Primary liver cancer ranks as the sixth most commonly diagnosed cancer and the third leading cause of cancer-related mortality worldwide. It primarily consists of two pathological subtypes: hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) [[1], [2], [3]]. Compared to HCC, ICC is notably more aggressive and is associated with a distinctly poorer prognosis [4,5]. ICC typically necessitates more extensive hepatic resection and currently lacks effective targeted or immunotherapeutic strategies [6,7]. Consequently, accurately differentiating between HCC and ICC is critical for guiding personalized treatment decisions and improving patient outcomes [[8], [9], [10]].

Currently, histological biopsy remains the gold standard for diagnosis, directly revealing the microstructural differences between these two entities: HCC typically exhibits densely packed trabecular architecture, whereas ICC demonstrates loosely arranged glandular structures embedded within abundant fibrous stroma [[11], [12], [13]]. However, the invasiveness of biopsy and the time lag in obtaining a definitive pathological diagnosis limit its application in progressive diseases for timely diagnosis and treatment planning. Conventional imaging techniques continue to face challenges in resolving tumor microstructure: although dynamic contrast-enhanced MRI serves as the primary method for differentiating HCC from ICC, substantial overlap in imaging features persists [14,15]. Diffusion-weighted imaging (DWI)-derived apparent diffusion coefficient (ADC) values reflect water diffusion restriction and indirectly correlate with cellular density, yet ADC measurements are confounded by multiple factors including cell volume and membrane permeability, making them insufficient to reliably differentiate the microstructural disparities between HCC and ICC [[16], [17], [18]]. Intravoxel incoherent motion (IVIM) modeling decouples pure diffusion (D) from perfusion parameters (f, D*) via a bi-exponential approach, while the stretched exponential model (SEM) quantifies tissue heterogeneity through the heterogeneity index (α) and distributed diffusion coefficient (DDC) [[19], [20], [21]]. Although both techniques show potential in HCC-ICC differentiation, they remain incapable of directly characterizing the core pathological features of ICC.

In recent years, time-dependent diffusion magnetic resonance imaging (Td-dMRI) has emerged as a promising approach for characterizing tumor microstructure. This technique integrates oscillating gradient spin-echo (OGSE) and pulsed gradient spin-echo (PGSE) sequences, enabling the tuning of effective diffusion times from sub-millisecond to millisecond scales [[22], [23], [24]]. This significantly enhances sensitivity to structural features at the 1–10 μm scale. This study selected the IMPULSED model because the core pathological differences between HCC and ICC including cell diameter (d), intracellular volume fraction (Vin), extracellular diffusion coefficient (Dex), and cellularity are key biophysical properties that this model aims to quantify [25,26]. This enables the establishment of a direct link between imaging parameters and pathology. Compared to simplified models like ADC or IVIM that confound multiple effects, IMPULSED holds promise for delivering more specific microstructural insights. Combined with the IMPULSED model, Td-dMRI allows for the quantification of key microstructural parameters including cell diameter (d), intracellular volume fraction (Vin), extracellular diffusion coefficient (Dex), and cellularity, thereby providing a more accurate reflection of tumor microenvironment heterogeneity [25,26].

Although Td-dMRI has made significant advances in the diagnosis of prostate cancer, glioma, and breast cancer [[25], [26], [27], [28]], its application in liver tumors, particularly for the critical clinical differential diagnosis between HCC and ICC, remains very limited. Therefore, this study aims to investigate for the first time the ability of Td-dMRI combined with the IMPULSED model to non-invasively distinguish HCC from ICC in a nude mouse animal model. This approach seeks to establish novel imaging biomarkers based on tumor microstructure to support precision diagnosis.