Aerial images are formed with anisotropy due to diffraction from the array of micro-optics that compose the aerial imaging optics. The point spread function (PSF) is used as an indicator to evaluate the spatial resolution of aerial imaging optics, and 4 times standard deviation (D4σ), which can be adapted to non-Gaussinan distributions, is considered appropriate to represent the aerial image diameter. However, abnormal values called hot pixels, which do not perform their normal function on the imaging sensor have a significant impact on the output value of D4σ in terms of calculation. In this study, we aimed to clarify the usefulness of hot pixels elimination. To achieve this, hot pixels are defined as hot pixels as pixels that are above the top 0.01% of the maximum luminance value and compare the D4σ results before and after elimination. As a result, the accuracy of D4σ is confirmed to be improved by eliminating hot pixels. Therefore, the elimination of hot pixels is considered to be an effective method to improve the accuracy of PSF quantification.