Perineural invasion (PNI) is a common pathological feature in pancreatic cancer, closely associated with increased pain, higher rates of metastasis, and poorer prognosis. Accurate preoperative prediction of PNI is crucial for informing surgical strategies and optimizing individualized treatment planning. However, at present, the diagnosis of PNI in pancreatic cancer relies solely on postoperative pathological examination, with no effective preoperative assessment methods available. This study aims to develop a predictive model for PNI using a novel habitat analysis based on multiparametric MRI (MP-MRI), combined with traditional radiomics features, to improve decision-making and patient outcomes.