[2603.23041] HUydra: Full-Range Lung CT Synthesis via Multiple HU Interval Generative Modelling

Computer Science > Computer Vision and Pattern Recognition arXiv:2603.23041 (cs) [Submitted on 24 Mar 2026] Title:HUydra: Full-Range Lung CT Synthesis via Multiple HU Interval Generative Modelling Authors:António Cardoso, Pedro Sousa, Tania Pereira, Hélder P. Oliveira View a PDF of the paper titled HUydra: Full-Range Lung CT Synthesis via Multiple HU Interval Generative Modelling, by Ant\'onio Cardoso and 3 other authors View PDF Abstract:Currently, a central challenge and bottleneck in the deployment and validation of computer-aided diagnosis (CAD) models within the field of medical imaging is data scarcity. For lung cancer, one of the most prevalent types worldwide, limited datasets can delay diagnosis and have an impact on patient outcome. Generative AI offers a promising solution for this issue, but dealing with the complex distribution of full Hounsfield Unit (HU) range lung CT scans is challenging and remains as a highly computationally demanding task. This paper introduces a novel decomposition strategy that synthesizes CT images one HU interval at a time, rather than modelling the entire HU domain at once. This framework focuses on training generative architectures on individual tissue-focused HU windows, then merges their output into a full-range scan via a learned reconstruction network that effectively reverses the HU-windowing process. We further propose multi-head and multi-decoder models to better capture textures while preserving anatomical consistency, with...