This study investigates the impact of computed tomography (CT) protocol optimization on image quality and dose in pediatric imaging using both standard (Catphan \(^{\text{\textregistered} }\) 600) and anthropomorphic 3D-printed pediatric phantoms. Eight protocols were evaluated on a hybrid PET/CT system, varying in tube voltage, current, slice thickness, and bow-tie filter selection. Image quality was assessed using task-based metrics including the task transfer function (TTF), noise power spectrum (NPS), and detectability index ( \(d'\) ), analyzed through IQWorks and ImQuest software platforms. Results demonstrate that pediatric protocols achieve higher spatial resolution and detectability at significantly lower radiation doses compared to adult protocols. The choice of bow-tie filter had a marked effect on noise distribution and dose uniformity, with pediatric (head) filters providing better image quality for small-size phantoms. The detectability index correlated positively with both contrast and dose, while further voltage reduction below 100 kV did not improve performance. These findings reinforce the importance of protocol tailoring for pediatric patients, highlighting the value of task-based metrics in CT quality optimization.