Self-supervised 3D Patient Modeling with Multi-modal Attentive Fusion

arXiv:2403.03217v1 Announce Type: new
Abstract: 3D patient body modeling is critical to the success of automated patient positioning for smart medical scanning and operating rooms. Existing CNN-based end-to-end patient modeling solutions typically require a) customized network designs demanding large amount of relevant training data, covering extensive realistic clinical scenarios (e.g., patient covered by sheets), which leads to suboptimal generalizability in practical deployment, b) expensive 3D human model annotations, i.e., requiring huge amount of manual effort, resulting in systems that scale …

abstract arxiv automated clinical cnn cs.cv data designs fusion medical modal modeling multi-modal network patient smart solutions success training training data type