In the ever-evolving realm of dental imaging, the significance of 2D dental panoramic images has undergone a notable surge, particularly in the realms of dental diagnosis and surgical planning. The meticulous segmentation of these images stands out as a pivotal task, endowing healthcare professionals with indispensable anatomical insights and paving the way for the progression of computer-aided diagnosis. Through a meticulously refined segmentation process, practitioners are empowered to attain heightened precision in diagnosing dental lesions, thereby fostering the evolution of more efficacious treatment plans. In our research endeavors, we chose to leverage the nnU-Net as our baseline network. The post-processing stage within our methodology assumed a role of paramount importance, integrating connected domain analysis. This strategic approach proved instrumental in honing the predicted images, aligning them more closely with established medical prior knowledge, and ensuring a heightened level of consistency in the segmentation outcomes. As a tangible testament to the efficacy of our approach, the preliminary-round score ultimately is 0.9475. This numerical validation underscores the success of our methodology in enhancing the precision of dental image segmentation, thereby laying the groundwork for advancements in computer-aided diagnosis within the realm of dental care.