标题： 从单目RGB图像中准确估计后背深度和脊柱形态的双特征提取器框架 显示英文标题

标题： A Dual-Feature Extractor Framework for Accurate Back Depth and Spine Morphology Estimation from Monocular RGB Images

摘要： 脊柱侧弯是一种常见的疾病，影响身体健康和外观，其中青少年特发性脊柱侧弯（AIS）是最常见的一种形式。 目前，主要的AIS评估工具是X光片，但存在显著的局限性，包括辐射暴露以及在贫困和偏远地区的可及性有限。 为了解决这个问题，当前的解决方案是使用RGB图像来分析脊柱形态。 然而，RGB图像极易受到环境因素的影响，例如光照条件，这会损害模型的稳定性和泛化能力。 因此，在本研究中，我们提出了一种新的流程，以准确估计裸露背部的深度信息，弥补二维信息的不足，然后通过整合深度和表面信息来估计脊柱形态。 为了精确捕捉背部表面的细微深度变化，我们设计了一个自适应多尺度特征学习网络，名为网格感知多尺度自适应网络（GAMA-Net）。 该模型使用双编码器提取块级和全局特征，这些特征通过基于块的混合注意（PBHA）模块进行交互。 自适应多尺度特征融合（AMFF）模块用于在解码器中动态融合信息。 结果表明，我们的深度估计模型在三个不同的评估指标上取得了显著的准确性，分别为近78.2%、93.6%和97.5%。 为了进一步验证预测深度的有效性，我们将表面和深度信息结合起来进行脊柱形态估计。 这种综合方法提高了脊柱曲线生成的准确性，达到了高达97%的出色性能。 显示英文摘要

摘要： Scoliosis is a prevalent condition that impacts both physical health and appearance, with adolescent idiopathic scoliosis (AIS) being the most common form. Currently, the main AIS assessment tool, X-rays, poses significant limitations, including radiation exposure and limited accessibility in poor and remote areas. To address this problem, the current solutions are using RGB images to analyze spine morphology. However, RGB images are highly susceptible to environmental factors, such as lighting conditions, compromising model stability and generalizability. Therefore, in this study, we propose a novel pipeline to accurately estimate the depth information of the unclothed back, compensating for the limitations of 2D information, and then estimate spine morphology by integrating both depth and surface information. To capture the subtle depth variations of the back surface with precision, we design an adaptive multiscale feature learning network named Grid-Aware Multiscale Adaptive Network (GAMA-Net). This model uses dual encoders to extract both patch-level and global features, which are then interacted by the Patch-Based Hybrid Attention (PBHA) module. The Adaptive Multiscale Feature Fusion (AMFF) module is used to dynamically fuse information in the decoder. As a result, our depth estimation model achieves remarkable accuracy across three different evaluation metrics, with scores of nearly 78.2%, 93.6%, and 97.5%, respectively. To further validate the effectiveness of the predicted depth, we integrate both surface and depth information for spine morphology estimation. This integrated approach enhances the accuracy of spine curve generation, achieving an impressive performance of up to 97%.