Contemporary screening rests on mammography, supplemented by MRI or ultrasound for high-risk groups. These modalities excel at detecting anatomically manifest lesions but cannot capture the pre-invasive molecular changes that precede visible tumor formation by months or even years. Imaging is, in essence, phenotypic: it records the downstream structural consequences of malignant transformation—tissue shape, texture, vascularity, and architectural distortion—without ever seeing the genomic events that set those changes in motion. Liquid biopsy, by contrast, interrogates tumor-derived material circulating in peripheral blood, including circulating tumor DNA (ctDNA), cell-free DNA methylation signatures, circulating tumor cells, microRNAs, and extracellular vesicles. It offers a biologically earlier detection window, but it is spatially blind and suffers from low sensitivity at early stages. The review argues that neither approach alone can reach the diagnostic ceiling that joint analysis might achieve.

The limitations of standalone imaging are well documented. Mammography loses sensitivity in dense breast tissue, which is present in 25 to 40 percent of screening-age women, because tumors and fibroglandular tissue appear similarly radio-opaque. Dense breasts confer a 1.6- to 2-fold elevated cancer risk while simultaneously reducing detection reliability. Approximately half of women undergoing annual mammography over a decade receive at least one false-positive result. Supplemental MRI halved interval cancer rates in the DENSE trial, but carries lower specificity, high cost, and access barriers that preclude universal deployment. A 2023 meta-analysis found MRI detected 25.7 cancers per 1,000 screenings versus 4.3 to 4.8 per 1,000 for ultrasound and tomosynthesis, yet the false-positive burden persisted across all modalities. Crucially, all imaging remains phenotypic and cannot visualize pre-invasive somatic mutations, epigenetic reprogramming, or clonal expansion.

Liquid biopsy faces complementary constraints. ctDNA constitutes less than 0.01 percent of total cell-free DNA in early-stage tumors, creating a severe signal-to-noise problem. Fragments are cleared in as little as 30 minutes, so a single blood draw captures only a transient molecular snapshot. No gold-standard assay exists for early-stage detection, and most clinical assays interrogate only one or two tumor DNA features, limiting specificity. In low-prevalence screening populations, even small specificity reductions generate many false positives, leading to unnecessary biopsies, patient anxiety, and increased costs. Clonal hematopoiesis of indeterminate potential—age-related mutations in blood cells that mimic early cancer signals—further complicates interpretation. And because liquid biopsy cannot localize a detected signal or characterize a lesion’s radiological phenotype, it is a complement rather than a replacement for imaging.

The central hypothesis of the review is that joint AI analysis of imaging phenotypes and blood-borne genotypes may eventually improve sensitivity and specificity for earlier-stage detection compared with either modality alone, though this remains to be validated in large prospective clinical studies. Specific alterations—including PIK3CA mutations, BRCA1/2 methylation changes, TP53 mutations, and HER2 amplification—can be detectable in plasma at the pre-invasive stage when imaging remains uninformative. The relationship is bidirectional: certain MRI texture and kinetic features correlate with receptor status and tumor mutational burden, while ctDNA methylation signatures carry tissue-of-origin signals that can direct organ-specific imaging. Combining both data streams can reduce false negatives from each modality’s blind spots and false positives by requiring corroborating evidence across independent data streams.

In quantitative terms, standalone liquid biopsy assays for early-stage disease typically show lower sensitivity—approximately 30 to 70 percent—despite high specificity above 90 percent, while unimodal imaging systems generally achieve AUC values of approximately 0.75 to 0.92 depending on the modality and population. In contrast, multimodal fusion methods integrating imaging and liquid biopsy data have reported relative AUC improvements of 10 to 20 percent and sensitivity gains of 15 to 25 percent in retrospective datasets, although these benefits vary across studies and depend strongly on cohort size and data quality. The review organizes fusion strategies into three tiers. Feature-level fusion extracts modality-specific representations independently and presents a joint vector to a shared downstream classifier, enabling the model to learn cross-modal interactions simultaneously but risking overfitting when one modality dominates the joint feature space. Decision-level fusion trains separate unimodal models and combines their output risk scores via majority voting, Bayesian averaging, or learned ensemble weights, yielding AUC increases of 3 to 8 percent over unimodal baselines while remaining robust to missing modalities. The largest improvements come from intermediate, attention-based architectures, with AUC increases of 10 to 20 percent in retrospective benchmarks.

Intermediate fusion architectures project modality-specific inputs onto a common lower-dimensional manifold where imaging and molecular representations are geometrically aligned. This shared latent space enables cross-modal biomarker discovery, imputation of missing modalities, and uncertainty quantification through probabilistic sampling. Emerging consensus favors hybrid intermediate designs where modality-specific encoders feed into a shared latent space fused through cross-attention mechanisms, offering the best balance between interaction learning and robustness to missing data. Explainable AI tools are essential for clinical adoption: Grad-CAM highlights spatial imaging regions driving predictions, SHAP quantifies the contribution of each molecular feature, and LIME provides locally faithful approximations around individual predictions. Transfer learning and federated learning address dataset scarcity, enabling model training across institutions without centralizing patient data—a critical consideration given privacy governance and medico-legal accountability concerns.

Both data streams become more informative when modeled longitudinally. Longitudinal imaging tracks changes in lesion size, morphology, and radiomic feature trajectories, identifying subtle progressive trends that are individually sub-threshold but collectively detectable. Serial liquid biopsy sampling through monthly or quarterly blood draws adds a temporal molecular dimension: a rising ctDNA mutant allele fraction may signal clonal expansion before any imaging-detectable lesion forms. ctDNA monitoring can predict breast cancer relapse with a lead time of up to three years. The core challenge is the asynchrony between imaging cycles and blood draw intervals, requiring careful co-registration onto a common time axis. Combining imaging and genomic data creates a high-dimensional joint feature space where the curse of dimensionality can cause models to memorize noise rather than learn generalizable signals, particularly when matched datasets contain only a few hundred cases. One pan-cancer survival study had to discard over 99 percent of variables before late-fusion ensembling.

Translating AI-driven multimodal platforms into clinical practice faces substantial barriers. Variability in imaging techniques—including differences in mammography kilovoltage, detector types, MRI field strength, and contrast dosing—leads to inconsistent radiomic outputs susceptible to vendor-specific artifacts and deep learning domain shift. Liquid biopsy is complicated by pre-analytical inconsistencies in tube types, centrifugation delays, and freeze-thaw cycles, as well as diverse analytical platforms including droplet-based digital PCR, next-generation sequencing panels, and methylome assays that resist direct comparison. Inter-institutional variability in demographics, scanners, and clinical workflows exacerbates harmonization challenges. A lack of matched longitudinal datasets limits the ability to train models capturing tumor evolution over time. Annotation noise arises from inter-radiologist variability, long follow-up periods required for ctDNA ground-truth labels, false positives from clonal hematopoiesis, and imaging artifacts in dense breast tissue. Because early breast cancer is rare in asymptomatic populations, even small false-positive rates can create substantial clinical burden.

Currently, no multimodal AI platform integrating breast imaging and liquid biopsy has been successfully incorporated into clinical workflows for population-level early breast cancer screening. Among the technologies discussed in the review, only AI-assisted imaging systems have received regulatory approval for clinical use. The FDA granted 510(k) clearance to nine AI mammography tools between 2017 and 2021, and Prognosia Breast achieved Breakthrough Device Designation in 2025. Liquid biopsies such as Guardant360 CDx are approved only for metastatic ESR1 detection in the US and Japan, with no endorsements for early screening. Future research must prioritize large-scale, prospective, multicenter trials validating multimodal AI systems across diverse populations—including different ethnicities, ages, and breast density subtypes. Standardized protocols for serial liquid biopsy sampling, synchronized with high-resolution imaging, will be essential. Personalized early-detection models integrating germline genetics, breast density, and hormonal history alongside imaging and molecular signals will improve sensitivity in high-risk subgroups, while longitudinal monitoring frameworks that continuously update risk scores using serial ctDNA trajectories and interval imaging represent the next generation of surveillance platforms. Scaling these innovations globally will require multimodal data fusion, federated learning, and validation across diverse populations to ensure equitable benefits.

Subject of Research: Integration of imaging with liquid biopsy using artificial intelligence for ultra-early detection of breast cancer

Article Title: Integration of imaging with liquid biopsy using artificial intelligence for ultra-early detection of breast cancer

Article References: Ramgir-Naidu, S., Govekar, A., Ojha, A., & Soni, M. (2026). Integration of imaging with liquid biopsy using artificial intelligence for ultra-early detection of breast cancer. Frontiers in Imaging, 5, Article 1870528. https://doi.org/10.3389/fimag.2026.1870528

Image Credits: AI Generated

DOI: 10.3389/fimag.2026.1870528

Keywords: Integration, imaging, liquid, biopsy, artificial, intelligence, ultra-early, detection, breast, cancer, scientific research