3DMAU-Net: liver segmentation network based on 3D U-Net

Abstract

Considering the three-dimensional (3D) U-Net lacks sufficient local feature extraction for image features and lacks attention to the fusion of high- and low-level features, we propose a new model called 3DMAU-Net based on the 3D U-Net architecture for liver region segmentation. Our model replaces the last two layers of the 3D U-Net with a sliding window-based multilayer perceptron (SMLP), enabling better extraction of local image features. We also design a high- and low-level feature fusion dilated convolution block that focuses on local features and better supplements the surrounding information of the target region. This block is embedded in the entire encoding process, ensuring that the overall network is not simply downsampling. Before each feature extraction, the input features are processed by the dilated convolution block. We validate our experiments on the liver tumor segmentation challenge 2017 (Lits2017) dataset, and our model achieves a Dice coefficient of 0.95, which is an improvement of 0.015 compared to the 3D U-Net model. Furthermore, we compare our results with other segmentation methods, and our model consistently outperforms them.

References

[1] AghamohammadiA, RanjbarzadehR, NaiemiF, et al.. TPCNN: two-path convolutional neural network for tumor and liver segmentation in CT images using a novel encoding approach. Expert systems with applications, 2021, 183: 115406 J]

[2] SiriS K, KumarS P, LatteV M. Threshold-based new segmentation model to separate the liver from CT scan images. IETE journal of research, 2022, 68(6): 4468-4475 J]

[3] RonnebergerO, FischerP, BroxT. U-net: convolutional networks for biomedical image segmentation. 18th International Conference on Medical Image Computing and Computer-Assisted Intervention, October 5–9, 2015, Munich, Germany, 2015, Heidelberg, Springer International Publishing: 234-241[C]

[4] ZhuF, CuiJ, ZhuB, et al.. Semantic segmentation of urban street scene images based on improved U-Net network. Optoelectronics letters, 2023, 19(3): 179-185 J]

[5] ZhouZ, RahmanS M M, TajbakhshN, et al.. Unet++: a nested U-net architecture for medical image segmentation. Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, September 20, 2018, Granada, Spain, 2018, Heidelberg, Springer International Publishing: 3-11 C]

[6] ValanarasuJ M J, PatelV M. Unext: MLP-based rapid medical image segmentation network. International Conference on Medical Image Computing and Computer-Assisted Intervention, October 8–12, 2022, Vancouver, BC, Canada, 2022, Cham, Springer Nature Switzerland: 23-33[C]

[7] ÇiçekÖ, AbdulkadirA, LienkampS S, et al.. 3D U-Net: learning dense volumetric segmentation from sparse annotation. 19th International Conference on Medical Image Computing and Computer-Assisted Intervention, October 17–21, 2016, Athens, Greece, 2016, Heidelberg, Springer International Publishing: 424-432[C]

[8] MILLETARI F, NAVAB N, AHMADI S A. V-net: fully convolutional neural networks for volumetric medical image segmentation[EB/OL]. (2016-06-15) [2024-01-23]. https://arxiv.org/abs/1606.04797.

[9] AhmedS F, RahmanF S, TabassumT, et al.. 3D U-Net: fully convolutional neural network for automatic brain tumor segmentation. 22nd International Conference on Computer and Information Technology (ICCIT), December 18–20, 2019, Dhaka, Bangladesh, 2019, New York, IEEE: 1-6[C]

[10] AFIFA K, AHMED M, KUN H, et al. 3D-DenseUNet: a new model for brain segmentation in magnetic resonance images[EB/OL]. (2023-12-06) [2024-01-23]. https://doi.org/10.21203/rs.3.rs-3710057/v1.

[11] KemassiO, MaamriO, BouananeK, et al.. Dilated convolutions based 3D U-net for multi-modal brain image segmentation. International Conference on Artificial Intelligence and its Applications, January 24–26, 2021, 2021, Cham, Springer International Publishing: 428-436[C]

[12] KeT, YangX, ZhangX, et al.. VAC-UNet: visual attention convolution U-Net for 3D medical image segmentation. IEEE International Conference on Medical Artificial Intelligence, November 18–19, 2023, Beijing, China, 2023, New York, IEEE: 435-440[C]

[13] YANG J, MARCUS D S, SOTIRAS A. Dynamic U-net: adaptively calibrate features for abdominal multi-organ segmentation[EB/OL]. (2024-03-14) [2024-04-13]. https://arxiv.org/abs/2403.07303.

[14] CHEN J, MEI J, LI X, et al. 3D TransUNet: advancing medical image segmentation through vision transformers[EB/OL]. (2024-03-14) [2024-04-13]. https://arxiv.org/abs/2310.07781.

[15] ABOUSSALEH I, RIFFI J, EL FAZAZY K, et al. 3DUV-NetR+: a 3D hybrid semantic architecture using transformers for brain tumor segmentation with multimodal MR images[J]. Results in engineering, 2024: 101892.

[16] XieE, WangW, YuZ, et al.. SegFormer: simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems, 2021, 34: 12077-12090[J]

[17] WooS, ParkJ, LeeJ Y, et al.. CBAM: convolutional block attention module. Proceedings of the European Conference on Computer Vision, September 8–14, 2018, Munich, Germany, 2018, Heidelberg, Springer International Publishing: 3-19[C]

[18] SalehiS S M, ErdogmusD, GholipourA. Tversky loss function for image segmentation using 3D fully convolutional deep networks. International Workshop on Machine Learning in Medical Imaging, September 10, 2017, Quebec City, QC, Canada, 2017, Cham, Springer International Publishing: 379-387[C]

[19] XiaoX, HuQ V, WangG. Edge-aware multi-task network for integrating quantification segmentation and uncertainty prediction of liver tumor on multi-modality non-contrast MRI. International Conference on Medical Image Computing and Computer-Assisted Intervention, October 8–12, 2023, Vancouver, BC, Canada, 2023, Cham, Springer Nature Switzerland: 652-661[C]

[20] ChenY, LiuL, LiJ, et al.. MetaLR: meta-tuning of learning rates for transfer learning in medical imaging. International Conference on Medical Image Computing and Computer-Assisted Intervention, October 8–12, 2023, Vancouver, BC, Canada, 2023, Cham, Springer Nature Switzerland: 706-716[C]

[21] YouC, DaiW, MinY, et al.. Implicit anatomical rendering for medical image segmentation with stochastic experts. International Conference on Medical Image Computing and Computer-Assisted Intervention, October 8–12, 2023, Vancouver, BC, Canada, 2023, Cham, Springer Nature Switzerland: 561-571[C]

[22] OhN, KimJ H, RhuJ, et al.. Automated 3D liver segmentation from hepatobiliary phase MRI for enhanced preoperative planning. Scientific reports, 2023, 13(1): 17605 J]

[23] AnsariM Y, YangY, MeherP K, et al.. Dense-PSP-UNet: a neural network for fast inference liver ultrasound segmentation. Computers in biology and medicine, 2023, 153: 106478 J]

RIGHTS & PERMISSIONS

Tianjin University of Technology