DL-based 医学图像数据预处理和训练的一般过程

一、Semi-supervised Medical Image Segmentation through Dual-task Consistency

Observing thatmulti/dual-task learning attends to various levels of information which have inherent prediction perturbation, we ask the question in this work: can we explicitly build task-level regularization rather than implicitly constructing networks and/or data-level perturbation and then regularization for SSL? To answer this question, we propose a novel dual-task-consistency semi-supervised framework for the first time. Concretely, we use a dual-task deep network that jointly predicts a pixel-wise segmentation map and a geometry-aware level set representation of the target. The level set representation is converted to an approximated segmentation map through a differentiable task transform layer. Simultaneously, we introduce a dual-task consistency regularization between the level set-derived segmentation maps and directly predicted segmentation maps for both labeled and unlabeled data. Extensive experiments on two public datasets show that our method can largely improve the performance by incorporating the unlabeled data. Meanwhile, our framework outperforms the state-of-the-art semi-supervised learning methods. Code is available at: HiLab-git/DTC

Pre-processing

In pre-processing, we use the soft tissue CT window range of [−125, 275] HU (Zhou et al. 2019a)¹, and resample all images to an isotropic resolution of 1.0×1.0×1.0mm. Finally, we crop the images centering at the pancreas region based on the ground truth with enlarged margins (25 voxels) and normalize them as zero mean and unit variance.

Training

Implementation Details and Evaluation Metrics: We implement our framework in PyTorch (Paszke et al. 2019)², using an NVIDIA 1080TI GPU. In this work, we use VNet (Milletari, Navab, and Ahmadi 2016)³ as the backbone for all experiments, and we implement dual-task VNet by adding a new regression layer at the end of the original VNet. The framework is trained by an SGD optimizer for 6000 iterations, with an initial learning rate (lr) 0.01 decayed by 0.1 every 2500 iterations. The batch size is 4, consisting of 2 labeled images and 2 unlabeled images.
…
To avoid over-fitting, we use the standard on-the-fly data augmentation methods during training stage (Yu et al. 2019)⁴.

---

1. Prior-aware neural network for partially-supervised multi-organ segmentation. In ICCV ↩︎

2. Pytorch: An imperative style, high-performance deep learning library. In NeurIPS ↩︎

3. V -net: Fully convolutional neural networks for volumetric medical image segmentation. In 3DV, 565–571. IEEE. ↩︎

4. Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation. In MICCAI ↩︎