FA-Net: A hierarchical feature fusion and interactive attention-based network for dose prediction in liver cancer patients.

Dose prediction is a crucial step in automated radiotherapy planning for liver cancer. Several deep learning-based approaches for dose prediction have been proposed to enhance the design efficiency and quality of radiotherapy plan. However, these approaches usually take CT images and contours of organs at risk (OARs) and planning target volume (PTV) as a multi-channel input and is thus difficult to extract sufficient feature information from each input, which results in unsatisfactory dose distribution. In this paper, we propose a novel dose prediction network for liver cancer based on hierarchical feature fusion and interactive attention. A feature extraction module is first constructed to extract multi-scale features from different inputs, and a hierarchical feature fusion module is then built to fuse these multi-scale features hierarchically. A decoder based on attention mechanism is designed to gradually reconstruct the fused features into dose distribution. Additionally, we design an autoencoder network to generate a perceptual loss during training stage, which is used to improve the accuracy of dose prediction. The proposed method is tested on private clinical dataset and obtains HI and CI of 0.31 and 0.87, respectively. The experimental results are better than those by several existing methods, indicating that the dose distribution generated by the proposed method is close to that approved in clinics. The codes are available at https://github.com/hired-ld/FA-Net.

Deep learning based automatic internal gross target volume delineation from 4D-CT of hepatocellular carcinoma patients.

BACKGROUND: The location and morphology of the liver are significantly affected by respiratory motion. Therefore, delineating the gross target volume (GTV) based on 4D medical images is more accurate than regular 3D-CT with contrast. However, the 4D method is also more time-consuming and laborious. This study proposes a deep learning (DL) framework based on 4D-CT that can achieve automatic delineation of internal GTV. METHODS: The proposed network consists of two encoding paths, one for feature extraction of adjacent slices (spatial slices) in a specific 3D-CT sequence, and one for feature extraction of slices at the same location in three adjacent phase 3D-CT sequences (temporal slices), a feature fusion module based on an attention mechanism was proposed for fusing the temporal and spatial features. Twenty-six patients' 4D-CT, each consisting of 10 respiratory phases, were used as the dataset. The Hausdorff distance (HD95), Dice similarity coefficient (DSC), and volume difference (VD) between the manual and predicted tumor contour were computed to evaluate the model's segmentation accuracy. RESULTS: The predicted GTVs and IGTVs were compared quantitatively and visually with the ground truth. For the test dataset, the proposed method achieved a mean DSC of 0.869 ± 0.089 and an HD95 of 5.14 ± 3.34 mm for all GTVs, with under-segmented GTVs on some CT slices being compensated by GTVs on other slices, resulting in better agreement between the predicted IGTVs and the ground truth, with a mean DSC of 0.882 ± 0.085 and an HD95 of 4.88 ± 2.84 mm. The best GTV results were generally observed at the end-inspiration stage. CONCLUSIONS: Our proposed DL framework for tumor segmentation on 4D-CT datasets shows promise for fully automated delineation in the future. The promising results of this work provide impetus for its integration into the 4DCT treatment planning workflow to improve hepatocellular carcinoma radiotherapy.

TD-Net: A Hybrid End-to-End Network for Automatic Liver Tumor Segmentation From CT Images.

Liver tumor segmentation plays an essential role in diagnosis and treatment of hepatocellular carcinoma or metastasis. However, accurate and automatic tumor segmentation remains a challenging task, owing to vague boundaries and large variations in shapes, sizes, and locations of liver tumors. In this paper, we propose a novel hybrid end-to-end network, called TD-Net, which incorporates Transformer and direction information into convolution network to segment liver tumor from CT images automatically. The proposed TD-Net is composed of a shared encoder, two decoding branches, four skip connections, and a direction guidance block. The shared encoder is utilized to extract multi-level feature information, and the two decoding branches are respectively designed to produce initial segmentation map and direction information. To preserve spatial information, four skip connections are used to concatenate each encoder layer and its corresponding decoder layer, and in the fourth skip connection a Transformer module is constructed to extract global context. Furthermore, a direction guidance block is well-designed to rectify feature maps to further improve segmentation accuracy. Extensive experiments conducted on public LiTS and 3DIRCADb datasets validate that the proposed TD-Net can effectively segment liver tumor from CT images in an end-to-end manner and its segmentation accuracy surpasses those of many existing methods.