Table 1.
List of instruments and devices evaluated, including their underlying technology and basic instrument specifications.
Fig 1.
Illustration of the basic processes for qualitative spectral comparison and quantitative analysis.
(A) Process for reference library creation and spectral comparison analysis. From top to bottom: (i) spectra are collected from different batches of the same medicine and compiled into a mean spectrum representative of that medicine. (ii) This mean spectrum is used to build a “library” or database that serves as the comparator against which test samples are compared. (iii) Test samples are scanned and then (iv) the test sample spectra are overlaid with the reference spectrum for visual or computational comparison to determine a pass or fail. (B) Illustration of a basic quantitative experiment. From left to right and top to bottom. (i) A set of standard calibration samples with increasing API concentration is prepared along with a solution of the test sample that should fit in the concentration range of those standards. (ii) All solutions are then tested on the instrument and (iii) the data collected. (iv) The data obtained is then used to build a calibration curve via linear least-squares regression. Interpolation of the peak area of the questioned sample into this curve yields the estimated API concentration.
Table 2.
Sensitivity and specificity to correctly determine quality of medicine samples for the 12 tested devices.
Fig 2.
Comparison of NIR spectra obtained for ofloxacin-containing simulated medicines.
Spectra were collected for ofloxacin-containing simulated medicines using the (A) Neospectra 2.5, (B) NIR-S-G1, and (C) MicroPHAZIR RX spectrometers. The black trace is of a falsified simulated medicine tablet containing only starch. The blue trace is of a simulated good quality ofloxacin sample that contained starch as the bulk excipient.
Fig 3.
Comparison of Raman spectra obtained for Artesun artesunate powder.
Raman spectra were collected with the (A) Progeny and (B) TruScan RM spectrometers for Artesun artesunate powder for injection. Spectra are provided for 1) a scan of the bottom of the Artesun glass vial containing no artesunate (blue trace), 2) a sample containing 60 mg of artesunate powder, scanned through the bottom of the glass vial (orange trace), and 3) the artesunate powder transferred to a polypropylene bag and compacted into a more localized area to enable more focused analysis (green trace).
Fig 4.
Receiver operating characteristic (ROC) curves for substandard analysis with the spectrometers.
ROC curves were created for the (A) 4500a, (B) MicroPHAZIR RX, (C) Progeny, and (D) TruScan RM spectrometers. ROC curves were based only on the results for simulated substandard and good quality medicines. Each legend identifies the threshold chosen for each point, with the one labelled “Stock” being the threshold used for the study. The stock thresholds for the MicroPHAZIR RX’s correlation coefficient, Progeny’s correlation coefficient, and TruScan RM’s p-value were the default values set by the manufacturer. The 4500a stock threshold was selected for the study since that instrument did not output pass/fail results.