A functional imaging constrained diffusion (FICD) framework is proposed to synthesize high-quality 3D brain PET images from structural MRI, through a new constrained diffusion model (CDM) that ensures voxel-wise alignment between synthetic PET and ground truth.
구조적 MRI 데이터를 이용하여 기능적 PET 영상을 고품질로 합성하는 새로운 기능적 영상 제약 확산 모델 (FICD) 프레임워크를 제안한다.