when the detector is the major source of noise. A spectrum-estimate may be further improved by taking into account any nonidealities in the system. In this paper, observations concerning the errors associated with such estimates are presented, with the use of results obtained from computer simulations. Three spectrum-recovery techniques for an HT spectrometer having a nonideal electro-optic mask are considered in terms of the mean-square error (MSE) associated with a given estimate. The discussion of the MSE is with respect to the input spectrum to be estimated, the detector noise, the transmittances of the nonideal mask, and the use of coaddition. Included is a review of the computational efficiency and the statistical bias of each method. The relative performances of the spectrum-recovery methods are presented with examples to help identify the sources of error for each of the techniques.

B. K. Harms

Papers

A Fast Spectrum-Recovery Method for Hadamard Transform Spectrometers Having Nonideal Masks

An Efficient Method for Recovering the Optimal Unbiased Linear Spectrum-Estimate from Hadamard Transform Spectrometers Having Nonideal Masks

An introduction to Hadamard spectrometry and the multiplex advantage

Implementation problems in Hadamard transform spectrometry

On the Mean-Square Error of Various Spectrum-Recovery Techniques in Hadamard Transform Spectrometry