analiza aiqbench
The paper proposes generating synthetic training data from archived routine samples so that diagnostic models can run directly on novel-modality inputs, avoiding the hallucination risk, compute cost, and regulatory overhead of generating images at inference time. The evidence is a single proof-of-concept on unstained prostate histopathology, with agreement against H&E benchmarks that is better than inference-time virtual staining but still moderate for raw tissue. Readers should treat it as a promising framing for data-scarce medical imaging, not a validated clinical result.
Training-Time Domain Transfer for Generator-Free Diagnostics on Novel Modalities in Computational Pathology
arXiv:2611.02931v1 Announce Type: cross Abstract: Novel imaging modalities promise improved diagnostics, but AI development is constrained by scarce training data. Inference-time domain transfer can address this by mapping novel modalities to established ones, enabling reuse of existing diagnostic models. However, this introduces hallucination risks, computational overhead, and regulatory complexity. Inspired by sim-to-real transfer, we propose training-time domain transfer, generating synthetic target-domain training data from archived routine samples to enable direct diagnosis on real inputs without inference-time generation. Using unstained prostate histopathology as a proof-of-concept, we compare this approach with inference-time virtual staining and conventional H&E-based cancer detection. Training-time transfer achieved higher concordance with the conventional benchmark than virtual staining for both raw (Cohen's $\kappa$=0.59 vs. 0.37; Gwet's AC1=0.73 vs. 0.29) and deparaffinized tissue ($\kappa$=0.66 vs. 0.61; AC1=0.73 vs. 0.64), supporting its potential for emerging diagnostic modalities.
Abstract:Novel imaging modalities promise improved diagnostics, but AI development is constrained by scarce training data. Inference-time domain transfer can address this by mapping novel modalities to established ones, enabling reuse of existing diagnostic models. However, this introduces hallucination risks, computational overhead, and regulatory complexity. Inspired by sim-to-real transfer, we propose training-time domain transfer, generating synthetic target-domain training data from archived routine samples to enable direct diagnosis on real inputs without inference-time generation. Using unstained prostate histopathology as a proof-of-concept, we compare this approach with inference-time virtual staining and conventional H&E-based cancer detection. Training-time transfer achieved higher concordance with the conventional benchmark than virtual staining for both raw (Cohen's $\kappa$=0.59 vs. 0.37; Gwet's AC1=0.73 vs. 0.29) and deparaffinized tissue ($\kappa$=0.66 vs. 0.61; AC1=0.73 vs. 0.64), supporting its potential for emerging diagnostic modalities.
4 pages, 2 figures. Submitted to IEEE ISBI 2027
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2611.02931 [cs.CV]
(or arXiv:2611.02931v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2611.02931 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 240 cuvinte