Seeing Cancer at the Molecular Level: How Raman Spectroscopy Is Rewriting the Rules of Medical Imaging
By Swati Tanwar, Ph.D. Assistant Research Scientist, Department of Mechanical Engineering, Johns Hopkins University, Baltimore, MD
Every year, millions of cancer diagnoses hinge on a biopsy. An invasive, painful procedure that samples a tiny fragment of tissue and hopes it tells the full story. But tumors are not uniform. They are heterogeneous landscapes of cells with different behaviors, different enzyme profiles, and different degrees of aggression. A needle inserted into one corner of a tumor may completely miss what is happening three millimeters away. We have built extraordinary machines to image the human body, yet we still struggle to see cancer at the level that matters most: the molecule.
That is the problem that drew me into research in the first place. And the technology I am most excited about is one that most clinicians have still never encountered: Raman spectroscopy.
A Different Kind of Imaging
Most medical imaging tells us where something is. Raman spectroscopy tells us what it is.
When a laser illuminates biological tissue, most of the light scatters back unchanged. But a tiny fraction, roughly one in every hundred million photons, scatters at a shifted frequency, carrying the unique vibrational fingerprint of the molecules it encountered. This is the Raman signal.1 Each molecule, a protein, a lipid, an enzyme, has its own spectral signature, as distinctive as a fingerprint. By reading those signatures, Raman spectroscopy can map the molecular chemistry of a cell or tissue without staining it, cutting it, or destroying it.
What makes this so powerful for cancer detection is that cancer is fundamentally a molecular disease. Tumors overexpress specific enzymes, alter their metabolic profiles, and carry biomarkers that distinguish aggressive cells from indolent ones. Raman spectroscopy can detect those differences, not just structurally, but chemically, with sub-cellular spatial resolution. Raman peaks are narrow, stable, and highly specific, making them ideal for multiplexed imaging of multiple targets simultaneously.
Cancer has always been one step ahead because it operates at a molecular level we could not see clearly enough. Raman spectroscopy is beginning to change that.
From Concept to Cancer Cell: What We Built at Johns Hopkins
My work at Johns Hopkins has focused on the intersection of Raman spectroscopy, nanotechnology, and cancer biology to develop a new generation of smart molecular probes that can detect and monitor cancer with a precision that conventional imaging simply cannot match.
Our first major innovation was the nanoSABER probe,2 a self-assembling, bioorthogonal Raman nanoprobe designed to activate inside cancer cells in response to a specific tumor enzyme called legumain. Legumain is overexpressed in aggressive prostate, breast, and gastric cancers, making it an ideal molecular target. The nanoSABER probe remains silent until it encounters legumain inside the cell, at which point it undergoes a chemical condensation reaction, self-assembles into a larger nanoparticle, and generates a distinct Raman signal in the cell-silent spectral window, a region of the spectrum free from background biological noise. In live mouse models, we demonstrated that nanoSABER could selectively light up aggressive tumors while leaving normal tissue dark, achieving a level of tumor specificity that is difficult to match with existing techniques.
Building on this, we developed a DNA origami-based nanodevice3 that combined Raman imaging with targeted drug delivery. By coating a programmable DNA scaffold with an enzyme-responsive peptide sequence conjugated to doxorubicin, a front-line chemotherapy drug, we created a system that could deliver the drug specifically to legumain-overexpressing cancer cells, while simultaneously allowing us to track exactly where the drug went using Raman imaging. This was a novel demonstration of a DNA origami platform functioning as a Raman-active theranostic device, a tool that both treats and images in one integrated system.
Most recently, we addressed one of Raman’s most persistent clinical limitations, imaging speed, by engineering bimodal SERS-dark-field nanoprobes.4 Surface-enhanced Raman scattering (SERS) amplifies Raman signals enormously by concentrating light in the nanogaps of plasmonic gold nanoparticles.5 However, acquiring a full SERS map of a single cell can take many minutes, far too slow for a clinical workflow. Our solution was to combine SERS with dark-field microscopy: the dark-field modality rapidly pinpoints where the nanoprobes have aggregated inside cancer cells, and SERS then validates the molecular identity with precision. The result was complete, specific imaging of a single cancer cell in under two minutes.
The Challenges I Faced and What They Taught Me
None of this came easily. The path from a promising spectroscopic technique to a functional imaging platform in living biological systems is long and unforgiving.
The first challenge was sensitivity in a noisy biological environment. Living cells are biochemically complex; endogenous molecules generate Raman signals that compete with your probe signal. Our solution, introducing vibrational tags that absorb in the cell-silent spectral region between ~1800–2800 cm⁻¹, was conceptually simple but technically demanding, requiring iterative probe redesign and rigorous spectral validation.
The second challenge was specificity, and our solution came from asking a different question. Instead of asking how to find cancer cells, we asked: what makes the most dangerous cancer cells dangerous? The answer was legumain, an enzyme overexpressed in aggressive, metastatic tumors. By designing probes that respond exclusively to legumain activity, we created a system that is not just cancer-targeted, it is aggression-targeted, lighting up precisely the cells that matter most while ignoring everything else.
The third challenge is one the entire field shares: bridging the gap between a promising laboratory result and a clinically usable tool. In vitro results can be beautiful. But moving probes into living animals introduces a gauntlet of new variables: immune response, renal clearance, off-target accumulation, and the sheer complexity of the tumor microenvironment. Our biocompatibility studies showed no signs of toxicity in major organs, and our in vivo Raman results were compelling, but I remain clear-eyed about the significant engineering and regulatory work that lies between a mouse model and a hospital. The path forward requires not just better probes but also closer collaboration among engineers, biologists, and clinicians from the very earliest stages of probe design.
The Positive Impact and the Honest Limitations
The positive case for Raman spectroscopy in medical imaging is strong. It is non-invasive. It requires no ionizing radiation. It can differentiate between molecularly different cancer subtypes that appear the same under a traditional microscope. It can monitor whether a drug has reached its target and whether it is working. In a healthcare system where misdiagnosis and chemotherapy toxicity remain significant problems, these capabilities are not incremental improvements. They are paradigm shifts.
Raman signals are inherently weak, requiring sophisticated equipment and long acquisition times compared to fluorescence. Miniaturized, clinically deployable Raman systems are still maturing. The regulatory pathway for nanoparticle-based imaging agents is complex and slow. And the cost of transitioning any new imaging modality into routine clinical use, training, infrastructure, and reimbursement is not trivial.
Looking Forward
What excites me most about the next decade is convergence. Raman spectroscopy is not advancing in isolation; it is being amplified by artificial intelligence, by advances in plasmonic nanomaterials, by DNA nanotechnology, and by a growing clinical appetite for precision diagnostics.
Machine learning models are already being trained to extract molecular signatures from complex Raman datasets faster and more reliably than any human analyst. Miniaturized Raman probes are being developed for endoscopic and intraoperative use, bringing real-time molecular imaging into the surgical suite. The multiplexing capability of Raman, the ability to track many molecular targets simultaneously with a single laser, may one day allow clinicians to profile the full molecular landscape of a tumor in a single imaging session.
We are not there yet. But having watched a self-assembling nanoprobe light up a cancer cell in a mouse for the first time, knowing that the signal appeared precisely where aggressive tumor cells were, and nowhere else, I believe the destination is worth the journey. Cancer has always been one step ahead because it operates at a molecular level that we could not see clearly enough. Raman spectroscopy is beginning to change that.
In practical clinical settings, the ultimate value of Raman imaging may lie not in replacing existing modalities, but in augmenting them. Surgeons could potentially use Raman-guided probes to distinguish tumor margins during operations, while pathologists may one day combine AI-assisted Raman analysis with traditional histology to improve diagnostic accuracy and molecular profiling.
References
(1) Cialla-May, D.; Krafft, C.; Rösch, P.; Deckert-Gaudig, T.; Frosch, T.; Jahn, I. J.; Pahlow, S.; Stiebing, C.; Meyer-Zedler, T.; Bocklitz, T.; et al. Raman Spectroscopy and Imaging in Bioanalytics. Analytical Chemistry 2022, 94 (1), 86-119. DOI: 0.1021/acs.analchem.1c03235.
(2) Tanwar, S.; Ghaemi, B.; Raj, P.; Singh, A.; Wu, L.; Yuan, Y.; Arifin, D. R.; McMahon, M. T.; Bulte, J. W. M.; Barman, I. A Smart Intracellular Self-Assembling Bioorthogonal Raman Active Nanoprobe for Targeted Tumor Imaging. Advanced Science 2023, n/a, 2304164. DOI: 10.1002/advs.202304164.
(3) Tanwar, S.; Date, S.; Goel, L.; Wu, L.; Chatterjee, A.; Barman, I. Raman Imaging of Targeted Drug Delivery with DNA-Based Nano-Optical Devices. Small 2025, 21 (4), 2402631. DOI: 10.1002/smll.202402631.
(4) Tanwar, S.; Zheng, P.; Wu, L.; Barman, I. Stimuli-responsive ‘On–Off’ SERS–darkfield bimodal plasmonic nanoprobes for selective cancer cell illumination. Biosensors and Bioelectronics 2025, 286, 117615. DOI: https://doi.org/10.1016/j.bios.2025.117615.
(5) Laing, S.; Jamieson, L. E.; Faulds, K.; Graham, D. Surface-enhanced Raman spectroscopy for in vivo biosensing. Nature Reviews Chemistry 2017, 1 (8), 0060. DOI: 10.1038/s41570-017-0060.