Interaction of immune system and cancer cells plays a crucial role in defining the physical and chemical characteristics of the tumor microenvironment. Consequently, monitoring and analyzing these interactions may prove essential for cancer diagnosis and prognosis. While standard techniques assessing cellular interaction are technically complicated and usually expensive, engineering solutions can be employed to introduce novel methods and effective techniques. In this paper, we presented a new blood-based breast cancer hallmark for people suspected of breast tumor disease (BTD) by time-lapse microscopy imaging from the interaction between the patient’s blood and MDA-MB-231 breast cancer cell lines. The detection protocol is based on the quantifying invasion of the WBCs to the breast cancer cell line. Many cytological, molecular, and immunofluorescent assays were carried out to approve the hypothesis. Blood immune cells showed meaningful invasion patterns to breast cancer cell lines in reverse correlation by the cancerous stage of the patients. Hence, we believe that the immune system is cognizant of the neoplastic nature of breast tumor disease. To eliminate all the human-related limitations, a convolutional neural network (CNN) architecture was used for invasion recognition. The proposed CNN architecture showed an accuracy of approximately 86%, making it a reliable, fast, and easy way for intelligent detection of invasion patterns to decide on the tumor stage. Results made us present the hypothesis that people with more aggressive breast cancer tumors have less strong immune cells to invade cancer cells which could be a start in the clinical use of the cellular-based immune system for cancer investigation. As WBCs were isolated from the blood with no pre-processing, this method would shed new light as a simple complementary method for better clarification of tumor nature.