The study uses NLP to extract key details from pathological reports, focusing on tumor staging and lymph node involvement. Automating this process improves diagnostic accuracy and treatment planning and minimizes human errors. A key contribution is the extracting detailed information about T4 tumor stage and lymph node (N1-N3) characteristics, areas often underexplored in prior studies. The performance of the proposed NLP framework was evaluated using different Machine Learning (ML) and Deep Learning (DL) models. The Support Vector Machine (SVM) demonstrated the highest accuracy for the T1 and T2 stages, achieving 98.5% and 97.9%, respectively. In contrast, the Naive Bayes (NB) model delivered the best performance for the T3 and T4 stages, with 94.7% and 98.1% accuracy rates, respectively. Additionally, lymph node-related information was effectively identified. Gated Recurrent Unit achieved accuracy for N0 (98.26%), N2 (98.1%), and N3 (96.8%), and Bidirectional Long short-term memory achieved accuracy for N1(99.23%), showcasing the model’s ability to understand the tumor progression. This approach demonstrates significant potential for improving cancer diagnostics and streamlining clinical workflows.