Ψ-Net: Focusing on the border areas of intracerebral hemorrhage on CT images

Zhuo Kuang,Xianbo Deng,Li Yu,Hongkui Wang,Tiansong Li,Shengwei Wang

DOI: https://doi.org/10.1016/j.cmpb.2020.105546

IF: 6.1

2020-10-01

Computer Methods and Programs in Biomedicine

Abstract:<p>Background and objective: The volume of the intracerebral hemorrhage (ICH) obtained from CT scans is essential for quantification and treatment planning. However,a fast and accurate volume acquisition brings great challenges. On the one hand, it is both time consuming and operator dependent for manual segmentation, which is the gold standard for volume estimation. On the other hand, low contrast to normal tissues, irregular shapes and distributions of the hemorrhage make the existing automatic segmentation methods hard to achieve satisfactory performance. Method: To solve above problems, a CNN-based architecture is proposed in this work, consisting of a novel model, which is named as Ψ-Net and a multi-level training strategy. In the structure of Ψ-Net, a self-attention block and a contextual-attention block is designed to suppresses the irrelevant information and segment border areas of the hemorrhage more finely. Further, an multi-level training strategy is put forward to facilitate the training process. By adding the slice-level learning and a weighted loss, the multi-level training strategy effectively alleviates the problems of vanishing gradient and the class imbalance. The proposed training strategy could be applied to most of the segmentation networks, especially for complex models and on small datasets. Results: The proposed architecture is evaluated on a spontaneous ICH dataset and a traumatic ICH dataset. Compared to the previous works on the ICH sementation, the proposed architecture obtains the state-of-the-art performance(Dice of 0.950) on the spontaneous ICH, and comparable results(Dice of 0.895) with the best method on the traumatic ICH. On the other hand, the time consumption of the proposed architecture is much less than the previous methods on both training and inference. Morever, experiment results on various of models prove the universality of the multi-level training strategy. Conclusions: This study proposed a novel CNN-based architecture, Ψ-Net with multi-level training strategy. It takes less time for training and achives superior performance than previous ICH segmentaion methods.</p>

engineering, biomedical,computer science, interdisciplinary applications,medical informatics, theory & methods

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to quickly and accurately segment the intracerebral hemorrhage (ICH) area in CT images. Specifically:

1. **Challenges of Manual Segmentation**: Although manual segmentation is regarded as the gold standard for volume estimation, it is time - consuming and operator - dependent, which is impractical in clinical applications.
2. **Difficulties of Automatic Segmentation**: Existing automatic segmentation methods are difficult to achieve satisfactory performance due to low contrast, irregular shapes and distributions.
3. **Partial Volume Effect**: Especially for traumatic ICH, the partial volume effect makes it difficult to distinguish the bleeding area from normal tissues.
4. **Detection of Small Bleeding Areas**: Small bleeding areas are easily overlooked.

To solve these problems, the author proposes a new architecture based on convolutional neural network (CNN), named /Psi1 - Net, and adopts a multi - level training strategy. The model improves the segmentation accuracy of the bleeding boundary by introducing self - attention blocks and context - attention blocks, and optimizes the training process through a multi - level training strategy, thereby reducing the training time while ensuring accuracy.