In the treatment of esophageal cancer, the high postoperative complication rate is an issue because of the extensive lymph node resection for curative purposes. To optimize the extent of resection, we developed a forceps-type coincidence radiation detector based on the concept of positron emission tomography (PET) for precise intraoperative localization of metastases. The forceps-type coincidence radiation detector consists of a pair of miniaturized PET detectors mounted on the tip of the forceps. The forceps are clamped over a suspected metastatic site intraoperatively to measure the radiation and identify the site of metastasis based on 18F-labeled fluorodeoxyglucose accumulation above a certain threshold. However, the sensitivity of the forceps-type coincidence radiation detector depends on the distance between the pair of detectors mounted on the tips of the forceps, and maintaining a constant distance between these detectors is difficult because the handle is held in the palm of the surgeon operating the device. Achieving stable coincidence counting requires correction of the coincidence radiation sensitivity according to the distance between the detectors during measurement. Utilizing the fact that the distance between detectors depends on the tip angle of the forceps, we developed a system that executes the correction based on the forceps tip angle. A linear encoder was used to measure the displacement of the forceps shaft and estimate the forceps tip angle from the amount of displacement. We also evaluated the effect of correcting radiation detection sensitivity using an experimentally developed system. The radiation detection sensitivity was corrected to a mean (± standard deviation) dose of 22.3 ± 1.3 kBq compared to the true value of 21.3 kBq. The average correction error was 5%, indicating the correction capability of the developed system. However, the variations in correction values pose a risk of missing metastases. In the future, the accuracy of the correction system will be improved, and experiments will be conducted in the clinical environment.