NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal, Montreal, QC, Canada; Mila - Quebec AI Institute, Montreal, QC, Canada.
NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal, Montreal, QC, Canada; Mila - Quebec AI Institute, Montreal, QC, Canada; Functional Neuroimaging Unit, CRIUGM, Université de Montréal, Montreal, QC, Canada.
Med Image Anal. 2021 Jul;71:102038. doi: 10.1016/j.media.2021.102038. Epub 2021 Mar 18.
Most image segmentation algorithms are trained on binary masks formulated as a classification task per pixel. However, in applications such as medical imaging, this "black-and-white" approach is too constraining because the contrast between two tissues is often ill-defined, i.e., the voxels located on objects' edges contain a mixture of tissues (a partial volume effect). Consequently, assigning a single "hard" label can result in a detrimental approximation. Instead, a soft prediction containing non-binary values would overcome that limitation. In this study, we introduce SoftSeg, a deep learning training approach that takes advantage of soft ground truth labels, and is not bound to binary predictions. SoftSeg aims at solving a regression instead of a classification problem. This is achieved by using (i) no binarization after preprocessing and data augmentation, (ii) a normalized ReLU final activation layer (instead of sigmoid), and (iii) a regression loss function (instead of the traditional Dice loss). We assess the impact of these three features on three open-source MRI segmentation datasets from the spinal cord gray matter, the multiple sclerosis brain lesion, and the multimodal brain tumor segmentation challenges. Across multiple random dataset splittings, SoftSeg outperformed the conventional approach, leading to an increase in Dice score of 2.0% on the gray matter dataset (p=0.001), 3.3% for the brain lesions, and 6.5% for the brain tumors. SoftSeg produces consistent soft predictions at tissues' interfaces and shows an increased sensitivity for small objects (e.g., multiple sclerosis lesions). The richness of soft labels could represent the inter-expert variability, the partial volume effect, and complement the model uncertainty estimation, which is typically unclear with binary predictions. The developed training pipeline can easily be incorporated into most of the existing deep learning architectures. SoftSeg is implemented in the freely-available deep learning toolbox ivadomed (https://ivadomed.org).
大多数图像分割算法都是针对每个像素的分类任务进行训练的，使用的是二进制掩模。然而，在医学成像等应用中，这种“黑白”方法过于受限，因为两种组织之间的对比度通常不明确，即位于物体边缘的体素包含组织的混合物（部分容积效应）。因此，分配单个“硬”标签可能会导致不利的近似。相反，包含非二进制值的软预测将克服该限制。在这项研究中，我们引入了 SoftSeg，这是一种深度学习训练方法，它利用软真实标签，不受二进制预测的限制。SoftSeg 的目标是解决回归问题而不是分类问题。这是通过以下三种方法实现的：(i)预处理和数据扩充后不进行二值化，(ii)归一化 ReLU 最终激活层（而不是 sigmoid），以及 (iii)回归损失函数（而不是传统的 Dice 损失）。我们评估了这三个特性对三个开源 MRI 分割数据集（脊髓灰质、多发性硬化症脑病变和多模态脑肿瘤分割挑战）的影响。在多次随机数据集拆分中，SoftSeg 优于传统方法，导致灰质数据集的 Dice 评分提高了 2.0%（p=0.001），脑病变提高了 3.3%，脑肿瘤提高了 6.5%。SoftSeg 在组织界面产生一致的软预测，并提高了对小物体（例如多发性硬化症病变）的敏感性。软标签的丰富性可能代表了专家间的可变性、部分容积效应，并补充了模型不确定性估计，而这通常在二进制预测中不明确。所开发的培训管道可以轻松地纳入大多数现有的深度学习架构中。SoftSeg 是在免费的深度学习工具包 ivadomed（https://ivadomed.org）中实现的。
