Liver tumor semantic segmentation is a crucial task in medical image analysis that requires multiple MRI modalities. This paper proposes a novel coarse-to-fine fusion segmentation approach to detect and segment small liver tumors of various sizes. To enhance the segmentation accuracy of small liver tumors, the method incorporates a detection module and a CSR (convolution-SE-residual) module, which includes a convolution block, an SE (squeeze and excitation) module, and a residual module for fine segmentation. The proposed method demonstrates superior performance compared to conventional single-stage end-to-end networks. A private liver MRI dataset comprising 218 patients with a total of 3605 tumors, including 3273 tumors smaller than 3.0 cm, were collected for the proposed method. There are five types of liver tumors identified in this dataset: hepatocellular carcinoma (HCC); metastases of the liver; cholangiocarcinoma (ICC); hepatic cyst; and liver hemangioma. The results indicate that the proposed method outperforms the single segmentation networks 3D UNet and nnU-Net as well as the fusion networks of 3D UNet and nnU-Net with nnDetection. The proposed architecture was evaluated on a test set of 44 images, with an average Dice similarity coefficient (DSC) and recall of 86.9% and 86.7%, respectively, which is a 1% improvement compared to the comparison method. More importantly, compared to existing methods, our proposed approach demonstrates state-of-the-art performance in segmenting small objects with sizes smaller than 10 mm, achieving a Dice score of 85.3% and a malignancy detection rate of 87.5%.

Diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) imaging are essential for the diagnosis of hepatocellular carcinoma (HCC) but suffer from severe displacement due to respiratory and cardiac motions. The currently available registration algorithms do not align the multi-contrast liver images satisfactorily, which hinders the efficacy of automatic lesion segmentation and radiomics analysis. We proposed a deep learning-based generative adversarial registration network (GARNET) for multi-contrast liver image registration and explored the added value on HCC lesion segmentation in patients with cirrhosis. The proposed model applied a pre-trained generative adversarial network to synthesize DCE images from DWI and then used diffeomorphic based symmetric image normalization (SyN) algorithm for registration. Furthermore, an attention-based U-net (AU-NET) was used to segment the HCC lesion from registered DWI and DCE images. GARNET was developed and validated on 901 patients collected between January 2013 and February 2021 across five medical centers, including 517 HCC patients, total of 597 HCCs, and every case was interpolated to 32 slices. GARNET achieved the highest liver registration Dice (0.812 ± 0.045) among all comparative methods, and the lowest tumor target registration error (3.470 ± 1.652 mm) for DWI/DCE registrations. In addition, the performance of HCC segmentation was significantly improved when using registered images from GARNET, with Dice improvements from 0.36 ± 0.22 to 0.45 ± 0.22; 0.53 ± 0.25 to 0.58 ± 0.21; and 0.44 ± 0.26 to 0.56 ± 0.19 in three validation centers. Experimental results showed that GARNET significantly improved the liver registration performances among multi-contrast images, thereby improving the further identification and segmentation of HCCs.