In recent years, various methods for hotspot detection during optical proximity correction (OPC) verification have been studied. They try to predict hotspots by analyzing optical features of aerial image such as peak intensity. However, detection accuracy in these conventional methods is still not sufficient. We cannot distinguish hotspots from nonhotspots by only focusing on aerial image of hotspot because one often becomes hotspot and the other does not despite of the same aerial images. On the other hand, optical features of pattern next to the hotspot are different even in such a case. Therefore, optical features which are extracted from surrounding patterns of hotspot are one of the promising metrics for hotspot detection. In this paper, we propose a new method to detect hotspots more accurately. A new metric, Surrounding Optical Feature (SOF), is introduced. SOF indicates optical features which are extracted from surrounding pattern of the evaluated pattern. The optical feature includes critical dimension (CD), normalized image log-slope (NILS), integral intensity, peak intensity of optical image. The proposed method consists of two steps. In step 1, appropriate SOF is extracted by using training data. In step 2, OPC verification is carried out with the SOF. The effectiveness of the proposed method is confirmed in the experimental comparisons.