The analysis of radiological images, especially computed tomography (CT) scans, is crucial for the efficient detection and segmentation of pathological areas in human body. This approach applies dynamical systems theory, phase space reconstruction, and graph analysis to enhance medical diagnostics. Utilizing entropy as a complexity measure allows for precise differentiation between healthy and pathological regions with a specific focus on kidney cancer, where exceptional results have been achieved in segmentation and anomaly detection. The implementation of the methodology in Python enables automated image processing and visualization of key patterns. In addition, entropy analysis reveals significant differences between healthy and pathological tissues. The results expand on existing methods based on fuzzy logic and neural networks, demonstrating potential for application to other medical images, such as magnetic resonance imaging (MRI) and sonography.