Deep Learning Models for Skin Cancer Detection Show Significant Performance Drop on Clinical Data
Researchers developed and evaluated a two-stage cascade deep learning system for classifying dermoscopic skin lesion images, comparing four neural network architectures across binary and multi-class schemes. While the system achieved strong internal performance (ROC-AUC 0.952–0.966), accuracy dropped substantially when tested on independent Russian clinical datasets, with sensitivity falling to 0.53–0.67 and calibration errors rising sharply. The findings highlight that AI skin cancer tools trained on international open datasets require external clinical validation and recalibration before real-world deployment.
A study posted to arXiv evaluated four deep learning architectures—ViT-B/16, Swin-S, ConvNeXt-S, and EfficientNetV2-S—for classifying dermoscopic images of skin neoplasms into benign, melanoma (MEL), squamous cell carcinoma (SCC), and basal cell carcinoma (BCC) categories. Three classification schemes were tested: binary (malignant vs. benign), single-stage four-class, and a novel two-stage cascade that first triages lesions as malignant or benign, then differentiates among malignant subtypes. Models were trained on aggregated ISIC Archive open data and validated both on a held-out internal sample and two independent Russian clinical datasets (Melanoscope AI and Sechenov University). Internally, performance was strong, but on clinical data ROC-AUC dropped to 0.797–0.893, sensitivity fell to 0.53–0.67, and expected calibration error rose from 0.02 to 0.27–0.39, with models systematically underestimating malignancy. The cascade approach improved macro F1 over single-stage classification for most architectures, with a statistically significant benefit for ViT-B/16, by recovering malignant lesions incorrectly assigned to the dominant benign class. The authors conclude that a tunable triage threshold offers clinically meaningful sensitivity control unavailable in standard argmax classification, but emphasize that the persistent generalization gap makes external validation and recalibration mandatory before any clinical deployment.
What's missing
The study does not report demographic or skin-tone diversity characteristics of the clinical validation datasets, which is relevant given known performance disparities of dermoscopy AI across skin types. It is also unclear whether the Melanoscope AI and Sechenov University datasets were prospectively collected or retrospectively curated, which could affect the representativeness of the generalization gap estimate. Dataset size for the two clinical validation sets is not specified in the abstract, limiting assessment of statistical power.
What different sources said
- arXiv cs.AICenter
Cascade Classification of Dermoscopic Images of Skin Neoplasms with Controllable Sensitivity and External Clinical Validation
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.