Method for analyzing intact capsules and tablets by near-infrared reflectance spectrometry

• US 4,893,253 A
• Filed: 03/10/1988
• Issued: 01/09/1990
• Est. Priority Date: 03/10/1988
• Status: Expired due to Fees

• Current Assignee: ADVANCED RESEARCH & TECHNOLOGY INSTITUTE
• Original Assignee: Indiana University Foundation
• Inventors: Lodder, Robert A.
• Primary Examiner(s): Lall, Parshotam S.
• Assistant Examiner(s): Teska, Kevin J.
• Application Number:
  US07/166,233
• Time in Patent Office: 670 Days
• Field of Search: 364/496-499, 364/413.01, 364/413.02, 364/525, 250/338.1, 250/338.5, 250/339-343, 356/36, 356/51, 356/30, 356/319, 356/328, 356/334
• US Class Current: 702/28
• CPC Class Codes: G01N 21/3577 for analysing liquids, e.g. polluted water
• CPC Class Codes: G01N 21/359 using near infrared light

Abstract

The BEAST method, when used in conjunction with NIRS data at as few as one or two wavelengths, is able to quickly detect a wide variety of adulterants in capsules, obviating the need to open them for analysis. The ability of this technique to detect the absence of components that should be present as well as the presence of components that should be absent enables it even to signal the presence of adulterants that have no near-infrared absorption. The technique generally comprises the steps of obtaining spectra for a training set of unadulterated samples, representing each specturm as a point in a hyperspace, creating a number of training set replicates and a bootstrap replicate distribution, calculating the center of the bootstrap replicate distribution, obtaining a spectrum for an adulterated sample, transforming this spectrum into a point in hyperspace, and identifying the adulterated sample as abnormal based on a relationship between the adulterated sample'"'"'s hyperspatial point and the bootstrap replicate distribution.

Claims

1. An analytical method for detecting adulterants in samples using a Near Infrared Reflectance Analysis (NIRS) instrument comprising the steps of:

(a) obtaining spectra at an appropriate number of wavelengths for a training set of an appropriate number of unadulterated samples;

(b) performing a transformation of each spectrum by representing such spectrum as a single point in hyperspace, and thus yielding a cluster of points in n-dimensional hyperspace, where n is the number of wavelengths utilized;

(c) creating a large set of training set sample replicates, each replicate being the same size as the training set, by randomly selecting samples from the training set, with replacement from the training set, and thus creating a bootstrap replicate distribution;

(d) calculating the center of the bootstrap replicate distribution;

(e) obtaining a spectrum for an adulterated sample;

(f) transforming this spectrum into a point in hyperspace;

(g) forming a hyperline in hyperspace between the point in hyperspace yielded from the spectrum of the adulterated sample and the center point of the bootstrap replicate distribution;

(h) forming a hypercylinder around the hyperline containing points of the training set sample replicates;

(i) projecting the multivariate points of the training set sample replicates onto the hyperline to form a univariate distribution of points;

(j) using this univariate distribution of points to construct confidence limits;

(k) identifying as abnormal the adulterated sample, the point of which falls three central 68% quantiles outside of the cluster of points yielded by the replicate spectra obtained from the training set sample replicates of the training set.

2. An analytical method for detecting inhomogeneous samples using a Near Infrared Reflectance Analysis (NIRS) instrument comprising the steps of:

(a) obtaining spectra at an appropriate number of wavelengths for a training set of an appropriate number of homogeneous samples;

(b) performing a transformation of each spectrum by representing such spectrum as a single point in hyperspace, and thus yielding a cluster of points in n-dimensional hyperspace, where n is the number of wavelengths utilized;

(c) creating a large set of training set sample replicates, each replicate being the same size as the training set, by randomly selecting samples from the training set, with replacement from the training set, and thus creating a bootstrap replicate distribution;

(d) calculating the center of the bootstrap replicate distribution;

(e) obtaining a spectrum for an inhomogeneous sample;

(f) transforming this spectrum into a point in hyperspace;

(g) forming a hyperline in hyperspace between the point in hyperspace yielded from the spectrum of the inhomogeneous sample and the center point of the bootstrap replicate distribution;

(h) forming a hypercylinder around the hyperline containing points of the training set sample replicates;

(i) projecting the multivariate points of the training set sample replicates onto the hyperline to form a univariate distribution of points;

(j) using this univariate distribution of points to construct confidence limits;

(k) identifying as abnormal the inhomogeneous sample, the point of which falls three central 68% quantiles outside of the cluster of points yielded by the replicate spectra obtained from the training set sample replicates of the training set.

3. An analytical method for detecting adulterants using a Near Infrared Reflectance Analysis (NIRS) instrument comprising the steps of:

(a) obtaining spectra at an appropriate number of wavelengths for a training set of an appropriate number of adulterated samples;

(b) performing a transformation of each spectrum by representing such spectrum as a single point in hyperspace, and thus yielding a cluster of points in n-dimensional hyperspace, where n is the number of wavelengths utilized;

(c) creating a large set of training set sample replicates using a Monte Carlo integration, each replicate being the same size as the training set, by randomly selecting samples from the training set, with replacement from the training set, and thus creating a bootstrap replicate distribution;

(d) calculating the center of the bootstrap replicate distribution;

(e) obtaining a spectrum for an adulterated sample;

(f) transforming this spectrum into a point in hyperspace;

(g) forming a hyperline in hyperspace between the point in hyperspace yielded from the spectrum of the adulterated sample and the center point of the bootstrap replicate distribution;

(h) forming a hypercylinder, the radius of which is 2 orders of magnitude greater than the length, around the hyperline containing points of the training set sample replicates;

(i) projecting the multivariate points of the training set sample replicates by triangulation onto the hyperline to form a univariate distribution of points;

(j) using this univariate distribution of points to construct confidence limits;

(k) identifying as abnormal the adulterated sample, the point of which falls three central 68% quantiles outside of the cluster of points yielded by the replicate spectra obtained from the training set sample replicates of the training set.