Quantum information processing, that is, information processing that relies its principles on the laws of quantum mechanics can go beyond the limitations of today's information technologies. One of the challenging problems toward realizing practical quantum information applications is to tackle the intervention of quantum noise. The currently available quantum technologies, however, contain noise in all the steps such as preparation, manipulation, and measurement of qubits, dubbed noisy intermediate-scale quantum (NISQ) technologies. In NISQ systems, errors may occur due to unwanted interactions both with an environment and within systems, for which little is yet known about verification, characterization, and quantification of the errors. In the present thesis, we present the verification of noise existing in a measurement of multiple qubits in NISQ systems by performing quantum tomography on detectors in IBMQ and Rigetti via cloud-based quantum computing services. It turns out that detectors in the NISQ devices share unwanted correlations when multiple qubits are measured, and cause the so-called measurement crosstalk errors. The correlations existing in a quantum measurement also contain entanglement whenever detectors work for more than two qubits. We also provide a method of analyzing and quantifying crosstalk errors in a quantum measurement.