Accurate vertebrae detection in 3D magnetic resonance (MR) images is crucial for disease diagnoses and treatment decisions. Despite remarkable progress in organ detection, automatic vertebrae detection remains challenging due to the complexity of pathological abnormalities and the repetitive nature of the spine. Therefore, recent vertebrae detection approaches utilizing convolutional neural networks detect the vertebrae sequentially or in multiple stages, resulting in complex frameworks and an accumulation of prediction errors. Moreover, the characteristics of routine lumbar spine MR images can make the detection model time-consuming and over-learning of inter-slice information. In this study, we propose an end-to-end hybrid-dimension height-driven attention-based detection network (HHANet) comprising a hybrid-dimension hierarchical feature extractor (HHF-Extractor) and a height-attention detection head (HA-Head). The HHF-Extractor consists of an anisotropic backbone for generating feature representations containing sufficient within-slice information and moderate inter-slice context and a projection-based asymptotic feature pyramid network for feature dimension reduction and spinal information enhancement. The HA-Head takes the enhanced feature maps as input and simultaneously performs vertebrae centroid localization and bounding box offset regression. The discrimination of the similar-appearing vertebrae is facilitated by using the height-driven attention network to model the spine anatomy and calibrate the predicted labels of bounding boxes based on their height positions. Extensive experimental results demonstrate that the proposed HHANet outperforms most state-of-the-art methods under different evaluation metrics from the aspects of localization accuracy and detection precision, and the inference time of an image of size 320 can achieve 0.146 seconds. Code will be available at https://github.com/Chenyangfan4/HHANet .

Automated assessment of coronary artery (CA) lesions via Coronary Computed Tomography Angiography (CCTA) is essential for the diagnosis of coronary artery disease (CAD). However, current deep learning approaches confront several challenges, primarily regarding the modeling of long-range anatomical dependencies, the effective decoupling of plaque texture from stenosis geometry, and the utilization of clinically prevalent mixed-grained annotations. To address these challenges, we propose a novel mixed-grained hierarchical geometric-semantic learning network (MG-HGLNet). Specifically, we introduce a topology-aware dual-stream encoding (TDE) module, which incorporates a bidirectional vessel Mamba (BiV-Mamba) encoder to capture global hemodynamic contexts and rectify spatial distortions inherent in curved planar reformation (CPR). Furthermore, a synergistic spectral–morphological decoupling (SSD) module is designed to disentangle task-specific features; it utilizes frequency-domain analysis to extract plaque spectral fingerprints while employing a texture-guided deformable attention mechanism to refine luminal boundary. To mitigate the scarcity of fine-grained labels, we implement a mixed-grained supervision optimization (MSO) strategy, utilizing anatomy-aware dynamic prototypes and logical consistency constraints to effectively leverage coarse branch-level labels. Extensive experiments on an in-house dataset demonstrate that MG-HGLNet achieves a stenosis grading accuracy of 92.4% and a plaque classification accuracy of 91.5%. The results suggest that our framework not only outperforms state-of-the-art methods but also maintains robust performance under weakly supervised settings, offering a promising solution for label-efficient CAD diagnosis.