Developmental dysplasia of the hip (DDH) is a common pathological condition in the neonatal period, and early diagnosis is critical for treatment effectiveness. The Graf sonographic method is a validated diagnostic approach for the first three to four months of life; however, correct examination requires high technical expertise and is strongly operator-dependent. Recent advances in deep learning, particularly convolutional neural networks, have shown promising results in automated medical image segmentation. However, existing approaches for neonatal hip ultrasound lack multi-class segmentation aligned with Graf nomenclature and do not integrate acquisition quality control, highlighting a gap that the present work aims to address.