Westlake University Handheld Device Detects Lung Cancer at 94.9 Percent Accuracy

LONDON, 25 June 2026 — A compact, handheld diagnostic instrument developed at Westlake University's School of Engineering can detect early-stage lung cancer biomarkers from a single drop of blood with 94.9 per cent accuracy — a figure that compares favourably with the 74.7 per cent accuracy achieved by the standard enzyme-linked immunosorbent assay test under equivalent conditions. The device is approximately 10,000 times more sensitive than standard ELISA methods, which have been the clinical workhorse for protein biomarker quantification since their introduction in 1971. The research, led by Associate Professor Wen Liaoyong, was conducted in collaboration with Xiamen University and validated across 171 clinical serum samples from lung cancer patients, with findings reported by the South China Morning Post.

• 94.9 per cent accuracy for early-stage lung cancer detection against a 74.7 per cent ELISA baseline
• 92.1 per cent accuracy for post-operative monitoring
• Approximately 10,000-fold improvement in sensitivity over standard ELISA
• Manufacturing cost of US$5 per chip, down from hundreds of dollars using conventional methods

The Global Cancer Detection Problem

According to the World Health Organisation, cancer was responsible for approximately 9.7 million deaths worldwide in 2022. GLOBOCAN 2022 data, published in CA: A Cancer Journal for Clinicians, recorded 20 million new cancer cases globally that year, with lung cancer accounting for 2.5 million diagnoses — making it the most commonly diagnosed cancer worldwide and the leading cause of cancer death, responsible for 1.8 million fatalities. Early-stage detection remains the single most powerful lever for improving survival: five-year survival rates for localised lung cancer exceed 60 per cent, compared with roughly 7 per cent for distant-stage disease, according to the National Cancer Institute.

The diagnostic gap is particularly acute in resource-limited settings. Drain et al. (Lancet Infectious Diseases, 2014) documented how point-of-care testing in resource-limited environments has consistently been hindered by the cost, complexity, and infrastructure requirements of laboratory-grade equipment. The problem is compounded for liquid biopsy applications — the analysis of circulating biomarkers in blood — where extreme sensitivity is required to detect the vanishingly small quantities of tumour-derived molecules present in early-stage patients.

Siravegna et al. (Nature Reviews Clinical Oncology, 2017) established that liquid biopsy techniques hold substantial promise for non-invasive cancer diagnostics, but translation to the clinic has been restricted by the need for laboratory infrastructure — spectroscopy equipment that is routinely the size of a double-door refrigerator and costs tens of thousands of dollars. Wen's device collapses that barrier.

Related: Pathway Labs Gains FDA Clearance for AI Heart Tool

Inside the Technology: Q-Modulated Refractometric Sensing

The device is built around a mechanism called Q-modulated refractometric sensing, which distinguishes it from conventional spectroscopy-based optical detection. Traditional spectroscopy measures the wavelength of light to identify molecular signatures — a method that requires prisms, spectrometers, and complex optical paths. Wen's approach instead measures light intensity, exploiting the fact that when biomarkers bind to the chip's surface, they alter the local refractive index of the sample in a measurable, highly reproducible manner.

The sensing surface is a three-dimensional chip fabricated from metamaterials — engineered nanostructures designed to manipulate light in ways that natural materials cannot. The chip operates using a photonic concept known as bound states in the continuum (BIC), resonance modes that confine light within the chip's nanostructure with extraordinary precision. When biomarkers are present in a blood sample, they perturb this confined resonance in a manner proportional to their concentration, generating a light-intensity signal that a simple photodetector can read. This is conceptually analogous to using a precise pressure gauge rather than a chromatograph to determine a gas composition — a fundamental simplification that allows the optical system to be reduced to three components: a 3D BIC sensing chip, an LED light source, and a photodetector.

The BIC architecture also provides the sensitivity advantage. Knight et al. (ACS Nano, 2014) demonstrated that aluminium-based nanostructures can sustain plasmonic resonances at shorter wavelengths than gold or silver alternatives, enabling high-precision optical confinement across a wider manufacturing scale range — from nanometre features to macroscopic chip dimensions. Wen's team used aluminium for the master template underlying the new chip design, building on findings published in Nature Materials that demonstrated high-precision metasurface manufacturing at scale.

Related: OpenAI LifeSciBench Tests AI Against 750 Real Drug Discovery Tasks

From Double-Door Refrigerator to Pocket Device

The manufacturing breakthrough that enables miniaturisation is a shift from electron beam lithography — which writes chip features one stroke at a time, like copying a book word by word — to a nano-imprint process that produces a single master template and then stamps thousands of consistent copies, akin to movable-type printing. Where conventional metasurface chips produced through electron beam lithography cost hundreds of US dollars each, Wen's team can now print thousands of chips on a single eight-inch silicon wafer at a cost of US$5 per chip. This is not incremental cost reduction; it is a structural shift in the manufacturing economics of optical biosensors.

The resulting system is compact enough for home use: a 3D BIC chip, an LED, and a photodetector. No spectrometers. No cooling systems. No calibration laboratory. The device takes one drop of blood and returns a quantitative biomarker reading. For health systems managing large asymptomatic populations — in primary care clinics, community hospitals, or mobile screening programmes in lower-middle-income countries — this architecture represents a fundamentally different screening model.

Related: Siemens Healthineers and GE HealthCare Advance AI Imaging Adoption

Related: MidJourney Disrupts The Imaging Industry with the MidJourney Scanner

Clinical Validation: 171 Serum Samples, 94.9 Per Cent Accuracy

Wen's team used the device to detect small extracellular vesicles (sEVs) — membrane-bound nanoparticles shed by tumour cells into the bloodstream — associated with lung cancer. sEVs are a clinically recognised class of liquid biopsy biomarker: they carry surface proteins and nucleic acids characteristic of their cell of origin, enabling tumour detection without tissue biopsy. However, sEV concentrations in the serum of early-stage lung cancer patients are extremely low, making detection with conventional ELISA unreliable at the diagnostic threshold. Yoshioka et al. (Nature Communications, 2014) established this sensitivity challenge as the principal barrier to clinical deployment of sEV-based liquid biopsy at scale.

In validation testing across 171 clinical serum samples, the Westlake device achieved:

• 94.9 per cent accuracy for distinguishing early-stage lung cancer from healthy controls
• 92.1 per cent accuracy for post-operative monitoring — tracking whether tumour burden had been adequately reduced after surgery
• 74.7 per cent accuracy achieved by conventional ELISA on the same sample set

The 20.2 percentage-point superiority over ELISA at the early-detection task is clinically material. In population screening at scale, a 20-point lift in accuracy translates to tens of thousands of patients correctly identified for further investigation per million screened. The National Lung Screening Trial (NEJM, 2011) demonstrated that screening high-risk populations reduces lung cancer mortality by 20 per cent — a statistically validated effect achieved with low-dose CT scanning that requires dedicated radiology infrastructure. A blood-based screening alternative with comparable accuracy and a US$5 consumable cost changes the deployment economics of that screening model entirely.

Related: Varda Space Industries Crystallizes Ritonavir in Space

Competitive Landscape: Point-of-Care Cancer Diagnostics

The handheld cancer biomarker detection space is nascent but competitive. Several platforms are pursuing lateral flow, electrochemical, or Raman-spectroscopy approaches to achieve laboratory-grade sensitivity at point-of-care. The Westlake device's principal differentiation is its chip cost and its 10,000-fold sensitivity advantage over ELISA — a gap that competing platforms have not yet published equivalent data to challenge.

| Platform Type | Technology | Sensitivity vs ELISA | Portable | Cost per Test | |---|---|---|---|---| | Westlake Q-modulated BIC | Metamaterial refractometric | ~10,000× | Yes | ~US$5 chip | | Lateral flow immunoassay | Antibody strip | 1–10× | Yes | US$1–10 | | Electrochemical biosensor | Impedance / amperometric | 100–1,000× | Partially | US$10–50 | | Portable SERS | Raman spectroscopy | 100–10,000× | Partially | US$20–200 | | Standard ELISA (baseline) | Enzyme-linked antibody | 1× (baseline) | No | US$5–20 |

Related: OpenAI Enhances ChatGPT Health Intelligence for Personalised Care

What It Means for Healthcare and Investors

For health system operators and hospital procurement teams, the device addresses a structural gap in oncology screening infrastructure. Current lung cancer screening guidelines — including those maintained by the American Lung Association and endorsed by the American Cancer Society — focus on low-dose CT scanning for high-risk individuals defined by smoking history and age. Blood-based screening at this accuracy level could broaden the eligible screening population and reduce dependence on radiology capacity, which remains acutely constrained in emerging markets. The IARC Global Cancer Observatory projects that low- and middle-income countries will account for approximately 70 per cent of cancer deaths by 2030, precisely where CT infrastructure is least available.

For the regulatory pathway, the device would likely require 510(k) clearance or De Novo classification in the United States, depending on the specific intended use and predicate devices. The FDA's existing framework for liquid biopsy in vitro diagnostic devices — which the agency has been developing since 2018 — creates a defined regulatory route, though clinical evidence requirements for a novel detection modality will require pivotal trial data beyond the 171-sample validation published to date.

For venture capital and strategic diagnostics acquirers, the manufacturing economics are the primary investment thesis. A US$5 chip that outperforms a US$20 ELISA at a key diagnostic task, on a platform that does not require laboratory infrastructure, is a scalable commercial product. Westlake University has not yet disclosed commercialisation arrangements. The WHO's framework for cancer early diagnosis identifies affordability and accessibility as the two systemic constraints on global screening uptake — both of which this device is designed to address.

Related: Novellia: $18M Series A Targets Pharma's $50B Real-World Data Problem

Related: Pair Team Secures Spot in Medicare AI Initiative

Forward Outlook

The research paper indicates the platform is not limited to lung cancer biomarker detection — the Q-modulated sensing architecture can be adapted to detect different analytes by functionalising the chip surface with alternative antibodies or aptamers. Wen has cited broader applications in tumour screening across multiple cancer types as a near-term research objective. Clinical translation will require expanded trial cohorts, regulatory engagement in target markets, and manufacturing partnerships to realise the wafer-scale production economics described in the paper. Westlake University's track record of technology transfer — anchored in a model that co-locates academic research with commercialisation infrastructure — positions the team to move from laboratory to spinout without the delays typical of conventional academic technology transfer offices.

Related: Oura 2026: Ring 5 Launches Blood Pressure Tracking Under FDA Carve-Out

Related: Augmedix, Tempus AI and Aidoc Lead Top 10 AI Healthcare Applications to Watch in 2026

---

BUSINESS 2.0 has no commercial relationship with Westlake University, Xiamen University, or any company mentioned in this article.

Frequently Asked Questions

How does the Westlake handheld cancer detection device work?

The device uses Q-modulated refractometric sensing: a 3D chip made from metamaterials confines light via bound-in-continuum resonance modes. When cancer biomarkers bind to the chip surface, they shift the local refractive index, altering light intensity in a measurable way. Because the system only needs to measure intensity — not wavelength — it requires only a 3D BIC chip, an LED, and a photodetector, enabling handheld form factor and US$5 chip cost.

What biomarkers does the device detect, and how was it validated?

The device detects small extracellular vesicles (sEVs) associated with lung cancer — membrane-bound nanoparticles shed by tumour cells into the bloodstream. Clinical validation was conducted on 171 serum samples from lung cancer patients in collaboration with Xiamen University, achieving 94.9 per cent accuracy for early-stage detection and 92.1 per cent accuracy for post-operative monitoring, versus 74.7 per cent for standard ELISA on the same sample set.

When might the device become available for clinical use?

The device is at the research validation stage. Clinical translation would require expanded pivotal trials, regulatory clearance (likely 510(k) or De Novo pathway in the US), manufacturing scale-up, and commercial partnerships. Westlake University has not yet announced a commercialisation timeline or spinout arrangement. The manufacturing breakthrough — US$5 per chip at wafer scale — substantially reduces the capital requirement for scaling production once regulatory approvals are secured.