WING: A Window-Prior-Based Generative Network with Gated Inception for Cross-Modality CT Synthesis

2026-07-07 · Source: Takara TLDR - Daily AI Papers · Field: Health & Wellbeing — Clinical Care & Medical Practice, Medical Devices & Health Technology, Health & Medical Research · Depth: Expert, quick

Summary

WING, a WINdow-prior-based Generative network, offers a state-of-the-art solution for synthesizing CT volumes from MRI and CBCT scans. This technology significantly improves treatment planning in adaptive radiotherapy by avoiding additional radiation exposure. It addresses the challenge of direct CT intensity regression, which struggles with high dynamic range and long-tailed distributions that obscure clinically important structures. WING reformulates the regression target into multiple windowed representations, leveraging the inductive prior that CT intensities are structure-deterministic and window-separable. The network integrates a Gated Inception Generator for multi-window predictions, a Fuse-and-Refine Transformer for output aggregation and detail refinement, and a joint adversarial training objective for enhanced realism. WING demonstrates superior performance on MRI-to-CT and CBCT-to-CT benchmarks, supporting multi-anatomy synthesis with a single compact model.

Key takeaway

For AI Scientists and Research Scientists developing medical image synthesis solutions, WING presents a robust, high-performing architecture. You should consider integrating its window-prior-based reformulation and multi-component generative approach to overcome challenges with high dynamic range in CT synthesis. This method offers a path to more accurate and clinically relevant CT generation, particularly for adaptive radiotherapy, while supporting diverse anatomical structures with a single model.

Key insights

WING synthesizes CT from MRI/CBCT using windowed representations and a multi-component generative network for improved radiotherapy planning.

Principles

• CT intensities are structure-deterministic and window-separable.
• Windowed views exhibit smoother distributions.
• Multi-shape kernel interactions capture cross-modality correspondence.

Method

WING employs a Gated Inception Generator for multi-window predictions, a Fuse-and-Refine Transformer for output aggregation and refinement, and joint adversarial training for realism.

In practice

• Generate CT from MRI for radiotherapy planning.
• Synthesize CT from CBCT for treatment adaptation.
• Support multi-anatomy synthesis with one model.

Topics

• CT Synthesis
• Adaptive Radiotherapy
• MRI-to-CT
• CBCT-to-CT
• Generative Networks
• Medical Imaging

Best for: AI Scientist, Research Scientist

Related on AIssential

• Recover Semantics First, Generate Better: Improved Latent Modeling for 3D MRI Reconstruction and Cross-Contrast Synthesis — Prioritizing semantic recovery in 3D MRI latent space compression significantly improves reconstruction fidelity and cross-contrast synthesis.
• Scaling Generative Foundation Models for Chest Radiography with Rectified Flow Transformers — A 1.3B-parameter generative foundation model for chest radiographs achieves expert-indistinguishable synthesis and controllable editing.
• Dual-Domain Equivariant Generative Adversarial Network for Multimodal CT-PET Synthesis — Dual-domain learning and rotational equivariance significantly enhance multimodal medical image synthesis accuracy and robustness.
• From Baseline to Follow-Up: Counterfactual Spine DXA Image Synthesis in UK Biobank Using a Causal Hierarchical Variational Autoencoder — The CHVAE model synthesizes anatomically plausible, causally consistent spine DXA images by intervening on age, addressing anatomical variation challenges.
• Transformer-Guided Graph Attention for Direct Cardiac Mesh Reconstruction: A Structural Digital Twin Framework — The framework directly generates simulation-ready cardiac meshes from medical images, bypassing traditional multi-stage, manual post-processing.
• Anatomy-Guided Residual Motion Diffusion for Controllable 4D Cardiac MRI Synthesis — A cascaded latent diffusion model synthesizes controllable, anatomically consistent 4D cardiac MRI, improving AI model robustness with limited data.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Takara TLDR - Daily AI Papers.