Italian Researchers Validate Method to Test Cocaine Evidence Without Opening the Bag

In a recent study, a team of researchers from Sapienza Università di Roma demonstrated that two near-infrared (NIR) analytical techniques can identify, classify, and quantify cocaine and its common cutting agents directly through sealed polyethylene evidence bags without opening the packaging. The findings, which were published in Analytica Chimica Acta, were validated on real cocaine samples seized in four separate law enforcement operations in Italy.¹

What was the main problem that this study was trying to address?

One of the main problems in forensic drug analysis is that current methods require analysts to open sealed evidence to extract samples and test them.¹ This is a concern because it raises questions of evidence alteration before trial, which can complicate verdicts.^1,2 At the very least, it can destroy a piece of evidence that supports a key fact in a criminal case.² In Italy, non-repeatable forensic analyses are governed by Article 360 of the Code of Criminal Procedure, which requires that the original state of a sample be preserved.

How did the researchers attempt to solve this problem?

The research team explored using two spectroscopic methods to solve the abovementioned problem. The two methods used were Fourier transform near-infrared (FT-NIR) spectroscopy and near-infrared hyperspectral imaging (NIR-HSI).¹ Both methods use NIR light, which can pass through polyethylene packaging, to generate a chemical signature of a sample's contents without physical contact.¹ Because no solvents, reagents, or sample extraction are involved, the researchers describe the approach as non-destructive, solvent-free, and compliant with evidentiary preservation requirements.¹

Once they had their method, the researchers built their test set, which included cocaine from four independent seizures alongside six adulterants commonly used to dilute or "cut" the drug before street sale: creatine, caffeine, levamisole, lidocaine, lactose, and mannitol.¹ All samples remained sealed in their original polyethylene bags throughout testing.

How did the researchers interpret spectral data?

The researchers applied chemometric modeling in interpreting the spectral data. Chemometric modeling is a branch of data analysis that extracts chemically meaningful patterns from large, complex spectral data sets.³ The team also used exploratory principal component analysis (PCA) was used to visualize how cocaine and adulterant spectra differed from one another, while a supervised classification method called soft independent modeling of class analogy (SIMCA) was trained to recognize cocaine's spectral signature and flag anything that did not match it.¹ In testing, that classification model achieved 100% sensitivity and specificity, meaning it correctly identified every cocaine sample and correctly rejected every non-cocaine adulterant, without false positives or false negatives.¹

Meanwhile, the researchers constructed a quantitative model using partial least squares (PLS) regression to predict the cocaine content of the seized samples with a coefficient of determination of 0.95 and a prediction error of 2.40%.¹ This result suggests that the technique can estimate drug purity, not just confirm identity.

The hyperspectral imaging component added a further capability to the method. Because NIR-HSI captures a full spectrum at every pixel of an image rather than a single averaged reading, it allowed the researchers to map the chemical composition across the surface of a sample and distinguish cocaine, adulterants, and background packaging material within the same image, without manually isolating the sample first.¹

What are the key takeaways from this study?

The researchers caution that the results should be treated as proof of concept rather than a finalized, universally applicable model. The study drew on a limited number of seizures, and the range of cutting agents and packaging types tested does not capture the full variability forensic labs encounter in practice. The authors say future work should expand the dataset to include more seizures, a wider range of cocaine purity levels and adulterant combinations, and testing across different instruments and environmental conditions to confirm the method's reliability for routine deployment.

References

1. Spinelli, E.; Casamassima, R.; Marini, F. Non-destructive Forensic Identification and Quantification of Cocaine through Sealed Packaging Using FT-NIR Spectroscopy and Hyperspectral Imaging. Anal. Chim. Acta 2026, 1414, 345679. DOI: 10.1016/j.aca.2026.345679
2. Schiro, G. Collection and Preservation of Blood Evidence from Crime Scenes. Crime Scene Investigator Network, 2026. https://www.crime-scene-investigator.net/blood.html (accessed August 14, 2026).
3. Weber, J.; Latza, A.; Ryabchykov, O.; Valet, O.; Darina Storozhuk, D.; Rathmell, C.; Bingemann, D.; Creasey, D. Real-Time Chemometric Analysis of Multicomponent Bioprocesses Using Raman Spectroscopy. Spectroscopy 2024, 39 (5), 14–22. DOI: 10.56530/spectroscopy.eo3187v4