A research team led by Professor Xiaochun Li of the Institute of Biomedical Precision Testing and Instrumentation at Taiyuan University of Technology, working in collaboration with Director Lizhong Zhang of Shanxi Bethune Hospital, has now unveiled a platform that tackles both halves of this problem at once. Writing in the journal Biomedical Analysis, the team describes a system that pairs a specially engineered microfluidic chip with a deep learning model to enrich and identify rare tumor cells without the use of molecular labels. Dr. Weizhi Liu of Taiyuan University of Technology, Professor Li, and Director Zhang serve as co-corresponding authors, with master’s student Junyi Ouyang as first author and Dr. Haiqin Li also contributing to the study. The work, published under the title Label-free Enrichment and Identification of Circulating Tumor Cells Integrating Inertial Microfluidics and Deep Learning, represents a proof-of-concept demonstration of how physics and artificial intelligence can be woven together into a single analytical pipeline.

The first stage of the platform addresses the separation problem, and it does so by exploiting the most basic physical difference between tumor cells and blood cells: their size. Most conventional CTC isolation techniques depend on antibodies that bind to specific molecules on the surface of tumor cells, most famously the epithelial marker EpCAM. This strategy has a fundamental weakness. Tumor cells are extraordinarily diverse, and different subpopulations of CTCs, particularly those that have undergone a transition toward a more invasive state, may shed or never express the very markers the technology is designed to detect. Cells that lack the marker simply slip through the filter, taking their secrets with them. Label-free approaches sidestep this limitation by relying on intrinsic physical characteristics such as cell diameter and deformability, which are far harder for a cell to disguise.

The chip designed by the Taiyuan team is a spiral microfluidic device containing carefully engineered contraction-expansion structures along its length. The underlying principle is inertial microfluidics, a technique in which fluid flowing through narrow channels at high speed generates lift forces that push cells into predictable equilibrium positions across the channel cross-section. Because CTCs are generally larger than blood cells, with typical diameters of roughly 12 to 25 micrometers compared with 7 to 12 micrometers for white blood cells and 6 to 8 micrometers for red blood cells, each cell population experiences a different balance of forces as it travels through the microchannel. The contraction-expansion geometry amplifies these differences, steering tumor cells and blood cells into separate flow paths so that tumor cells can be collected at a dedicated outlet without any dependence on surface markers.

The team validated the separation concept through computer simulations and laboratory experiments before testing it on artificial blood samples spiked with MCF-7 breast cancer cells and white blood cells. The results were striking. Tumor cells were concentrated at the central outlet while the majority of blood cells were routed toward the side outlets. Quantitatively, the system achieved an MCF-7 cell recovery rate of 89.2 percent, with a standard deviation of 3.1 percent, and a white blood cell removal rate of 86.9 percent, with a standard deviation of 1.4 percent. Perhaps most tellingly, the proportion of tumor cells in the collected fraction rose from 9 percent before separation to 38 percent after enrichment. That fourfold increase in purity dramatically reduces the background the identification stage must contend with, and it does so without staining, labeling, or chemically altering the cells in any way.

Separation, however, is only half the battle. Once the enriched sample has been collected, researchers still face the question of which cells among the mixture are truly tumor cells. In many existing workflows this determination requires fluorescent staining or antibody-based labeling, which adds processing steps, consumes time, requires specialized reagents and equipment, and can compromise the viability or molecular integrity of the very cells researchers hope to study downstream. Fluorescence-based identification also ties the analysis to specific markers, reintroducing at the detection stage the same bias that label-free separation was designed to eliminate. The Taiyuan team’s answer to this dilemma is a deep learning model based on the YOLOv8 architecture, trained to recognize tumor cells directly from ordinary bright-field microscope images, the kind of imagery any standard laboratory microscope can produce.

YOLOv8, whose name stands for You Only Look Once, is a family of object detection models known for speed and accuracy in computer vision tasks. During the development of the platform, fluorescence images were used to establish the ground-truth identity of different cell types and to support the annotation of training data, but once training was complete the model performed all recognition using bright-field images alone. This is a meaningful distinction from many image-analysis approaches that require individual cells to be manually isolated or segmented before classification. YOLOv8 can locate, classify, and count cells directly within complete microscope images, treating the identification task as a single integrated detection problem rather than a multi-step pipeline vulnerable to accumulated error.

The performance figures reported for the identification stage are impressive for a label-free method. The model achieved 96.0 percent overall accuracy in distinguishing tumor cells from non-tumor cells, with precision and recall both at 94.6 percent and specificity at 95.4 percent. In practical terms, this means the system correctly identifies nearly all tumor cells present while rarely misclassifying blood cells as tumor cells, a balance that is critical when the starting population is so heavily skewed toward blood cells. The results demonstrate that artificial intelligence can extract subtle morphological information from standard bright-field images, information that would previously have required fluorescent labeling to access, thereby reducing cost, complexity, and the risk of damaging precious clinical samples.

By combining physical cell sorting with artificial intelligence, the platform addresses the two major bottlenecks that have constrained CTC analysis: finding rare tumor cells in a complex blood environment and recognizing them accurately after enrichment. The label-free workflow may simplify sample processing and, because the cells are never stained or chemically modified, help preserve their integrity for downstream applications such as single-cell sequencing and drug susceptibility testing. Those applications matter enormously, since a viable CTC captured intact can reveal the genetic mutations driving a patient’s tumor and predict which therapies are likely to work before any drug is administered. More broadly, the study illustrates how the marriage of microfluidic technology and deep learning can open new possibilities for analyzing rare biological cells of many kinds, not only tumor cells.

The researchers are careful to frame the current work as a proof of concept rather than a finished clinical tool. The platform was validated using MCF-7 cell lines and artificial blood samples rather than patient-derived clinical samples, and the enrichment and recognition modules have not yet been fully integrated into a single automated system. Future studies, the team says, will focus on evaluating additional tumor types and real patient samples while improving system integration and automation. Corresponding author Xiaochun Li summarized the motivation plainly, noting that the challenge of circulating tumor cell analysis is not only finding these rare cells but also identifying them accurately after separation, and that by combining microfluidics with artificial intelligence the team hopes to provide a simpler and more flexible approach for label-free CTC analysis and future downstream applications. If subsequent validation in clinical samples confirms these early results, the day when a routine blood draw can expose the hidden travelers of metastatic cancer may be considerably closer than it once seemed.

Subject of Research: Label-free detection of circulating tumor cells using inertial microfluidics and deep learning

Article Title: A tiny chip and AI team up to find hidden tumor cells in blood

Article References: A tiny chip and AI team up to find hidden tumor cells in blood. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: circulating tumor cells, liquid biopsy, microfluidics, inertial microfluidics, deep learning, YOLOv8, label-free detection, cancer diagnostics, bright-field imaging, cell sorting, MCF-7, biomedical analysis

The fund operates through a highly competitive process that weighs several explicit criteria, including the availability of intellectual property protection, marketability, commercial potential, technical feasibility and the breadth of a project’s impact. Since its launch in 2011, the program has invested more than 5.1 million dollars and advanced the commercial readiness of 99 innovations born at SUNY campuses. That seed money has catalyzed an additional 41 million dollars in follow-on investment from government agencies, industry licensees and early-stage investors, a leverage ratio that underscores why university systems increasingly treat translational funding as core infrastructure rather than a peripheral perk. Chancellor King framed the program in sweeping terms, describing SUNY as a national leader in interdisciplinary research that develops state-of-the-art technologies, saves lives, transforms industries and serves the public good. The Board of Trustees echoed the sentiment, noting that SUNY research uses revolutionary breakthroughs to fuel economies and empower communities throughout New York State and beyond.

Among the most technically ambitious awards is a project at the University at Albany led by Dr. Spyros Galis, who is building a scalable platform for quantum sensing and imaging that operates at room temperature. Current quantum photonic devices typically require cryogenic cooling to function properly, a constraint that has limited their widespread adoption since refrigeration hardware adds cost, bulk, complexity and substantial energy consumption to any deployment. By re-engineering quantum photonic platforms to work without cooling, the approach promises to reduce all of those barriers simultaneously, opening the door to real-world applications in sensing and imaging that would be impractical if each device needed a cryostat. Quantum sensors exploit delicate quantum states to measure physical quantities with extraordinary precision, and removing the thermal bottleneck is widely regarded as a prerequisite for moving such instruments out of specialized laboratories and into hospitals, factories and field instruments.

Also at Albany, Dr. Gary Saulnier is developing a product called CurrentView, a bedside monitoring system that provides real-time, non-invasive, three-dimensional monitoring of pulmonary perfusion and ventilation. The clinical target is the care of neonatal and pediatric patients with congenital heart disease and other cardiopulmonary conditions, populations in which complete, continuous data can be decisive for treatment quality. The system uses non-invasive electrical measurements to give caregivers a continuous assessment of lung function, drastically improving their ability to make data-informed decisions at the bedside. In intensive care settings for the smallest patients, where repeated imaging or invasive sampling carries real risk, a continuous non-invasive window into how blood and air are moving through the lungs represents a meaningful shift in the information available to clinicians during critical moments.

At Binghamton University, Dr. Nancy Guo has developed ClinSegAI, a secure artificial intelligence platform designed to predict molecular biomarkers directly from standard pathology imaging. Turnaround time for treatment decisions remains a major roadblock to efficient cancer care, because molecular characterization of a tumor often requires additional laboratory workflows that delay therapy. ClinSegAI applies machine learning to detect biomarkers within routine histopathology images, rapidly reducing the time needed to inform treatment decisions and marking what the university describes as a major improvement in the standard of care. Because the model works from images that pathologists already produce, it could in principle slot into existing clinical workflows rather than demanding new tissue collection, and the platform’s emphasis on security addresses the data governance concerns that frequently slow the adoption of medical AI systems.

Binghamton’s second award goes to Dr. John Fetse, who is conducting in vivo validation of engineered lipid nanoparticles for RNA therapeutic delivery. RNA therapeutics offer promising treatment avenues for a wide variety of diseases, but current lipid nanoparticle methods suffer from poor delivery efficiency and raise toxicity concerns that limit proper dosing. Fetse’s innovation incorporates amino acids directly into the lipid molecules themselves, a chemical modification that reduces toxicity risk while improving delivery rates. The approach matters because delivery remains the central bottleneck of the entire RNA medicine field: therapeutic RNA is fragile, quickly degraded, and must be escorted into the right cells in sufficient quantities for a meaningful dose while avoiding harmful accumulation elsewhere. Validating such formulations in living organisms is the essential step toward demonstrating that the chemistry performs outside the controlled conditions of cell culture.

The University at Buffalo is home to two projects that pair semiconductor engineering with human health. Dr. Uttam Singisetti is developing gallium oxide semiconductor transistors for electric vehicle and AI power applications, responding to surging demand from electrified transport and the data center boom for electronics that deliver electricity more efficiently. Gallium oxide is a wide-bandgap material, and transistors built from it can handle high voltages and switch power with less wasted energy than conventional silicon devices, improving performance while lowering costs. The second Buffalo project, led by Dr. Margarita L. Dubocovich, targets advanced phase circadian disorders, in which the body’s internal biological clock is misaligned with a person’s daily schedule or environment. Such misalignment elevates the risk of sleep disruption, cardiovascular disease, depression, cancer and chronic pain. Dubocovich is working toward a first-in-class, orally administered therapeutic that realigns the underlying clock mechanism and restores healthy rhythms, rather than merely masking symptoms with sedatives or stimulants.

At Stony Brook University, Dr. M. Hassan Arbab has developed a handheld, portable terahertz spectral imaging scanner for the diagnosis and triage of skin burns. Burn patients frequently undergo multiple reconstructive surgeries because current clinical techniques assess burn depth with only around 60 to 65 percent accuracy, forcing surgeons to make consequential decisions with incomplete information. Arbab’s device achieves 93 to 95 percent or better accuracy, a dramatic expansion of diagnostic capability that allows clinicians to determine which burns will heal on their own and which require intervention. Terahertz radiation sits between microwave and infrared frequencies and is sensitive to the water content and structural changes in tissue, which makes it well suited to distinguishing viable from non-viable skin without contact or ionizing radiation. Stony Brook President Andrea Goldsmith highlighted the device as an exemplar of the campus’s record of translating research breakthroughs into real-world medical diagnostics.

The final award supports Dr. Yamin Li at SUNY Upstate Medical University, who is developing a lung-targeting drug formulation for sepsis-induced acute respiratory distress syndrome, a life-threatening condition causing lung injury and breathing difficulty that affects roughly three million patients annually. Li’s formulation uses lipid nanoparticles engineered to target the lung, aiming to reduce mortality, shorten intensive care unit stays and decrease ventilator use for patients with the condition. The project addresses a critical unmet need, since treatment options for ARDS remain largely supportive. Campus leaders across the system framed the awards in similar terms: University at Albany President Havidán Rodríguez emphasized transforming promising ideas into real-world impact, Binghamton’s Anne D’Alleva pointed to safer and more effective treatments, Buffalo’s Caroline Attardo Genco cited the strength of the university’s innovation ecosystem in semiconductors and the life sciences, and Upstate’s Dr. Mantosh Dewan described the research mission as improving the human condition. Together, the nine projects illustrate how modest, well-targeted seed funding can push a diverse portfolio of early-stage technologies toward the market readiness that attracts the investors and partners needed to bring them to patients.

Subject of Research: SUNY Technology Accelerator Fund seed grants for commercializing university research technologies

Article Title: SUNY Chancellor King announces funding for groundbreaking technologies to improve lives and protect New Yorkers

Article References: SUNY Chancellor King announces funding for groundbreaking technologies to improve lives and protect New Yorkers. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: SUNY, Technology Accelerator Fund, quantum sensing, artificial intelligence, cancer diagnostics, lipid nanoparticles, RNA therapeutics, gallium oxide semiconductors, terahertz imaging, circadian rhythm disorders, ARDS, technology commercialization

A multidisciplinary team at the University of California, Los Angeles has now received a major boost in its effort to resolve this diagnostic dilemma. The group, led by Dr. Steven Dubinett, dean of the David Geffen School of Medicine at UCLA, Associate Vice Chancellor for Research, and an investigator at the UCLA Health Jonsson Comprehensive Cancer Center, has been awarded the inaugural Early Detection Award from the LUNGevity Foundation and the Rising Tide Foundation for Clinical Cancer Research. The $1 million award will fund a three-year study aimed at developing a noninvasive approach to determine whether indeterminate pulmonary nodules are cancerous, potentially transforming how clinicians manage one of the most common and consequential findings in modern chest imaging.

The scale of the problem is difficult to overstate. Lung cancer remains the leading cause of cancer-related death in the United States, claiming more lives each year than colon, breast, and prostate cancers combined. The paradox that has long frustrated oncologists is that lung cancer caught early is often curable, yet the very screening technologies designed to catch it early, most notably low-dose CT scanning, generate enormous numbers of ambiguous findings. A nodule of a few millimeters may be nothing at all, or it may be the first visible sign of a malignancy that will spread within months. Current clinical guidelines rely on nodule size, growth rate, and risk models built from population data, but these tools leave a wide gray zone in which neither aggressive intervention nor simple watchful waiting is clearly correct.

The UCLA team’s strategy is to attack that gray zone from two directions simultaneously. The three-year study will combine advanced computational analysis of CT images, a field known as radiomics, with a blood-based test that examines cell-free DNA methylation. Radiomics uses algorithms to extract quantitative patterns from medical images, features such as texture, shape, and density characteristics that are invisible to the human eye but may correlate with underlying tumor biology. DNA methylation analysis, meanwhile, looks for chemical modifications to fragments of DNA shed into the bloodstream, patterns that can differ between cancer-derived DNA and DNA released by healthy tissue. By integrating these two complementary streams of data, the investigators hope to generate a risk assessment that is more precise than either imaging or molecular testing could achieve alone.

This convergence of disciplines is no accident. The research team brings together experts in cancer biology, pulmonary medicine, computational imaging, artificial intelligence, and molecular diagnostics. Alongside Dr. Dubinett, the project includes Dr. Ramin Salehi-Rad, Dr. Xianghong J. Zhou, Dr. William Hsu, and Dr. Linh M. Tran, each contributing a distinct layer of expertise that spans the biological, technological, and clinical dimensions of the problem. The design reflects a growing conviction in cancer research that the hardest problems in early detection, those sitting at the intersection of imaging physics, genomics, and clinical decision-making, are best solved not by isolated laboratories but by teams whose members can translate findings across disciplinary boundaries in real time.

The clinical validation plan is ambitious in scope. The team will test its combined approach in 500 patients receiving care at UCLA Health and Veterans Affairs medical centers, populations that include both community patients and veterans, a group with elevated lung cancer risk. The central question is whether fusing imaging-derived features with blood-based molecular biomarkers can meaningfully improve risk stratification compared with existing methods. Success would mean that clinicians could more accurately identify which patients require prompt evaluation and treatment, while sparing others from biopsies or surgeries that carry real risks of complications, anxiety, and cost. In a health system where millions of nodules are detected annually, even a modest improvement in diagnostic accuracy could prevent tens of thousands of unnecessary invasive procedures each year.

The implications extend beyond the individual patient encounter. Unnecessary biopsies and surgeries represent a substantial burden on the health care system, both financially and in terms of clinical resources. A bronchoscopy or needle biopsy of the lung carries risks including bleeding, collapsed lung, and infection, and surgical resection of a nodule that proves benign exposes patients to the morbidity of major thoracic surgery for no benefit. Conversely, the cost of a false reassurance, a cancer dismissed as benign that later presents at an advanced stage, is measured in lives. A validated noninvasive test that pushes diagnostic confidence in either direction would allow clinicians to allocate invasive resources where they matter most and to monitor low-risk nodules with greater confidence and less patient anxiety.

The award itself marks a notable milestone for the two supporting organizations. The Early Detection Award is the inaugural grant of its kind jointly offered by the LUNGevity Foundation, the largest nonprofit dedicated to lung cancer research and patient support, and the Rising Tide Foundation for Clinical Cancer Research, which funds translational projects designed to move scientific discoveries toward clinical application. By directing the award toward pulmonary nodule assessment, the funders have targeted what many in the field consider the single most consequential bottleneck in lung cancer early detection: the moment after a scan reveals a spot and before anyone knows what it means.

For Dr. Dubinett, the project represents the culmination of years of work at the frontier of lung cancer biology and early detection research. In a statement accompanying the announcement, he emphasized the collaborative character of the effort. “This award recognizes the power of team science and the value of bringing together experts from multiple disciplines to tackle a critical problem in lung cancer detection,” he said. “By integrating advances in imaging science, artificial intelligence, and molecular diagnostics, we hope to improve assessment of pulmonary nodules and help ensure that patients receive the right care at the right time.” The statement captures the underlying philosophy of the project: that no single technology, however sophisticated, will resolve the nodule dilemma on its own, but that a carefully engineered synthesis of imaging and molecular data might.

If the three-year study succeeds, the consequences could ripple well beyond UCLA. A validated imaging-plus-blood test for nodule assessment would fit naturally into existing lung cancer screening programs, which already perform annual low-dose CT scans on millions of high-risk individuals, and could be deployed at the moment of the very first suspicious finding. It could also inform the design of future early-detection tools for other cancers, where similar combinations of radiomic image analysis and liquid biopsy biomarkers are under active investigation. For the 1.5 million Americans who each year face the anxious limbo of an indeterminate lung nodule, the UCLA team’s work offers the prospect of something deceptively simple yet profoundly valuable: a faster, safer, and more accurate answer to the question every one of them asks, is it cancer?

Subject of Research: Noninvasive detection of cancerous indeterminate pulmonary nodules using CT radiomics and blood-based cell-free DNA methylation analysis

Article Title: UCLA research team awarded $1 million to develop new approach for detecting lung cancer

Article References: UCLA research team awarded $1 million to develop new approach for detecting lung cancer. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: lung cancer, early detection, pulmonary nodules, radiomics, cell-free DNA methylation, artificial intelligence, liquid biopsy, CT imaging, UCLA, LUNGevity Foundation, clinical validation, cancer diagnostics

The biology behind the excitement is genuine. c-MET, the mesenchymal-epithelial transition factor, is a receptor that becomes activated when hepatocyte growth factor binds its extracellular domain, triggering a cascade of downstream signalling through partners such as Gab1, PI3K, MAPK and STAT3. This signalling drives cell motility, proliferation and survival, the hallmarks of invasive growth. In healthy tissue, c-MET expression is low and restricted mainly to epithelial and endothelial cells involved in repair and development. In cancer, however, the pathway is frequently dysregulated, producing receptor overexpression that is tightly linked to metastasis, therapy resistance and poorer survival across multiple tumour types. Reported overexpression reaches 26 to 82 percent in oral cavity squamous cell carcinoma, up to 87 percent in penile cancer, 15 to 67 percent in colorectal cancer, 17 to 81 percent in non-small cell lung cancer, 56 to 80 percent in renal cell carcinoma and 14 to 54 percent in breast cancer.

Critically for imaging, tumour tissue can show c-MET levels up to eleven times higher than surrounding epithelium, with a median ratio of 3.4. Imaging scientists generally consider a target diagnostically useful when it can be detected with a signal-to-background ratio above 1.5 to 2, placing c-MET squarely within the range worth pursuing. The review team conducted their search of PubMed, Embase and Scopus according to PRISMA 2020 guidelines, with the protocol registered on PROSPERO, and ultimately included 63 eligible reports: 50 preclinical animal studies published between 2002 and 2026 and 13 clinical reports from 2015 to 2026. The preclinical literature described 62 unique tracers, of which 44 carried a radioactive label and 18 a fluorescent one, distributed across three broad compound classes: monoclonal antibodies, peptides and small molecules.

Each scaffold carries its own pharmacokinetic personality. Monoclonal antibodies, at roughly 150 kilodaltons, showed the highest binding affinities and, in selected studies, the highest tumour-to-background ratios, but their slow circulation meant tumour uptake peaked three to four days after injection, forcing reliance on long-lived radionuclides such as zirconium-89, used in nearly 62 percent of antibody designs. Radiolabelled antibodies achieved tumour accumulation of 3 to 47 percent of injected dose per gram, with reported tumour-to-background ratios ranging from a dismal 0.1 to an impressive 43. Onartuzumab-based designs dominated this category, accounting for 38 percent of antibody tracers. Fluorescent antibody variants conjugated to IRDye800CW delivered comparable affinities of 1.0 to 1.3 nanomolar and tumour-to-background ratios near 5, peaking four days after injection.

Peptides, by contrast, occupy a pharmacokinetic sweet spot. Weighing between 0.5 and 5 kilodaltons, they combine rapid tumour targeting with fast systemic clearance, producing high contrast within hours and permitting short-lived isotopes such as technetium-99m and fluorine-18. Sixteen radiolabelled peptide tracers were identified, dominated by cMBP-derived designs, with binding affinities spanning 0.9 to 326 nanomolar and the best variants, such as the macrocyclic HiP-8, approaching antibody-like binding at around 1 nanomolar. Tumour uptake generally peaked within one to two hours, with reported tumour accumulation of 0.7 to 9.4 percent of injected dose per gram and tumour-to-background ratios of 1.7 to 20. Fluorescent peptide tracers followed a similar pattern, with Cy5-analogues the most popular fluorophores and reported ratios reaching as high as 33. Small molecules, though fast and capable of crossing cell membranes, showed generally lower tumour accumulation and contrast, with ratios of just 0.2 to 3.3, and remain the least mature class.

Yet when the authors traced which designs actually reached the clinic, the pattern was unambiguous: all four translated tracers were peptides. EMI-137, a macrocyclic 26-amino-acid peptide engineered with intramolecular cyclisation for subnanomolar-range affinity and renal clearance, was the most extensively studied, evaluated across eight clinical reports involving 102 participants in five tumour types. cMBP-ICG, a minimalist 12-amino-acid linear peptide conjugated to indocyanine green, was tested in 60 patients with oral cavity cancer. The PET tracers gallium-68-EMP-100 and gallium-68-MetP together accounted for three reports and 25 participants. The reasons others stalled remain unclear, but the authors point to a combination of affinity, stability, manufacturability, regulatory feasibility and the clinical relevance of the animal models used, factors that tracer performance metrics alone cannot predict.

The clinical data reveal both promise and frustration. EMI-137 enabled fluorescence-guided tumour visualisation with tumour-to-background ratios of 1.3 to 9.7, and in its landmark colorectal application, second-pass fluorescence endoscopy identified nine additional adenomatous lesions, roughly 19 percent, that white-light colonoscopy had missed. In papillary thyroid cancer, fluorescence reclassified disease from unifocal to multifocal in four of five patients by detecting foci as small as 1.4 millimetres invisible on preoperative ultrasound. But specificity suffered badly: benign c-MET-expressing tissues lit up too, yielding a sample-size weighted average sensitivity of 81.7 percent but a specificity of just 39.3 percent across EMI-137 studies. In laparoscopic colorectal surgery, fluorescence discriminated only 4 of 9 primary tumours and detected no nodal metastases despite histological confirmation in over half the patients.

The standout performer was cMBP-ICG, applied topically in oral cavity squamous cell carcinoma. Because the tracer was rinsed into the mouth rather than injected, systemic exposure was minimal and imaging immediate. A randomised controlled trial of 50 patients compared fluorescence-guided biopsy-site selection against conventional white-light inspection in the same patients, and fluorescence won decisively on every metric: sensitivity of 88 versus 65 percent, specificity of 93 versus 76 percent, positive predictive value of 90 versus 68 percent, negative predictive value of 91 versus 74 percent, and overall diagnostic accuracy of 91 versus 72 percent, with statistical significance across the board. Weighted averages across both cMBP-ICG studies showed 85.5 percent sensitivity and 91.1 percent specificity, with tumour-to-background ratios of 2.7 to 4.1. The two PET tracers demonstrated feasible whole-body c-MET imaging, and gallium-68-MetP produced the first clinical evidence of a quantitative correlation between tracer uptake and immunohistochemical c-MET expression, with a correlation coefficient of 0.71. All four tracers showed favourable safety, with only mild adverse events reported.

The review’s authors are candid about the caveats. Risk-of-bias assessment using the ROBINS-I tool judged the clinical evidence to carry serious methodological limitations, stemming from small sample sizes, absent control groups, unblinded outcome assessment and the predominance of single-arm early-phase feasibility studies. Preclinical work leaned heavily on a handful of high c-MET-expressing cell lines in subcutaneous xenografts, models that poorly recapitulate the heterogeneous expression and microenvironmental complexity of real tumours. Only one radiolabelled tracer report and seven fluorescent tracer reports used orthotopic models. Heterogeneity in study designs, outcome definitions and reporting was so great that a formal meta-analysis proved impossible, forcing the team into narrative synthesis with weighted averages that they themselves caution should be interpreted cautiously.

What emerges is a field at an inflection point. The biological rationale for c-MET imaging is solid, the safety profile of the leading tracers is reassuring, and the single randomised trial in oral cancer demonstrates that molecular imaging can genuinely outperform conventional assessment. The path forward, the authors argue, lies in larger, standardised prospective studies designed to prove that c-MET-targeted imaging improves clinically meaningful endpoints: lesion detection, staging accuracy, treatment selection and surgical decision-making. They also point toward synergy with therapy, noting that c-MET status already guides first-line treatment with MET tyrosine kinase inhibitors and antibody-drug conjugates in non-small cell lung cancer, raising the prospect of matched imaging and therapeutic agents in a theranostic paradigm. Until such evidence arrives, the 62-tracer graveyard of preclinical promise stands as a warning that in molecular imaging, a beautiful animal study is only the beginning of a very long road.

Subject of Research: Systematic review of preclinical and clinical molecular imaging tracers targeting the c-MET receptor in cancer

Article Title: Molecular imaging tracers targeting c-MET: a systematic review of preclinical evidence and clinical translation

Article References: Verdijk, R. W. A., van der Mierde, S. M., Berehova, N., van Meerbeek, M. P., Brouwer, O. R., van der Poel, H. G., van Leeuwen, F. W. B., & Buckle, T. (2026). Molecular imaging tracers targeting c-MET: a systematic review of preclinical evidence and clinical translation. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08174-w

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08174-w

Keywords: c-MET, molecular imaging, fluorescence-guided surgery, PET imaging, EMI-137, cMBP-ICG, clinical translation, systematic review, tumour tracers, image-guided surgery, receptor tyrosine kinase, cancer diagnostics

Now, a team of researchers at the U.S. Food and Drug Administration’s Division of Imaging, Diagnostics, and Software Reliability has developed a new tool designed to catch these flaws before they can do damage. The system, called HistoART, is described in a study published in the journal Neural Computing and Applications. Led by Seyed Kahaki, with contributions from Alexander Webber, Ghada Zamzmi, Adarsh Subbaswamy, Rucha Deshpande, and Aldo Badano, the work compares three fundamentally different strategies for detecting artifacts in whole slide images, and finds that a large foundation model originally trained for general pathology tasks outperforms both a conventional deep learning network and a classical, hand-engineered approach.

The stakes are higher than they might first appear. As hospitals adopt digital pathology workflows, the reliability of downstream image analysis tasks, from tumor detection to grading, depends on the quality of the images feeding into them. An algorithm trained to recognize cancer cells may misinterpret a tissue fold as a suspicious structure, or an air bubble as a region of lost tissue. Previous studies have documented how quality control, or the lack of it, directly affects the accuracy of computational pathology systems. Yet many existing quality control tools rely on fixed rules and handcrafted measurements that may not generalize across the wide variety of scanners, staining protocols, and tissue types encountered in real-world laboratories.

HistoART tackles the problem with a three-way comparison. The first approach, the foundation model-based approach, fine-tunes UNI, a general-purpose foundation model for computational pathology that has been trained on massive collections of pathology images. Foundation models of this kind learn rich, general-purpose visual representations that can be adapted to many downstream tasks with relatively little additional training. The second approach, the deep learning approach, is built on a ResNet50 backbone, a widely used convolutional neural network architecture originally developed for general image recognition. The third, the knowledge-based approach, dispenses with learned representations altogether and instead relies on handcrafted features derived from texture, color, and frequency-based metrics, drawing on decades of classical image analysis research.

All three approaches were trained and evaluated to detect six of the most prevalent artifact types in whole slide images: tissue folds, out-of-focus regions, air bubbles, tissue damage, marker traces, and blood contamination. Each of these artifacts arises at a different stage of the workflow. Tissue folds occur when sections of tissue wrinkle during mounting on the slide. Out-of-focus regions result from imperfect autofocus during scanning. Air bubbles become trapped under the coverslip, marker traces are left by pens used to label slides, and blood contamination can obscure underlying tissue architecture. Detecting such a heterogeneous set of defects demands a system capable of recognizing very different visual patterns, from the sharp linear creases of a fold to the diffuse haze of an out-of-focus patch.

The evaluation was deliberately broad. The researchers assembled a dataset of more than 50,000 image patches sourced from diverse whole slide imaging scanners, including instruments from Hamamatsu, Philips, and Leica Aperio AT2, and drawn from multiple imaging sites. This diversity matters because artifacts can look different depending on the scanner that produced the image, and a detection system that works well on one manufacturer’s output may fail on another’s. By testing across scanners and sites, the team aimed to assess how well each approach would generalize beyond a single laboratory environment, a persistent weakness of many published machine learning studies in pathology.

The results were striking. The foundation model-based approach achieved a patch-wise area under the receiver operating characteristic curve, or AUROC, of 0.995, with a 95 percent confidence interval of 0.994 to 0.995, a level of performance that approaches the theoretical ceiling of a perfect classifier. The ResNet50-based deep learning approach reached an AUROC of 0.977, with a confidence interval of 0.977 to 0.978, while the knowledge-based approach, built on handcrafted features, trailed at 0.940, with a confidence interval of 0.933 to 0.946. The AUROC metric measures a classifier’s ability to distinguish between positive and negative cases across all possible decision thresholds, so the gap between 0.995 and 0.940 represents a substantial difference in real-world reliability, particularly for the subtle or ambiguous artifacts that matter most in clinical practice.

Why would a foundation model so decisively outperform a purpose-built convolutional network? The likely answer lies in the breadth of its pretraining. UNI was trained on enormous and varied collections of pathology imagery, meaning it has already encountered an extraordinary range of tissue appearances, staining variations, and imaging conditions before any artifact-specific fine-tuning begins. When subsequently adapted to the artifact detection task, it brings that prior knowledge to bear, allowing it to distinguish, say, a genuine tissue fold from a naturally occurring tissue edge with greater confidence than a network learning visual features from scratch. The handcrafted approach, meanwhile, is limited by the imagination of its designers: texture descriptors, color statistics, and frequency-domain measures capture certain artifact signatures well but may miss patterns that do not fit predefined mathematical descriptions.

Detection alone, however, is not enough. A laboratory needs to know what to do with the information. To bridge the gap between detection and actionable conclusions, the team developed a quality report scorecard that quantifies the number of high-quality image patches in a dataset and visualizes the distribution of artifact subgroups. This reporting layer transforms raw per-patch predictions into a summary that laboratory staff and algorithm developers can act upon: a slide riddled with folds can be rescanned or excluded, a scanner producing systematic out-of-focus regions can be serviced, and a dataset destined for training a diagnostic model can be filtered to remove compromised patches. In this way, HistoART functions not merely as a classifier but as a quality assurance pipeline for the entire digital pathology workflow.

The researchers describe the resulting system as a multi-branch pipeline that may enhance the reliability of whole slide image analysis by providing a scalable and interpretable approach for improving digital pathology workflows. Interpretability is a key selling point of the scorecard design, since regulators and clinicians are often wary of black-box systems that flag images without explaining why. By visualizing which artifact types dominate a slide or dataset, the tool offers a transparent window into the failure modes of the imaging process. Consistent with the team’s commitment to open science, all the data and methods from the work are available for download on GitHub, allowing other laboratories to adopt, scrutinize, and extend the system. As whole slide imaging continues its march toward becoming the default medium of diagnostic pathology, tools like HistoART suggest that the quality of the digital slide, long treated as an afterthought, is finally being engineered with the same rigor as the diagnoses that depend on it.

Subject of Research: Artifact detection in whole slide histopathology images using foundation models, deep learning, and handcrafted features

Article Title: HistoART: Histopathology artifact detection based on large foundation model

Article References: Kahaki, S., Webber, A., Zamzmi, G., Subbaswamy, A., Deshpande, R., & Badano, A. (2026). HistoART: Histopathology artifact detection based on large foundation model. Neural Computing and Applications, 38(18), Article 753. https://doi.org/10.1007/s00521-026-12454-9

Image Credits: AI Generated

DOI: 10.1007/s00521-026-12454-9

Keywords: histopathology, whole slide imaging, artifact detection, foundation model, UNI, ResNet50, digital pathology, quality control, machine learning, AUROC, cancer diagnostics, image analysis

Methylation is one of biology’s most powerful control systems. In DNA, the addition of a methyl group to the fifth position of cytosine, forming 5-methylcytosine (5mC), is catalyzed by DNA methyltransferases and governs transcriptional repression, genomic imprinting, chromatin accessibility, and cellular memory. Enzymes of the TET family further oxidize 5mC into 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxylcytosine, creating a layered code of demethylation intermediates. When these patterns go awry—through hypermethylation of tumor suppressor promoters such as RASSF1A, SHOX2, and SFRP1, or global hypomethylation of repetitive elements—the result can be gene silencing, genome instability, and malignant transformation. On the RNA side, N6-methyladenosine (m6A), the most abundant internal modification in messenger RNA, dynamically regulates splicing, stability, translation, and immune recognition, while related marks such as m5C, m1A, m7G, and m3C diversify the epitranscriptomic landscape and are increasingly implicated in cancer, neurological disease, and viral infection.

Despite decades of biological insight, measuring these marks in the clinic remains stubbornly difficult. The gold standard for DNA methylation, bisulfite sequencing, relies on harsh chemical treatment that degrades DNA and destroys sequence complexity—a fatal flaw when the sample is scarce, fragmented cell-free DNA from a blood draw. Methylation-specific PCR and pyrosequencing are locus-limited, while antibody-based enrichment methods lack single-base resolution. RNA methylation detection fares worse: MeRIP-seq maps modified regions only to roughly 100-nucleotide windows, mass spectrometry offers global averages without sequence context, and the definitive single-site method, SCARLET, is so labor-intensive and low-throughput that it is essentially confined to specialist laboratories. These constraints have kept methylation biomarkers largely out of point-of-care and decentralized testing, despite their measurable presence in body fluids.

CRISPR-based biosensing offers a radical alternative. Cas12 and Cas13 effectors, the same enzymes famous for genome editing, carry collateral cleavage activities: once activated by a matching target, they indiscriminately shred nearby fluorescent reporters, generating an amplified signal within minutes. Because guide RNAs can be designed against virtually any sequence, the recognition step is programmable by design. Crucially, mechanistic studies have revealed that methylation itself modulates CRISPR activation—methylated DNA can weaken Cas12a activation relative to unmethylated DNA, and m6A can alter reverse-transcription read-through or reshape RNA structure in ways that Cas13 can detect. Coupled with cascade amplification strategies such as Cas13-to-Cas12 or Cas13-to-Csm6, these systems reach attomolar sensitivity, capable of distinguishing methylation states across both DNA and RNA substrates.

For DNA methylation, the review identifies three mechanistically distinct paradigms. Chemical-conversion approaches rewrite methylation status into sequence changes: bisulfite treatment converts unmethylated cytosines to uracil, turning methylation differences into SNP-like targets that platforms such as HOLMESv2, built on the mismatch-sensitive Cas12b, can quantitatively resolve. Gentler bisulfite-free variants, exemplified by meHOLMES, use TET oxidation followed by APOBEC3A deamination to preserve DNA integrity while achieving comparable discrimination. Restriction-enzyme strategies employ methylation-sensitive nucleases like HpaII and HhaI, or the methylation-dependent enzyme GlaI, to selectively triage templates before amplification and Cas12 readout—architectures that have detected the SEPT9 colorectal cancer biomarker in human serum at sensitivities reaching 86.4 attomolar. Most strikingly, direct methylation-responsive detection eliminates preprocessing altogether: in the CRISPR-MeDNA Test, 5mC at specific positions within the crRNA-target duplex suppresses Cas12a’s collateral cleavage, allowing amplification-free discrimination of methylated cfDNA in plasma from cancer patients and healthy controls.

Detecting oxidized cytosine derivatives presents its own challenges, since 5hmC, 5fC, and 5caC preserve normal Watson-Crick base pairing and thus evade direct sequence recognition. Current solutions rely on modification-selective enrichment: in the CAICas12a system, 5hmC-containing DNA is biotin-labeled, captured on magnetic beads, amplified by rolling-circle amplification, and then read out by Cas12a with a detection limit of 11 femtomolar. Analogous approaches for the rarer derivatives 5fC and 5caC remain largely undeveloped, representing a clear gap in the field.

RNA methylation sensing exploits an even broader repertoire of signal-conversion chemistry. Reverse-transcription-mediated Cas12a assays convert modification-dependent polymerase pausing, truncation, or misincorporation into methylation-readable cDNA signatures; the XNA-RT-Cas12a system achieves single-nucleotide m6A discrimination without chemical derivatization by exploiting the destabilizing effect of m6A on A-U pairing. Alternatively, the CRISPRm6A assay uses the enzyme MazF, which cleaves only unmethylated ACA motifs, so that m6A-protected RNA survives to trigger a Cas13a-Csm6 tandem amplification cascade—preamplification-free and single-base precise. Direct structure-based sensing is also emerging: because m6A induces local conformational rearrangements known as m6A switches, and Cas13 activation is exquisitely sensitive to target accessibility, the enzyme can function as a structure-responsive biosensor. The principle extends to m1A, whose positive charge disrupts Watson-Crick pairing and directly suppresses Cas13a collateral cleavage, and potentially to m5C and pseudouridine, though detection of these marks remains nascent.

The translational stakes are high. Commercial DNA methylation diagnostics—including Cologuard for colorectal cancer screening, EpiCheck for bladder cancer surveillance, and confirmMDx for prostate cancer—have already validated the clinical utility of methylation biomarkers, yet all depend on bisulfite chemistry, centralized laboratories, and multistep workflows. CRISPR methylation assays, by contrast, could combine programmable recognition, rapid fluorescent or lateral-flow readout, and compatibility with short cfDNA fragments in portable formats. But the review is candid about the distance remaining: nearly all reported platforms exist at proof-of-concept stage, tested on synthetic targets or limited clinical specimens, with none approved as a clinical assay. Fragmented cfDNA can sever methylation sites from the PAM sequences Cas enzymes require; PAM-relaxed Cas variants may expand coverage at the cost of specificity; and collateral-cleavage signals are notoriously sensitive to guide efficiency, reaction kinetics, and sample-matrix inhibition, complicating quantification and multiplexing.

The path forward, the authors argue, runs through engineering, standardization, and rigorous clinical validation. Near-term priorities include amplification-free detection of rare methylated molecules using more active Cas effectors and optimized guide designs; multiplexed panels combining Cas12 and Cas13 to simultaneously profile 5mC, 5hmC, m6A, and m1A; and artificial-intelligence tools, from deep-learning guide-activity predictors to AlphaFold 3 modeling of Cas-nucleic-acid complexes, to accelerate assay design. Longer term, integration with droplet microfluidics, single-cell sequencing, and Cas9-guided nanopore enrichment could push the technology toward spatially resolved and long-read epigenetic profiling. Manufacturing will demand defined quality attributes for Cas proteins, guide RNAs, and methylation-processing enzymes, along with commutable reference materials that currently do not exist—particularly for RNA standards. Regulatory planning, the review stresses, must begin early, since intended use, specimen type, and testing environment determine the evidentiary bar.

The authors conclude that CRISPR methylation detection is unlikely to replace PCR- and sequencing-based methods outright, but its greatest near-term value may lie in rapid, targeted, decentralized testing where conventional infrastructure is limited—from community clinics to low-resource settings. If the field can convert its remarkable analytical sensitivity into reproducible, scalable, and quantitatively calibrated workflows, CRISPR-based methylation sensing could finally bring the epigenome—and the epitranscriptome—into routine, minimally invasive diagnostics, turning a subtle layer of chemical biology into a practical clinical instrument for early cancer detection and precision medicine.

Subject of Research: CRISPR-based biosensing platforms for detecting DNA and RNA methylation as epigenetic diagnostic biomarkers

Article Title: CRISPR technologies for detecting DNA and RNA methylation: Mechanisms, platforms, and translational opportunities

Article References: Chen, K., Yang, B., Sang, R., Chen, W., Xiang, T., & Deng, F. (2026). CRISPR technologies for detecting DNA and RNA methylation: Mechanisms, platforms, and translational opportunities. Bioengineering & Translational Medicine, Article e70174. https://doi.org/10.1002/btm2.70174

Image Credits: AI Generated

DOI: 10.1002/btm2.70174

Keywords: CRISPR, DNA methylation, RNA methylation, m6A, Cas12a, Cas13, epigenetics, liquid biopsy, biosensors, cancer diagnostics, 5-methylcytosine, epitranscriptomics

The problem the researchers set out to solve is one of the most persistent dilemmas in urologic oncology. PSA, the workhorse of prostate cancer screening for more than three decades, cannot reliably distinguish slow-growing tumors that may never harm a patient from aggressive cancers that demand immediate treatment. The consequence is a diagnostic cycle of overdiagnosis and overtreatment: elevated PSA triggers magnetic resonance imaging, and suspicious findings or persistent uncertainty lead to needle biopsies that frequently detect only indolent disease, or nothing at all. Meanwhile, a minority of genuinely dangerous cancers slip through the cracks. Clinically significant prostate cancer, defined by tumor grade and volume thresholds that correlate with metastasis and death, is the target that any meaningful diagnostic must hit without flagging every benign enlargement of the gland.

The 19-biomarker model approaches the problem from an entirely different angle. Rather than relying on a single protein produced by prostate tissue, it analyzes a panel of urinary peptides, fragments of proteins shed into urine as they are filtered through the kidneys or released from the urinary tract. Proteolysis, the enzymatic cleavage of proteins, is altered in cancer, and the resulting peptide patterns act as a downstream readout of tumor biology. Capillary electrophoresis separates these peptides with extraordinary resolution, and mass spectrometry identifies and quantifies each one by its mass-to-charge ratio. The 19 selected markers, combined into a weighted score, capture this proteolytic signature of malignancy in a non-invasive sample that patients provide without needles, discomfort, or radiation.

The prospective design is what gives the new findings their weight. Biomarker studies often falter when moved from retrospective datasets, where models are trained and tested on the same or similar patients, into real-world clinical settings. Here, consecutive biopsy-naïve patients with clinical suspicion of prostate cancer at two major Spanish hospitals underwent urine collection, multiparametric magnetic resonance imaging (mpMRI), and standard biopsy, allowing the test’s performance to be measured against gold-standard pathology rather than against itself. The researchers benchmarked the 19-biomarker model not only against PSA but also against the ERSPC risk calculator, prostate-specific antigen density (PSAD), and mpMRI. Every comparison favored the urine test: ERSPC reached an AUC of 0.63 (p = 0.0108), PSAD 0.61 (p = 0.0021), and mpMRI 0.65 (p = 0.0069), all substantially below the proteomics score.

Perhaps the most clinically striking result emerged in the subgroup of men whose mpMRI scans showed low PI-RADS scores, meaning radiologists saw little or nothing suspicious on imaging. In these patients, where the traditional pathway would often reassure both doctor and patient, the 19-biomarker model detected clinically significant cancer with 75 percent sensitivity and 90 percent specificity. This matters because MRI-invisible tumors are a recognized blind spot of modern diagnostics; studies have shown that a meaningful fraction of significant cancers go undetected when biopsy decisions rest solely on imaging findings. A urine test that can flag danger the scanner misses offers a genuine safety net, potentially catching aggressive disease before it has the chance to progress beyond cure.

The study also prospectively validated a previously described nomogram that combines the proteomics score with mpMRI results. The combined model achieved an AUC of 0.82, with 90 percent specificity and 64 percent sensitivity for clinically significant disease. The logic of this integration is important: imaging excels at locating lesions and estimating their burden, while the proteomic signature reflects biological aggressiveness independent of what is visible on the scan. Fusing the two layers of information produces a risk estimate more accurate than either alone, mirroring a broader shift in oncology toward multimodal risk stratification, where molecular, imaging, and clinical data are woven into a single decision framework.

To probe whether the test reflects true disease biology rather than statistical noise, the team turned to 100 patients who underwent radical prostatectomy, the surgical removal of the prostate. They correlated preoperative urine scores with the final pathology of the excised gland, including the International Society of Urological Pathology (ISUP) grade groups. The association between higher biomarker scores and more aggressive tumor features provides biological plausibility for the diagnostic signal, suggesting that the urinary peptides are not merely markers of prostate enlargement or inflammation but genuine echoes of malignant transformation and tumor grade.

Performance metrics alone do not determine clinical value; what matters is whether a test changes decisions in ways that benefit patients. A sensitivity of 75 percent and specificity of 90 percent in MRI-low-suspicion patients means that a negative urine result would allow many men to safely defer biopsy, avoiding the bleeding, infection, and sexual and urinary side effects that accompany prostate needle procedures, while a positive result directs resources toward those most likely to harbor significant disease. At the population level, where millions of men undergo PSA testing annually and a substantial fraction proceed to MRI and biopsy, even modest reductions in unnecessary procedures would translate into enormous savings in morbidity, anxiety, and healthcare expenditure.

The road from validation to routine practice still requires broader, multi-center, ethnically diverse cohorts and regulatory assessment, and the authors acknowledge limitations inherent in any single-center-pair study of moderate size. Yet the prospect of a urine-based test that rivals and exceeds the performance of PSA, risk calculators, and imaging represents a genuine inflection point in prostate cancer diagnostics. As proteomics matures from a discovery science into a clinical discipline, the 19-biomarker model stands as a demonstration that the molecular information flowing quietly through the human urinary tract can be harnessed to answer one of medicine’s most consequential questions: which cancers need to be found, and which needles can finally be left in the drawer.

Subject of Research: Prospective clinical validation of a urine-based proteomics biomarker test for predicting clinically significant prostate cancer in biopsy-naïve patients

Article Title: Prospective validation of a urine-based proteomics test for predicting clinically significant prostate cancer in biopsy-naïve patients

Article References: Morillo, A. C., Jobre, K. N., Lendinez Cano, G., Blanca-Pedregosa, A., Lopez Ruiz, D., Parada, J., Heidegger, I., Culig, Z., Lopez-Beltran, A., Carrasco-Valiente, J., Mischak, H., Medina, R. A., Campos Hernandez, J. P., Frantzi, M., & Gómez Gómez, E. (2026). Prospective validation of a urine-based proteomics test for predicting clinically significant prostate cancer in biopsy-naïve patients. British Journal of Cancer. https://doi.org/10.1038/s41416-026-03602-y

Image Credits: AI Generated

DOI: 10.1038/s41416-026-03602-y

Keywords: prostate cancer, urine test, proteomics, biomarkers, PSA, mass spectrometry, mpMRI, biopsy, clinical validation, cancer diagnostics, PI-RADS, liquid biopsy

Cancer cells frequently alter the number of copies they carry of particular DNA segments. A deletion can remove a gene that restrains cell growth, while a gain or amplification can increase the dosage of an oncogene, intensifying signals that promote proliferation or survival. These changes, collectively called copy-number variations, may span an entire chromosome or be confined to a small genomic region containing a clinically important gene. Their patterns can help identify tumor types, distinguish biologically different disease subgroups and reveal potential treatment targets. DNA methylation arrays were originally developed primarily to measure chemical tags attached to DNA, especially methyl groups that influence gene regulation. Yet the same arrays also provide indirect information about copy number because the intensity of signals from thousands of genomic probes changes when DNA is gained or lost. CnQuant is designed to extract and organize that secondary signal, turning a widely used epigenetic assay into a broader genomic profiling tool.

The platform builds on the team’s earlier EpiDiP system and incorporates the Mepylome toolkit for processing methylation and copy-number data. Its architecture separates analysis services from the visual interfaces used by clinicians. A coordinating component, CQmanager, directs files through a local application programming interface and can be incorporated into existing diagnostic workflows, including hospital systems that must keep patient data on site. CQcalc calculates copy-number alterations using array-specific, gender-balanced reference data, which are essential because normal signal levels vary between platforms and can be affected by sex-chromosome composition. The software stores reference information in compressed, checksum-verified form to reduce storage and computational demands. Once profiles have been generated, CQall_plotter can overlay data from multiple samples and array types, while the CQall and CQcase interfaces present cohort-level and individual-patient views through a web-based graphical environment.

That distinction between population patterns and single-patient inspection is central to CnQuant’s clinical design. CQall functions as an atlas of recurrent abnormalities, allowing users to examine how often a chromosomal gain or deletion appears within a reference cohort. Such frequency information can provide a plausibility check when a new diagnostic result seems unusual, and it may also help researchers investigate the genomic architecture of rare tumors. CQcase focuses on one specimen at a time, displaying selected genomic regions at high resolution and attaching gene-level annotations to the plot. A clinician can therefore move from a broad chromosome-wide pattern to a specific locus, such as ERBB2, MDM2 or PDGFRA, or to tumor-suppressor regions including CDKN2A. The interface is intended to be usable without specialist bioinformatics training. Annotated plots can be downloaded into electronic health records or shared through links during multidisciplinary tumor-board discussions, although the researchers emphasize that the links preserve visualization and annotation rather than exposing identifying patient information.

The system’s reference strategy also addresses a subtle problem in comparing data produced by different generations of methylation arrays. A reference cohort may combine samples analyzed on the older HumanMethylation450K platform with samples processed on EPIC arrays, whose probe content is not identical. If a comparison uses probes present on only one platform, apparent differences may reflect technology rather than tumor biology. CnQuant therefore restricts cross-platform cohort analyses to genomic probes shared by all included array types. That choice can reduce the number of measurements available, but it makes comparisons more conservative and helps avoid misleading conclusions, particularly in rare tumor entities where cohorts are small and heterogeneous. The researchers report that the software supports conventional methylation-array versions and supplies the corresponding copy-number-neutral reference data, enabling a unified approach rather than requiring each laboratory to assemble its own normalization framework.

Examples shown by the investigators illustrate how chromosomal signatures can mirror recognized tumor biology. In posterior fossa pilocytic astrocytomas, the platform identified recurrent gain of chromosome 7 associated with an internal tandem duplication. In diffuse midline gliomas carrying H3K27 alterations, it highlighted frequent gain involving PDGFRA. A recurrent loss of chromosome 7 together with gain of chromosome 10 appeared in RTK II glioblastomas that lacked IDH mutations, while sporadic amplification of the ERBB2 locus was visible in breast carcinomas. These patterns are not, by themselves, substitutes for a complete diagnosis. Instead, they provide genomic context that can be interpreted alongside histology, methylation-based tumor classification, sequencing and immunohistochemistry. At the individual-gene level, a copy-number plot may help explain a high or low variant allelic frequency in parallel sequencing, or clarify whether an apparent sequencing signal is consistent with a deletion, duplication or amplification in the surrounding DNA.

To test whether the visual output corresponded to established clinical results, the team examined four groups of tumors: breast-cancer metastases, H3K27-altered diffuse midline gliomas, IDH-wild-type RTK II glioblastomas and posterior fossa pilocytic astrocytomas. The 30 cases contained oncologically relevant copy-number changes and had already been assessed using accredited diagnostic methods. Depending on the tumor, the comparison data came from second-generation DNA or RNA sequencing panels, including the Oncomine Comprehensive Assay V2 and Archer FUSIONPlex Core Solid Tumor panel, or from HER2 fluorescence in situ hybridization and immunohistochemistry. Across 155 genomic loci assessed by visual comparison, 153 matched the routine results. The 98.71 percent figure is encouraging, but it comes from a small internal study and reflects visual concordance rather than a large prospective clinical trial with prespecified performance measures. Broader testing across institutions, tumor types and sample qualities will be needed before the software’s clinical reliability can be fully established.

CnQuant could also extend copy-number interpretation beyond microarrays, according to its developers. The cohort atlas may serve as a reference when clinicians interpret targeted sequencing or newer nanopore sequencing results, both of which can produce ambiguous evidence for gains and losses depending on coverage, assay design and tumor purity. Because the platform is locally installable through Docker or Windows Subsystem for Linux, it does not require a cloud-based analysis service or extensive computing infrastructure. Its open-source code and public reference resources may make it easier for laboratories to inspect, adapt and integrate the system into their own workflows. The authors argue that existing tools, including Conumee 2.0 and SeSAMe, do not combine interactive graphical exploration, on-the-fly gene annotation, high processing speed and low resource requirements in the same way. If independent validation confirms the initial results, a tool that makes chromosome-scale abnormalities immediately visible could help transform copy-number data from an underused by-product of methylation testing into a practical component of personalized cancer diagnosis and treatment planning.

Traditionally, THz imaging techniques have primarily exploited contrasts based on water content differences to distinguish healthy from diseased tissues. While this approach has been somewhat effective, it falls short when confronting the intricate heterogeneity found in pathological conditions like cancer and burn injuries. The reliance on hydration levels oversimplifies tissue complexity, often masking crucial microstructural changes that could serve as reliable biomarkers of disease progression. Polarimetric measurements of THz waves—analyzing changes in wave polarization after interaction with tissue—offer a promising alternative, capable of capturing nuanced architectural features. However, the biophysical mechanisms underlying these polarization changes remained elusive until the recent computational explorations provided new clarity.

The research team harnessed Monte Carlo simulations—a statistical technique well-suited for modeling complex scattering phenomena—to explore how THz waves interact with microscopic spherical particles embedded in strongly absorbing biological media. These particles effectively represent key pathological structures found in diseased tissue, such as clusters of tumor cells or the damaged microstructures seen in burns, including the destruction of hair follicles and sweat glands. The simulations revealed that both the intensity of diffusely scattered THz light and its degree of polarization exhibit predictable variations depending on the size and concentration of these scatterers. Intriguingly, these signatures enabled the characterization of tissue polarimetric properties through a single polarization measurement, streamlining what previously demanded multiple, complex measurements.

Complementing their simulations, the team manufactured tissue phantoms composed of gelatin imbued with polypropylene spheres varying in size to emulate the optical properties and scattering behavior of real tissue. These experimental validations confirmed the computational predictions: larger spheres consistently yielded stronger scattered light intensity and displayed characteristic polarization dips at specific terahertz frequencies. This frequency-dependent polarimetric response sets a foundation for non-destructive, detailed tissue assessment, which could dramatically enhance diagnostic accuracy in clinical settings.

The researchers further demonstrated the clinical relevance of their approach by applying THz polarimetric imaging to porcine skin samples with induced burns, uncovering distinctive contrast between injured and healthy tissue zones. This capability suggests that THz scattering and polarimetric measurements can serve as sensitive indicators of tissue damage, holding promise for monitoring wound healing and assessing burn severity without invasive biopsies or staining—techniques currently standard in medicine but often time-consuming and resource-intensive.

Importantly, the study’s findings extend beyond burn diagnostics, offering new avenues for oncological applications. Early detection of tumor budding, where small clusters of malignant cells dissociate from the primary tumor mass, is critical for prognosis and treatment planning. Traditional detection relies on biopsy coupled with histological staining, procedures that are not only invasive but also subject to sampling errors. THz polarimetric imaging’s ability to visualize microscopic clusters through inherent tissue scattering properties presents an innovative, potentially faster diagnostic pathway, bypassing lengthy sample preparation while maintaining high sensitivity.

From a technical perspective, the study underscores the power of combining advanced computational physics with experimental optics. Monte Carlo models account for the diffuse, multiple scattering environments typical of biological tissues, a challenging scenario that hampers many conventional imaging techniques. By simulating polarized THz light’s complex interactions with tissue phantoms mimicking realistic absorption and scattering conditions, the researchers not only demystified the origins of polarimetric signals but also established quantifiable relationships between tissue microstructure and measurable optical parameters.

Looking forward, the research group plans to expand their investigations into actual cancer tissue samples, deepening the understanding of how THz polarimetric signals correlate with diverse pathological features. The development of broadband THz systems will further enable resolution of even smaller tissue structures—potentially as minute as 10 to 30 micrometers—thereby broadening the scope of detectible disease-related changes. Such advances could usher in a new paradigm of label-free, real-time tissue characterization with broad implications for early diagnosis and personalized medicine.

The implications for the medical field are profound: by offering a non-invasive, rapid, and sensitive diagnostic method, THz polarimetric imaging could reduce dependency on biopsies, lower healthcare costs, and increase patient comfort. Moreover, as THz technology matures, integration into clinical workflows might enable continuous, bedside monitoring of disease progression or therapeutic response, a feat still unachievable with many existing imaging modalities.

This study marks a significant milestone in medical optics, bridging theoretical physics, computational modeling, and experimental validation to harness the full diagnostic potential of terahertz waves. As the field moves forward, collaboration among optical physicists, engineers, and clinicians will be essential to translate these promising discoveries into effective tools for daily medical practice, potentially transforming cancer detection, burn assessment, and beyond.

In summary, the research lays out a comprehensive framework for understanding and exploiting THz Mie scattering and polarization phenomena in tissues, backed by rigorous simulation and corroborated through experimental imaging. By illuminating the subtle, yet diagnostically meaningful, variations in tissue microstructure through a novel optical window, this work sets the stage for a new generation of medical imaging technologies with remarkable sensitivity, specificity, and clinical impact.

Subject of Research: Human tissue samples
Article Title: Terahertz Mie scattering in tissue: diffuse polarimetric imaging and Monte Carlo validation in highly attenuating media models
News Publication Date: 4-Jun-2025
Web References: https://www.spiedigitallibrary.org/journals/journal-of-biomedical-optics/volume-30/issue-06/066001/Terahertz-Mie-scattering-in-tissue–diffuse-polarimetric-imaging-and/10.1117/1.JBO.30.6.066001.full
References: E. Heller et al., “Terahertz Mie scattering in tissue: diffuse polarimetric imaging and Monte Carlo validation in highly attenuating media models,” J. Biomed. Opt. 30(6), 066001 (2025). DOI: 10.1117/1.JBO.30.6.066001
Image Credits: Heller et al., doi 10.1117/1.JBO.30.6.066001

Keywords

Imaging, Oncology, Applied optics, Medical tests, Tissue damage

Blood plasma, the liquid component of blood, is composed of numerous molecules, including proteins, metabolites, lipids, and salts. This rich mixture serves as a carrier for thousands of biomolecules that reflect the physiological state of the body and can potentially provide critical insights into various health conditions. For example, the presence of elevated prostate-specific antigen levels has long been associated with prostate cancer screening. In light of these biochemical markers’ potential, scientists have succeeded in creating a method that analyzes a wide-ranging array of molecules within plasma to establish specific patterns characteristic of different cancers.

Researchers, led by Mihaela Žigman, harnessed a technique known as electric-field molecular fingerprinting, which utilizes ultra-short bursts of infrared light to probe the complex molecular compositions found in blood plasma. Their study analyzed plasma samples from a robust cohort of 2,533 participants, which included individuals diagnosed with lung, prostate, breast, or bladder cancer, as well as those without any cancer diagnosis. By applying this novel technique, the researchers recorded the unique patterns of light emitted by the molecular mixtures in the plasma, thus creating what they termed an "infrared molecular fingerprint."

The insightful work did not merely stop at capturing these fingerprints. The next step involved employing machine learning technologies to decode and analyze the complex patterns of light associated with cancer and non-cancer samples. A sophisticated computer model was trained using these molecular signatures to learn the distinctions between the varying states of health and disease. This machine learning framework was subsequently tested on an independent sample subset to gauge its efficacy on unseen data, revealing a notable accuracy rate of up to 81% in correctly identifying lung cancer-specific infrared signatures.

This achievement represents a pivotal moment in oncological diagnostics, with the research highlighting the ability of the electric-field molecular fingerprinting technique to detect specific cancer signatures effectively. However, the research also illuminated challenges, as the machine learning model exhibited lower success rates when it came to identifying the other types of cancer within the study. With ongoing advancements and refinements, the researchers aim to broaden this technology’s application, targeting additional types of cancers and various other health conditions, underlining the technique’s substantial potential in future medical diagnostics.

Žigman commented on the significance of their findings, stating, "Laser-based infrared molecular fingerprinting detects cancer, demonstrating its potential for clinical diagnostics." The team emphasizes that with further technological refinements and independent validation through adequately powered clinical studies, this innovative method could reshape the landscape of cancer diagnosis and screening, offering quicker and less invasive options to patients.

The study is not merely an academic exercise; it holds the promise of fostering a paradigm shift in how we approach cancer diagnostics. The ability to quickly identify the presence of cancerous conditions using a simple blood draw could pave the way for not only timely interventions but also reduced healthcare costs associated with more traditional diagnostic methods. Furthermore, the implications of this work could extend beyond oncology, setting the foundation for similar approaches in addressing other health issues characterized by unique molecular fingerprints in blood plasma.

This significant research highlights the intersection of advanced technology and medical science, showcasing how machine learning and novel analytical techniques can collaborate to enhance patient care. As the scientific community continues to explore and validate these innovative approaches, it remains to be seen how rapidly they will integrate into everyday medical practice and what transformative impacts they will have on patient outcomes.

In conclusion, this pioneering research encapsulates the profound potential of leveraging pulsed infrared light in the early detection of cancer, a field where every moment counts. As the findings from the study are further validated and refined, they may usher in a new era of cancer diagnostics characterized by accuracy, efficiency, and patient-centered care. The collaboration of various technological advancements in medicine reflects hope for a future where cancer can be diagnosed swiftly and efficiently, reducing the emotional and financial toll on patients and families alike.

As researchers continue to build upon this foundation, collaborative efforts will be crucial, combining expertise from various fields to overcome current limitations and enhance the technology’s effectiveness across diverse contexts. The future may hold an expansive toolkit for cancer diagnostics, fundamentally altering our understanding of disease detection and fostering a new wave of therapeutics tailored to the individual nuances of each patient’s molecular profile.

In sum, the recent study unlocks not only a method for potential early cancer detection but also catalyzes broader discussions about the future of medical diagnostics, encouraging an innovative spirit within the scientific community aimed at improving patient outcomes and empowering individuals with timely information regarding their health.

Subject of Research: Cancer detection using pulsed infrared light
Article Title: Electric-Field Molecular Fingerprinting to Probe Cancer
News Publication Date: 9-Apr-2025
Web References:
References: 10.1021/acscentsci.4c02164
Image Credits: American Chemical Society

Keywords

Cancer research, Medical diagnostics, Blood plasma analysis, Machine learning, Infrared fingerprinting, Oncology