Hardware Co-Simulation Based Deep Neural Networks for Liver Diseases Detection in MRI Images
DOI:
https://doi.org/10.59992/IJCI.2026.v5n2p1الكلمات المفتاحية:
Liver MRI، Multi-Class Classification، Focal Liver Lesions، Quantization-Aware Training، FPGA Acceleration، Vitis HLS، C/RTL Co-Simulation، Edge AIالملخص
Accurate characterization of focal liver lesions on multiphasic magnetic resonance imaging (MRI) is central to early detection and treatment planning; however, clinical deployment of deep learning remains constrained by compute cost, latency, and the need for rigorous verification of hardware implementations. This paper presents LiverNet-Q, a hardware-aware, multi-phase deep neural network for multi-class liver disease detection from MRI, coupled with an end-to-end hardware co-simulation workflow that validates functional equivalence between the software model and the synthesized register-transfer level (RTL) design. The proposed pipeline first localizes the liver with a lightweight U-Net trained using public liver MRI annotations, then classifies lesions into five clinically relevant categories (normal liver, hepatocellular carcinoma, hemangioma, focal nodular hyperplasia, and simple cyst) using attention-based phase fusion. To enable resource-efficient inference, LiverNet-Q is trained with quantization-aware training and deployed in INT8 precision. The accelerator is implemented with Vitis HLS using a streaming dataflow micro-architecture that targets an initiation interval of one for core convolution operators. Experiments on public benchmarks demonstrate that INT8 deployment preserves diagnostic performance with a small loss relative to FP32 while providing substantial speedups. Hardware co-simulation reports confirm cycle-accurate latency and throughput, supporting reproducible, deployment-ready evaluation.
المراجع
1. O. Ronneberger, P. Fischer, and T. Brox, 'U-Net: Convolutional Networks for Biomedical Image Segmentation,' MICCAI, 2015. https://arxiv.org/abs/1505.04597.
2. F. Isensee et al., 'nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation,' Nature Methods, 2021. https://doi.org/10.1038/s41592-020-01008-z.
3. K. He, X. Zhang, S. Ren, and J. Sun, 'Deep Residual Learning for Image Recognition,' CVPR, 2016. https://arxiv.org/abs/1512.03385.
4. M. Sandler et al., 'MobileNetV2: Inverted Residuals and Linear Bottlenecks,' CVPR, 2018. https://arxiv.org/abs/1801.04381.
5. M. Tan and Q. Le, 'EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,' ICML, 2019. https://arxiv.org/abs/1905.11946.
6. A. E. Kavur et al., 'CHAOS [1] Challenge -Combined (CT-MR) Healthy Abdominal Organ Segmentation,' Medical Image Analysis, 2021. https://doi.org/10.1016/j.media.2020.102174.
7. M. Gross et al., 'LiverHccSeg: A publicly available multiphasic MRI dataset with liver and HCC tumor segmentations and inter-rater agreement analysis,' Data in Brief, 2023. https://doi.org/10.1016/j.dib.2023.109662.
8. H. Huber et al., 'LiverHccSeg (Zenodo dataset record),' 2023. https://zenodo.org/records/7957516.
9. LiMT Consortium, 'LiMT: a multi-task liver image benchmark dataset,' arXiv, 2025. https://arxiv.org/abs/2511.19889.
10. M. J. A. Jansen et al., 'Automatic classification of focal liver lesions based on MRI and clinical data,' PLOS ONE, 2019. https://doi.org/10.1371/journal.pone.0217053.
11. K. Wang et al., 'Fully automating LI-RADS on MRI with deep learning-guided lesion segmentation and classification,' 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10233056/.
12. Y. Wu et al., 'Deep learning LI-RADS grading system based on contrast-enhanced multiphase MRI,' Annals of Translational Medicine, 2020. https://cdn.amegroups.cn/journals/amepc/files/journals/16/articles/35992/public/35992-PB2-5679-R2.pdf.
13. M. Gatti et al., 'Benign focal liver lesions: The role of magnetic resonance imaging,' World Journal of Gastroenterology, 2022. https://pmc.ncbi.nlm.nih.gov/articles/PMC9157713/.
14. G. W. Hu et al., 'Diagnosis of liver hemangioma using magnetic resonance imaging diffusion-derived vessel density,' 2024. https://pmc.ncbi.nlm.nih.gov/articles/PMC11651954/.
15. R. Stollmayer et al., 'LI-RADS-based hepatocellular carcinoma risk mapping using deep learning,' EJNMMI Research, 2025. https://doi.org/10.1186/s40644-025-00844-6.
16. F. Quinton et al., 'A Tumour and Liver Automatic Segmentation (ATLAS) dataset for contrast-enhanced liver MRI [8],' Data, 2023. https://doi.org/10.3390/data8050079.
17. S. Wu et al., 'A Coarse-to-Fine Fusion Network for Small Liver Tumor Segmentation in MRI,' Diagnostics, 2023. https://doi.org/10.3390/diagnostics13152504.
18. J. Song et al., 'Liver-VLM: Enhancing Focal Liver Lesion Classification on Multiphase MRI,' Applied Sciences, 2025. https://doi.org/10.3390/app152312578.
19. S. Mulay et al., 'Liver Segmentation from Multimodal Images using HED and Mask R-CNN,' arXiv, 2019. https://arxiv.org/abs/1910.10504.
20. A. G. Howard et al., 'Searching for MobileNetV3,' ICCV, 2019. https://arxiv.org/abs/1905.02244.
21. B. Jacob et al., 'Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference,' CVPR, 2018. https://arxiv.org/abs/1712.05877.
22. S. Han, H. Mao, and W. J. Dally, 'Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding,' ICLR, 2016. https://arxiv.org/abs/1510.00149.
23. V. Sze et al., 'Efficient Processing of Deep Neural Networks: A Tutorial and Survey,' Proceedings of the IEEE, 2017. https://doi.org/10.1109/JPROC.2017.2761740.
24. Y. Chen et al., 'Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep CNNs,' ISSCC, 2016. https://doi.org/10.1109/ISSCC.2016.7418007.
25. M. Blott et al., 'FINN-R: An End-to-End Deep-Learning Framework for Fast Exploration of Quantized Neural Networks,' ACM TRETS, 2021. https://doi.org/10.1145/3441302.
26. FINN Project, 'FINN: A Framework for Fast, Scalable FPGA [7]-based QNN Inference,' GitHub repository. https://github.com/Xilinx/finn.
27. J. Duarte et al., 'Fast inference of deep neural networks in FPGA for particle physics,' JINST, 2018. (hls4ml) https://doi.org/10.1088/1748-0221/13/07/P07027.
28. AMD, 'Vitis High-Level Synthesis User Guide (UG1399): Running C/RTL Co-Simulation,' 2025. https://docs.amd.com/r/en-US/ug1399-vitis-hls/Running-C/RTL-Co-Simulation.
29. AMD, 'Vitis AI [5] User Guide,' 2025. https://docs.amd.com/r/en-US/Vitis-AI.
30. Xilinx, 'Vitis HLS [4] Introductory Examples,' GitHub repository. https://github.com/Xilinx/Vitis-HLS-Introductory-Examples.
