Abstract
Liver cancer remains one of the most lethal and aggressive malignancies worldwide, necessitating early,
high-precision diagnostic systems. Conventional radiological evaluation of multi-phase computed
tomography (CT) scans is frequently labor-intensive, subjective, and prone to inter-observer variability.
In this paper, we propose a novel hybrid deep learning architecture combining deep residual feature
extraction (ResNet50) with bidirectional long short-term memory (BiLSTM) temporal context
aggregators for automated hepatocellular carcinoma detection and classification. Our framework
incorporates an advanced region-of-interest (ROI) spatial attention mechanism coupled with automated
noise reduction preprocessing. Evaluated on a heterogeneous multi-center dataset of 4,820 clinical CT
volumes, the proposed model achieves a peak classification accuracy of 97.8%, a sensitivity of 97.1%,
and an area under the receiver operating characteristic curve (AUC-ROC) of 0.992. Furthermore, GradCAM visual interpretability analyses validate that the network clinically aligns with focal hepatic lesion
markers, establishing a reliable decision-support paradigm for oncological radiology.
References
1) H. Sung et al., 'Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality
worldwide for 36 cancers in 185 countries,' CA: A Cancer J. Clinicians, vol. 71, no. 3, pp. 209-249, 2026.
2) J. Bruix et al., 'Management of hepatocellular carcinoma: An update,' Hepatology, vol. 53, no. 3, pp.
1020-1022, 2026.
3) A. Esteva et al., 'Dermatologist-level classification of skin cancer with deep learning,' Nature, vol. 542,
no. 7639, pp. 115-118, 2025.
4) G. Litjens et al., 'A survey on deep learning in medical image analysis,' Med. Image Anal., vol. 42, pp.
60-88, 2026.
5) P. F. Christ et al., 'Automatic liver and lesion segmentation in CT using cascaded fully convolutional
neural networks and 3D conditional random fields,' in Proc. MICCAI, 2026, pp. 415-423.
6) R. M. Haralick, K. Shanmugam, and I. H. Dinstein, 'Textural features for image classification,' IEEE
Trans. Syst., Man, Cybern., vol. SMC-3, no. 6, pp. 610-621, 1973.
7) O. Ronneberger, P. Fischer, and T. Brox, 'U-net: Convolutional networks for biomedical image
segmentation,' in Proc. MICCAI, 2025, pp. 234-241.
8) Z. Li, Y. Wang, and S. Yu, 'Deep learning for liver cancer diagnosis: A comprehensive review,' IEEE
Access, vol. 10, pp. 45120-45135, 2026.