Abstract
The term "virtual biopsy" is gaining significant traction in medical literature. Since 2015, the number of publications referencing the concept in PubMed has doubled, reaching its highest level to date in 2021. This surge raises critical questions about the unique characteristics of a virtual biopsy and whether it should be classified as a software-based decision support tool or a standalone medical procedure.
The Limitations of the Physical Gold Standard
For decades, the physical biopsy has been the gold standard for diagnosing diseases like cancer. By extracting tissue samples for microscopic analysis, clinicians can identify cellular abnormalities with high specificity. However, this method has inherent drawbacks. It is invasive, carries risks of patient complications, and is susceptible to "sampling error." Because a needle only captures a minute fraction of a tumor, it may fail to represent the most aggressive regions of a heterogeneous mass.
This paper explores the conceptual and regulatory gap between AI as a "decision support tool"—software that assists a physician's reasoning—and a "medical procedure," which serves as a definitive diagnostic step. The central inquiry is whether AI-driven analysis of medical imaging can provide biological insights equivalent to, or better than, physical tissue samples.
Methodological Foundations
The transition toward virtual biopsies is supported by the synthesis of several advanced technical domains:
- Radiomics: The extraction of quantitative, high-dimensional features from medical images that are imperceptible to the human eye.
- Super-resolution AI: The application of deep learning to enhance image resolution beyond the physical limits of current hardware.
- Multi-modal Imaging: The integration of diverse data streams, such as 18F-FDG PET/MRI and sonography, to increase diagnostic certainty.
- Radiogenomics: The use of non-linear AI models to infer genetic traits and molecular phenotypes (such as immunotherapy targets) directly from imaging patterns.
Beyond technical feasibility, the paper evaluates the regulatory and economic frameworks—such as the UK’s NICE standards and Germany’s digital health laws—necessary to transform these algorithms into billable medical procedures.
Key Insights: Volume vs. Resolution
A primary advantage of the virtual biopsy is its ability to assess the entire tumor volume and its metastases (the "primarius and its filiae"). While a microscope offers superior resolution at the cellular level, imaging provides a more complete characterization of tumor heterogeneity.
The authors also address the terminology itself, noting that "virtual" is somewhat of a misnomer. Radiologic images reflect physical reality just as histology slides do; the term "virtual" refers to the simulation of the diagnostic insight rather than a lack of physical evidence. To move from a support tool to a recognized procedure, AI must transition toward "decision-making," where the software’s output is trusted enough to safely omit the physical biopsy.
Clinical Evidence and Results
Several "proof-of-concept" studies demonstrate the potential of this approach:
- Prostate Cancer: In a study of 291 patients, the combination of MRI and PSMA PET/CT showed such high detection rates for significant cancer that researchers suggested physical biopsies could be omitted in specific high-suspicion cases.
- Breast Cancer: Evidence suggests that if both PET/MRI and ultrasound are positive for axillary lymph node involvement, clinicians might dispense with invasive histopathologic sampling.
- Lung Cancer: AI has successfully linked radiomics features in non-small cell lung cancer (NSCLC) to the expression levels of immunotherapy targets, potentially guiding treatment without requiring new tissue samples.
The Path to Clinical Adoption
Despite these promising results, several hurdles remain. The majority of current evidence is retrospective; the authors emphasize that large-scale, prospective randomized controlled trials are the "central hurdle" for stakeholder acceptance.
Furthermore, the "billability" of virtual biopsies depends on health economics. For insurance companies and regulatory bodies to adopt these tools, they must see clear evidence of cost reduction and improved patient quality of life, such as the avoidance of biopsy-related complications. The transition from "low-risk" digital health applications to "high-risk" diagnostic procedures requires rigorous validation by bodies like the FDA or the Federal Institute for Drugs and Medical Devices in Germany.
Summary updated