analiza aiqbench
This research introduces MIRAGE, a specialized deep learning framework designed to synthesize post-contrast MRI images from pre-contrast scans while explicitly preserving critical tumor enhancement details often lost in standard generative models. By integrating multi-scale lesion supervision and asymmetric penalties for missed tumors, the method achieves superior performance across radiomics and segmentation metrics compared to baseline diffusion or pix2pix approaches. The study highlights a nuanced fidelity-utility trade-off where task-aware synthesis is essential for medical imaging applications requiring high diagnostic accuracy rather than just perceptual realism.
MIRAGE: Multi-scale Lesion-Informed Representation with Auxiliary Guidance for MRI Contrast Enhancement
arXiv:2607.19137v1 Announce Type: cross Abstract: Inferring contrast enhancement from one pre-contrast breast MRI slice is underdetermined: post-contrast appearance contains physiological information that is not uniquely encoded in baseline anatomy. Optimizing only paired pixel fidelity can suppress uncertain lesion enhancement, whereas adversarial or stochastic generative objectives can favor realistic post-contrast appearance without guaranteeing patient-specific lesion fidelity. We introduce MIRAGE, a residual 2D U-Net that combines global reconstruction and perceptual losses with three forms of lesion-aware supervision available only during training: an asymmetric penalty for missed tumor enhancement, multi-scale auxiliary tumor segmentation, and guidance through a frozen post-contrast tumor segmentation nnU-Net. We evaluate the method on 301 cases from the multi-centre MAMA-SYNTH data using eight complementary image-, region-, radiomics-, and segmentation-based metrics. MIRAGE ranks first on six metrics and markedly improves downstream lesion localization over tuned pix2pix, conditional diffusion, and latent bridge-matching baselines. The generative alternatives retain advantages in LPIPS or contrast classification, revealing a clear fidelity-utility trade-off. Leave-one-in and leave-one-out ablations show that the losses are partly redundant for lesion localization but exert distinct effects on appearance, radiomics, and boundary accuracy. These results support task-aware synthesis while also showing that its apparent optimality is conditional on the downstream models and metrics used to define utility.
Abstract:Inferring contrast enhancement from one pre-contrast breast MRI slice is underdetermined: post-contrast appearance contains physiological information that is not uniquely encoded in baseline anatomy. Optimizing only paired pixel fidelity can suppress uncertain lesion enhancement, whereas adversarial or stochastic generative objectives can favor realistic post-contrast appearance without guaranteeing patient-specific lesion fidelity. We introduce MIRAGE, a residual 2D U-Net that combines global reconstruction and perceptual losses with three forms of lesion-aware supervision available only during training: an asymmetric penalty for missed tumor enhancement, multi-scale auxiliary tumor segmentation, and guidance through a frozen post-contrast tumor segmentation nnU-Net. We evaluate the method on 301 cases from the multi-centre MAMA-SYNTH data using eight complementary image-, region-, radiomics-, and segmentation-based metrics. MIRAGE ranks first on six metrics and markedly improves downstream lesion localization over tuned pix2pix, conditional diffusion, and latent bridge-matching baselines. The generative alternatives retain advantages in LPIPS or contrast classification, revealing a clear fidelity-utility trade-off. Leave-one-in and leave-one-out ablations show that the losses are partly redundant for lesion localization but exert distinct effects on appearance, radiomics, and boundary accuracy. These results support task-aware synthesis while also showing that its apparent optimality is conditional on the downstream models and metrics used to define utility.
Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2607.19137 [eess.IV]
(or arXiv:2607.19137v1 [eess.IV] for this version)
https://doi.org/10.48550/arXiv.2607.19137 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 · 301 cuvinte
Subiecte conexe
The Entangled Dimensions of Decision Theory
LessWrong · 30 iul. 2026
Quantum phase estimation with optimal confidence interval using three control qubits
Quantum Journal · 19 aug. 2026
Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks
arXiv · 28 sept. 2026
Introducing Instella-MoE: A State-of-the-Art Fully Open Mixture-of-Experts Language Model
AMD · 24 iul. 2026