Cancer is one of the most and frequent causes of death around the world. Brain tumor is a critical and dangerous type and has a few difficulties of the techniques used for its detection; it is hard to determine its location when it is small at an early stage. The purpose of this work is to design a patch antenna sensor that is a low-cost microstrip which is suitable to detect a brain cancer tumor. The computer simulation technology CST Studio Suite 3D EM simulation and analysis was used to design a patch antenna with different frequencies of 2.8 GHz, 3.9 GHz, 5GHz and 5.6GHz to diagnose brain tumors. A comparison study between these resonance frequencies (lower-band (L-B) 2 GHz, middle-band (M-B) 3.9-5 GHz and upper-band (U-B) > 5 GHz) has been performed with six layers of brain phantom of fat, dura, brain, skin, CSF (Cerebrospinal Fluid) and skull. The designed patch sensor was assessed on both scenarios without and with a tumor cell on a brain phantom. Three parameters have been observed, the frequency phase shift, the deep amount of reflection return loss and power absorption were used to indicate the presence of the tumor cell. This study concludes that the middle-band (M-B) results in good penetration and better return loss depth around - 20dB. Meanwhile, the higher band provides high resolution of 21 MHz phase-shift but with only depth value of difference return loss of -0.1dB. The proposed work could provide a pathway on the design of patch sensors for biomedical applications.