AUTHENTICATING AN ARTICLE

Abstract

A method of authenticating an article, the method comprising: reading a label associated with the article, the label comprising article fingerprint information; retrieving, in dependence on the article fingerprint information, an article fingerprint, the fingerprint describing the resonance response of the article to an excitation signal; applying an excitation signal to the article; receiving the resonance response of the article to the excitation signal; comparing the resonance response to the fingerprint; and determining in dependence on the comparison whether the article is authentic.

Claims

1. A method of authenticating an article, the method comprising: reading a label associated with the article, the label comprising article fingerprint information; retrieving, in dependence on the article fingerprint information, an article fingerprint, the fingerprint describing the resonance response of the article to an excitation signal; applying an excitation signal to the article; receiving the resonance response of the article to the excitation signal; comparing the resonance response to the fingerprint; and determining in dependence on the comparison whether the article is authentic.

2.-36. (canceled)

Description

[0063] The present invention is now described, purely by way of example,with reference to the accompanying diagrammatic drawings, in which:

[0064] FIG. 1 shows examples of research results indicating the suitability of NQR for detecting counterfeit medicines;

[0065] FIG. 2 shows an overview of a medicines authentication system;

[0066] FIG. 3 is a flowchart describing the operation of the fingerprinting stage of the medicines authentication system;

[0067] FIG. 4 is a flowchart describing the operation of the authentication stage of the medicines authentication system;

[0068] FIG. 5 is a schematic of a pulsed-spin locking excitation sequence and the resulting NQR response;

[0069] FIG. 6 shows an actual time domain NQR response generated from a pulsed spin-locking multiple-pulse sequence;

[0070] FIG. 7 shows an example of how the NQR response is dependent on the excitation pulse sequence;

[0071] FIG. 8 shows schematic representations of Distance-Sensitive Pulse Spin-Locking (DSPSL) phase-encoded sequences;

[0072] FIG. 9 shows NQR responses using three-block Distance-Sensitive Pulse Spin-Locking (DSPSL) phase-encoded sequences; and

[0073] FIGS. 10 and 11 show an example of NQR responses being used to determine the distance of a sample.

OVERVIEW

[0074] Generally, medicine authentication involves assessing whether a purported medicine corresponds to what it is claimed to be, for example, whether it is as described on the label of its packaging.

[0075] Medicine authentication using NQR involves matching the NQR response of the purported medicine, acquired during an authentication event, against an expected NQR response or “fingerprint”, which has been generated previously from a genuine or authentic version of the medicine.

[0076] FIG. 2 shows an overview of a medicines authentication system, specifically the authentication module. This comprises the following: [0077] NQR console 200 [0078] Connection to cloud 210 [0079] Power amplifier 220 [0080] T/R (transmit-receive) switch 230 [0081] Antenna and label reader 240 [0082] Receiver pre-amplifier 250

[0083] Also shown is a container 260 of medicine undergoing the authentication process and the label/fingerprint 270 being used for authentication.

[0084] FIG. 3 shows a flowchart describing the operation of the fingerprinting stage of the medicines authentication system.

[0085] QR fingerprint generator [0086] Sf1. Select pulse sequence settings i, ii or iii [0087] Sf2. Acquire OR response of target using chosen setting [0088] Sf3. Use OR response to generate fingerprint [0089] Sf4. Incorporate fingerprint in label [0090] Sf5. a) print label on packet [0091] OR b) upload label to cloud

[0092] FIG. 4 shows a flowchart describing the operation of the authentication stage of the medicines authentication system. [0093] QR Authentication steps [0094] Sa1. Read packet label into device [0095] Sa2. Access OR fingerprint [0096] Sa3. EITHER a) QR fingerprint encoded in label [0097] OR b) QR fingerprint downloaded from cloud [0098] Sa4. Recover pulse sequence settings from fingerprint [0099] Sa5. Acquire QR response from target using pulse sequence [0100] Sa6. Compare response to OR fingerprint [0101] Sa7. Match? [0102] Sa8. EITHER a) Yes green light [0103] OR b) No=>red light

[0104] NQR Fingerprint

[0105] The fingerprint may be generated from the medicine in various forms, for example as a powder, a single pill, caplet or capsule, an oral suspension, or as a packaged product containing the medicine, for example a number of packaged pills, caplets or capsules in a blister pack, cardboard packet or bottle.

[0106] Ideally, the NQR response that is used for authentication is generated in the same configuration in which the fingerprint was generated. For example, the response of a single pill is compared with the fingerprint of a single pill, rather than against the fingerprint of a plurality of pills or packaged pill(s).

[0107] The fingerprint is typically derived from at least some of the following measureable and quantifiable characteristics of an NQR response, whether intrinsic to the material or resulting from a convolution of intrinsic and one or more extrinsic properties, eg. sample processing, pulse sequence parameters and/or amount of material present:

TABLE-US-00001 Characteristic Origin of the characteristic Line frequency Intrinsic to the material Temperature coefficient (Hz/K) Spin-lattice relaxation time: T.sub.1 Spin-Spin relaxation time: T.sub.2 T.sub.2*/linewidth intrinsic characteristic of the material + lineshape sample processing (e.g. pressure applied in creating pill) T.sub.2e intrinsic characteristic of the material + pulse sequence parameters signal intensity intrinsic characteristic of the material + amount of material present + pulse sequence parameters

[0108] An NQR response generally comprises a number of spectral lines, corresponding to frequencies, so there is a choice of which are used in defining the fingerprint.

[0109] T.sub.1 and T.sub.2 each require a series of experiments to measure and thus are unlikely to be a direct part of the fingerprint (albeit that T.sub.1 and T.sub.2 are present indirectly in T.sub.2e and T.sub.2).

[0110] T.sub.2e denotes a time constant associated with the decay in signal across the entire echo train of an NQR response; T.sub.2* is a time constant associated with the decay in signal of individual echo.

[0111] Knowledge of temperature is required for use of frequency information allied to the temperature coefficient, as line frequency, T.sub.1, T.sub.2 and T.sub.2e are temperature dependent.

[0112] In some embodiments, the fingerprint comprises an image of the expected NQR response (e.g. the spectrum or a spectral plot) and a direct comparison is made between the NQR response acquired during the authentication event and the fingerprint.

[0113] More likely, the fingerprint comprises a string of acceptable values for some or all of the response characteristics and a comparison is made with the corresponding values extracted by signal processing from the NQR response acquired during the authentication event.

[0114] The signal processing is similar to that used for determining whether an NQR signal is real or merely due to radio-frequency interference, based on measuring the same characteristics.

[0115] The following table shows typical characteristics of two different medicines, paracetamol tablets and sulfadiazine powder:

TABLE-US-00002 Paracetamol Sulfadiazine Characteristic tablets powder Frequency (of spectral lines at 2563.2 2563.7 294 K)/kHz Temperature coefficient/Hz/K 40 90 T.sub.2*/ms 0.18 0.16-0.55 T.sub.1/s 6 s 9 s T.sub.2e/ms (at 2τ = 2.4 ms) 450 2300 T.sub.2StarEst (estimated T.sub.2*) 1.65E−04 2.32E−04 etDampEst (estimated 1/T.sub.2e) 2.55E−06 9.56E−05 dampest (estimated 1/T.sub.2*) 0.036 0.207 T.sub.2effEst (estimated T.sub.2e) 2.361 0.502

[0116] As can be seen, at room temperature the NQR spectral lines of the two medicines are of very similar frequency. This means that they would not by themselves make for a good fingerprint. Typically, therefore, the fingerprint is built from multiple characteristics of the NQR response.

[0117] Although other lines in the spectra, which are not at the same frequencies, could be used for the fingerprint, this may not be necessary if there are other characteristics (e.g. in this case T.sub.2e) which are sufficiently different so as to be discernible.

[0118] Acquiring the NQR Response

[0119] Time is a factor in authentication. In order to authenticate medicines quickly, a pulse sequence, comprising a plurality of radiofrequency pulses is used as this will allow the signal processing to extract from the NQR response all the characteristics needed to compare against the fingerprint in a “single shot”.

[0120] The following table shows characteristics that can be extracted in a single shot from different classes of pulse sequence:

TABLE-US-00003 Class of pulse sequence “single shot” characteristics Single pulse (stochastic) Frequency, T.sub.2*, lineshape, signal intensity Pulsed spin-locking multiple- Frequency, T.sub.2*, T.sub.2e, lineshape, pulse signal intensity Steady-state multiple pulse Frequency, T.sub.2*, lineshape, signal intensity

[0121] In practice—as time is an issue—multiple-pulse pulse sequences are favoured for producing responses good signal-to-noise ratios (SNR) in a small amount of time.

[0122] Furthermore, as steady-state sequences work only within certain ranges of T.sub.1 and T.sub.2*, mostly, but not exclusively, a pulsed-spin locking sequence is used.

[0123] FIG. 5 is a schematic of a pulsed-spin locking excitation sequence and the resulting NQR response, showing the time-dependent signal intensity of a) excitation radiofrequency pulses 320 generated during an authentication event, and b) the resulting NQR response 300, comprising echoes 310 (the plurality of sequential echoes 310 defining an “echo train”).

[0124] FIG. 6 shows an actual time domain NQR response generated from a pulsed spin-locking multiple-pulse sequence, from which frequency, signal intensity T.sub.2* and T.sub.2e and lineshape may be determined.

[0125] While the response frequency and T.sub.2* are independent of the pulse sequence, i.e. the duration, intensity and/or relative “timings” of the radiofrequency pulses, both the response signal intensity and T.sub.2e are heavily influenced by these timings; most particularly, the values of these characteristics decrease with increasing time between the radiofrequency pulses. This allows for a further level of security to be introduced into the authentication system by way of Pulse Sequence Encoding.

[0126] Pulse Sequence Encoding

[0127] As mentioned previously, the value of certain measurable characteristics of the NQR response are influenced by features of the pulse sequences used ie. the selection choices made regarding eg. number of pulses, time between successive pulses (pulse spacing) etc. The same material, present in the same quantity may return different values for T.sub.2e and/or signal intensity depending on these selection choices, while T.sub.2* and line frequency are unaffected.

[0128] FIG. 7 shows an example of how the NQR response is dependent on the excitation pulse sequence. For two different pulsed spin-locking sequences A and B, where sequence A has a lower pulse spacing (known as 2τ in pulsed spin-locking sequences) than sequence B, the same material present in the same quantity returns a higher signal intensity response for sequence A than for sequence B.

[0129] Hence, first a fingerprint for a first batch of a medicine may be generated using sequence “A” and a second fingerprint for a second batch of the same medicine (i.e. same manufacturer, same product and same packaging) produced, for example, on a different day, is generated using sequence “B”.

[0130] By encoding information regarding which sequence (A or B) was used into the label or tag, the authentication device would be able to reproduce correctly the conditions used to generate the fingerprint specific to that batch—and hence identify the batch.

[0131] A change would also observed in T.sub.2e between the fingerprints generated using the two different sequences.

[0132] Other types of encoding which could be used, such as changing the number of pulses in the sequence, would affect signal intensity but not T.sub.2e.

[0133] Even if a counterfeiter were able to measure the intrinsic NQR characteristics of a medicine in an attempt to fake parts of a fingerprint (e.g. line frequency, T.sub.2*), they would not know the necessary details of the pulse sequence which had been used and so be unable to recreate the whole fingerprint.

[0134] Distance Correction

[0135] Another factor which influences the values of the characteristics which may form the fingerprint is effect of the distance between the antenna and the material being measured.

[0136] Effectively, the authentication device used to acquire the NQR response (for generating the fingerprint and/or for performing an authentication event) may be considered to operate in two modes: [0137] in a first mode, the pill or packaged medicine is dropped into the radiofrequency coil of the authentication device which surrounds it [0138] in a second mode, where the authentication device comprises a unilateral antenna brought up to the side of the pill or packaged medicine.

[0139] In the latter mode the effect of the distance between the antenna and material being measured affects the NQR response, in particular the intensity characteristic.

[0140] The difficulty arises when using a unilateral antenna on a sample of medicine in a sealed box, where it can be difficult to know the distance from the antenna to the sample.

[0141] As the intensity of both the radiofrequency field (B.sub.1) and the measured NQR response decrease with distance, when the distance from the antenna to the sample is unknown, it is difficult to know from an NQR response if the amplitude of the signal acquired is due to there being a small amount of the sample close to the antenna, or a large amount of the sample further away. This can make it difficult to conduct quantitative medicine authentication.

[0142] The problem may be solved by exploiting the drop-off in the B.sub.1 field with distance to encode the NQR response in such a way that the true signal intensity, corresponding to zero distance between antenna and sample, may be calculated independently of the distance between the antenna and sample, based on a graph of distance from coil versus correction factor.

[0143] Preferably, the solution: [0144] gives information on distance to the sample in a single measurement [0145] uses a preferred pulse sequence [0146] does not suffer any reduction in signal-to-noise ratio compared to a non-encoded version of a preferred pulse sequence

[0147] FIG. 8 shows schematic representations of Distance-Sensitive Pulse Spin-Locking (DSPSL) phase-encoded sequences.

[0148] FIG. 8(a) shows a two-block DSPSL pulse sequence, comprising the pulsed spin-locking sequence:

P1.sub.±x−(τ−P2.sub.+y−τ−).sub.n,

where pulses P1 and P2 are of phase +x, −x or +y; τ is a time period between pulses; and n is the number of times the combination of delays 2τ and P2 are repeated.

[0149] A third class of pulse—P3—is inserted midway in the trains of P2 pulses, such that:

P1.sub.±x−(τ−P2.sub.+y−τ−).sub.n/2−τ.sup.1−P3.sub.∓x−τ.sup.1−(τ−P2.sub.+y−τ−.sub.n/2

where τ.sup.1=[τ−(P3−P2)] in order to keep the time between the pulses at 2τ.

[0150] Applying the resulting encoded PSL sequence, signals acquired before pulse P3—in “block one” 540—are compared to those acquired after pulse P3—in “block two” 550.

[0151] The difference in phase between the signals acquired in the first data block 540 and those acquired in the second data block 550 is dependent on the flip angle of pulse P3 530.

[0152] The flip angle of P3 is in turn dependent on the duration of P3 and the B.sub.1 field at the sample.

[0153] If the pulse duration is fixed for a given authentication event, then the B1 field to which the sample is exposed is dependent upon the distance of the sample from the unilateral antenna (in the absence of RF shielding between sample and antenna).

[0154] Thus the encoded DSPSL pulse sequence can be used to determine the distance of a sample from the unilateral antenna via a comparison of the phase of the signals in the first data block 540 (associated with the second pulse (P2) 520-1) to that of the signals in the second data block 550 (associated with a subsequent P2 pulse 520-2).

[0155] The phase of the signal in the second data block 550 will be inverted with respect to that of the signal in the first data block 540 only when the sample is at a distance from the coil where the B.sub.1 field is such that the connecting P3 pulse 530 acts as an inversion pulse; at all other distances the phase difference will be something other than 180 degrees.

[0156] The P3 pulse 530 need not be halfway along the train of pulses; it can be placed at any point. It would, however, be desirable to place it as late in the train as possible to maximize the SNR acquired from the signals acquired before the P3 pulse.

[0157] There is further advantage in using two (or more) P3 pulses for a finer degree of distance information, as described below.

[0158] FIG. 8(b) shows a three-block DSPSL pulse sequence. This is essentially the two-block DSPSL sequence described above modified through the introduction of a second P3 B.sub.1-specific inversion pulse 530-2 and a third data block 560. Thus any magnetization knocked out of the xy plane by a first P3 pulse 530-1 will be brought back into the xy plane by the second P3 pulse 530-2.

[0159] FIG. 9 shows NQR responses using three-block DSPSL phase-encoded sequences, in this example from paracetamol pills placed 5 mm above the surface of the unilateral antenna.

[0160] Typically, whole echo trains are acquired for signal averaging; however, for clarity, here only the envelope of the echo train is shown (i.e. the data points have been averaged down to a single data point per echo) to allow the phase of the signal in each data block to be directly compared without the need for further signal processing.

[0161] Different numbers of echoes have been acquired in each block to illustrate different possible configurations of the sequence. The second 550 and third 560 blocks are primarily not to provide signal, but to provide distance information via the phase changes in the signals from first 540 to second 550 to third 560 blocks.

[0162] FIG. 9(a) shows a time-domain echo train with a single data point per echo (128 echoes per block).

[0163] FIG. 9(b) shows a time-domain echo train with a single data point per echo (256 echoes in the first block 540 and 64 echoes in the second block 550 and third blocks 560). The number of echoes in each of the second and third blocks is a quarter of that in the first block, so there is greater T.sub.2e decay going from the first block to second block etc. This is considered the optimal set-up provided the SNR is sufficient in the second and third blocks to allow the phase to be measured.

[0164] FIGS. 10 and 11 show an example of NQR responses being used to determine the distance of a sample.

[0165] FIG. 10 shows two boxes Box 1 and Box 2 of sample medicine (boxes of pills) in two different arrangements with respect to the antenna.

[0166] FIG. 11 shows NQR echo train envelopes of: [0167] i) on the left, the signals returned by the DSPSL sequence with the sample placed progressively further away from the antenna; [0168] ii) on the right, the signals returned by the DSPSL sequence with the two boxes A and B.

[0169] By varying the distance of the sample from the antenna, it is possible to build a chart of the relative phase between the three blocks with increasing distance between the sample and the antenna. This chart is then used to estimate the position of the sample in the two different boxes.

[0170] A comparison of the plots on the right with those on the left indicates that the pills in the second box, Box 2 are further from the antenna than the pills in the first box, Box 1—in this case this is due to the second box being made of thicker cardboard being too big to fit within the rim of the shield plate, meaning it cannot be placed directly on the Perspex covering the antenna

[0171] Alternatives and Modifications

[0172] Alternative embodiments may use techniques other than NQR, for example: X-rays; NMR; THz radiation; Raman scattering; near infrared radiation; information on the granularity or colour of a medicine under different wavelengths of light; and/or chromatographic techniques.

[0173] In some embodiments pulses of other function and/or design for pulse P3 are used, for example sequences that place P3 in different places along the train and/or which use different numbers of P3 pulses.

[0174] It will be understood that the present invention has been described above purely by way of example, and modifications of detail can be made within the scope of the invention.

[0175] Each feature disclosed in the description, and (where appropriate) the claims and drawings may be provided independently or in any appropriate combination.

[0176] Reference numerals appearing in the claims are by way of illustration only and shall have no limiting effect on the scope of the claims.