Study Reveals Hidden AI Vulnerabilities in Lung Cancer Screening: Acquisition Parameters Affect Measurements Invisibly
A new preprint study shows that CT reconstruction kernel and image noise levels can significantly distort AI-based lung nodule measurements and detection, even when DICOM metadata appears identical. Researchers using a LUNA16-trained MONAI RetinaNet detector found that kernel differences alone flipped a clinically significant Fleischner size category in 5.2% of nodules, while noise degraded detection confidence for small nodules without affecting measurement. The findings suggest current AI governance frameworks for medical imaging have a critical unmonitored gap at the input level.
Researchers have published a preprint on arXiv demonstrating that CT scan acquisition parameters — specifically reconstruction kernel and image noise — act as structured, measurable variables that can silently undermine AI lung-nodule detection and measurement systems. Using a LUNA16-trained MONAI RetinaNet model, the team found that switching reconstruction kernels (B30f vs. B80f) on real paired CT scans flipped Fleischner nodule size categories in 5.2% of cases (8 of 155 nodules), even though detection confidence scores were statistically unchanged. Controlled perturbation experiments on the LIDC-IDRI dataset revealed a dissociation: noise degraded detection confidence — particularly for nodules under 6 mm — while frequency/kernel changes corrupted diameter measurements, with each axis leaving the other largely unaffected. Critically, a 4-feature pixel-level fingerprint could recover reconstruction identity with AUC around 0.95 on real CT and 0.995 on a phantom, whereas the standard DICOM ConvolutionKernel metadata tag was uninformative, carrying identical labels across differing reconstructions. The fingerprint also generalized across four CT manufacturers with leave-one-vendor-out AUC of 0.94–0.98. The authors argue that current accreditation frameworks, including the 2026 ACR-SIIM Practice Parameter and the ACR Assess-AI registry, monitor AI outputs but lack this input-side acquisition-aware validation layer.
What's missing
As a preprint, this work has not yet undergone formal peer review. The study relies on one specific detector architecture (MONAI RetinaNet trained on LUNA16), and it is unclear how well findings generalize to other AI models or clinical tasks beyond lung nodule detection. The clinical impact of the 5.2% Fleischner category flip rate in real-world workflows — including downstream patient management decisions — is not quantified. The proposed pixel fingerprint approach has not been validated in a prospective clinical deployment setting.
What different sources said
- arXiv cs.AICenter
Acquisition state behaves as a structured, measurable variable governing lung-nodule AI: kernel-driven measurement instability and noise-driven detection fragility, invisible to DICOM metadata
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.