The recently emerging conditional diffusion models seem promising for mitigating the labor and expenses in building large 3D medical imaging datasets. However, previous studies on 3D CT generation primarily focus on specific organs characterized by a local structure and fixed contrast and have yet to fully capitalize on the benefits of both semantic and textual conditions. In this paper, we present GuideGen, a controllable framework based on easily-acquired text prompts to generate anatomical masks and corresponding CT volumes for the entire torso—from chest to pelvis. Our approach includes three core components: a text-conditional semantic synthesizer for creating realistic full-torso anatomies; an anatomy-aware high-dynamic-range (HDR) autoencoder for high-fidelity feature extraction across varying intensity levels; and a latent feature generator that ensures alignment between CT images, anatomical semantics and input prompts. Combined, these components enable data synthesis for segmentation tasks from only textual instructions. To train and evaluate GuideGen, we compile a multi-modality cancer imaging dataset with paired CT and clinical descriptions from 12 public TCIA datasets and one private real-world dataset. Comprehensive evaluations across generation quality, cross-modality alignment, and data usability on multi-organ and tumor segmentation tasks demonstrate GuideGen's superiority over existing CT generation methods.

Accurate segmentation of lymph node metastases in colorectal cancer is crucial for disease staging, prognosis evaluation, and treatment planning. However, the identification of metastatic lymph nodes in colorectal cancer from CT images remains extremely challenging due to their small size, indistinguishable appearance, and the potential for distal spread from the primary tumor, making them difficult to differentiate within the complex abdominal cavity. Based on the metastatic patterns of tumors, where lymph node metastases in colorectal cancer often accompany the mesenteric vessels, we propose to leverage the mesenteric vasculature as guidance for the identification of metastatic lymph nodes. Specifically, we utilize vessel segmentation maps of the superior and inferior mesenteric vessels as mesenteric vascular guidance. To explicitly model the spatial relationships between metastatic lymph nodes and mesenteric vessels, we further employ explicit distance relationship modeling to represent the distance from each voxel in the CT image to the nearest mesenteric vessel. In addition, to incorporate mesenteric vascular guidance and explicit distance relationship modeling into the segmentation process of metastatic lymph nodes in colorectal cancer, we integrate a guidance encoder and a guidance signal fusion module into the U-Net segmentation network. Inference-time vascular localization is also employed to assist in localizing the region of interest and filtering segmentation results. Experimental results on the colorectal cancer lymph node metastasis dataset demonstrate the effectiveness of our proposed method. The code will be released at https://github.com/YuriYu12/VesselGuidedSegmentation.