This study presents a new method to characterize the tumor microenvironment (TME), focusing on its dynamic interactions that impact cancer progression. Despite advancements in computational pathology, certain limitations still exist. We propose defining TME through interaction zones between tumor, necrotic, and stroma tissues, classified with an SVM (80% accuracy). To validate TME's relevance, we associated it with survival using multilayer perceptron (MLP) and Cox regression modeling. Results showed significant correlations with handcrafted TME features, yielding 80% accuracy in the test set and 81% in validation for MLP. The Cox regression unveiled a noteworthy hazard ratio of 0.36 (95% CI: 0.23-0.55, p !' 0.05), indicating a significant TME-survival link. This methodology suggests that TME organization could serve as a predictive marker in serous carcinoma of the ovary, providing valuable insights into the role of the tumor microenvironment in disease development.