Artificial Intelligence-assisted MRI and CT-based diagnostic anatomical evaluation of experimental rat disease models

Artificial intelligence (AI) emerged as the transformative technology in biomedical imaging to improve the accuracy, efficiency and reproducibility of disease diagnosis. Integration of multimodal imaging data for the automated anatomical assessment in preclinical disease models remains insufficiently explored. This study aimed to develop and validate an AI-assisted framework integrating the magnetic resonance imaging (MRI) and computed tomography (CT) for comprehensive anatomical evaluation of the experimental rat disease models. Count of 180 Sprague-Dawley rats were allocated into healthy control and disease-induced groups representing neurological, pulmonary, hepatic and musculoskeletal disorders. High-resolution MRI and CT datasets were acquired longitudinally and processed using the deep-learning architectures for image segmentation, feature extraction, lesion detection and disease classification. Multimodal imaging repository constitutes 4320 MRI and 3960 CT image volumes, 96.8% met predefined quality criteria for analysis. Quantitative imaging revealed the significant increases in lesion volume, edema burden, tissue heterogeneity, structural distortion and tissue density across disease groups compared with controls (p < 0.001). Automated segmentation achieved Dice similarity coefficients ranging from 0.91 to 0.94, indicating excellent agreement with the expert annotations. Integrated MRI-CT AI model demonstrated superior diagnostic performance, achieving accuracy of 96.8%, sensitivity of 95.4%, specificity of 97.6% and area under the curve of 0.987, outperforming MRI-only and CT-only models. Strong correlations were observed between the AI-derived imaging biomarkers and histopathological severity scores (r = 0.86-0.93, p < 0.001). The findings demonstrated that AI-assisted multimodal MRI-CT integration provides highly accurate, reproducible and biologically relevant anatomical characterization of experimental disease models and supporting potential application in advanced preclinical imaging, translational research and future precision diagnostic systems.