Cancer remains a leading cause of death. Our goal is to enhance the understanding of the biological processes underlying cancer development, with the potential to identify early biomarkers and enable personalized treatment. This project focuses on analyzing spatial single-cell RNA expressions across samples at various cancer stages. In combination with the Copy number variance analysis, we employ neural networks to detect gene signatures that drive cell-cell communication within pre-malignant and malignant areas. By analyzing these gene signatures, we aim to identify biomarkers or drug targets to prevent or treat cancer at the early stage. The project utilizes spatial and non-spatial single-cell RNA expression data from various cancers, sourced from previously published studies.