During the tests, these noise sources are observed in the results. First of all, the thermal noise was easily seen in data after 30 seconds with CMOS sensor. A dark environment without any radiation and visible light sources as well as other electromagnetic noise sources was required to measure thermal noise. The test environment was covered with Cu plates and foils in order to reflect EM sources from outside, and the environment did not contain any light sources. A heavy ion source and a COTS CMOS sensor were used in a shielded environment, and the results pointed out that an incidence on a pixel creates secondary electrons which excite the neighbor pixels. Therefore, a clustering code added to find the sum of the deposited energy from the ions. Also, the code determines the thermal noise level and the energy threshold to sense an event on the detector. Besides other radiation sources, leaky circuits and heat from electronics can cause noise. Even though the power consumption of the electronics is really low, it was enough to heat up the sensor.

Also, a simple observation pointed out single-pixel outbursts occurs time to time, and most of them are near the edge pixels. Therefore, we focused on the central pixels in this work. This noise source could be detected and surpassed; however, for the sake of the real-time application, the computational complexity was limited in a certain extent.

Reading the chip faster resulted in less noise. There are some advantages of pixelated sensors such as high flux cope capability, high spatial resolution, possibility to identification of particles (cluster identification), and requiring low operation frequency. The results showed that more complex methods decrease the sampling frequency because of computation period. Real-time measurement efficiency increases with the sampling frequency in our primary tests as expected. Further tests will continue in a vacuum chamber under different conditions, and the optimization results will be improved according to these tests.