4D deep image prior: dynamic PET image denoising using an unsupervised four-dimensional branch convolutional neural network

• Published in: Physics in medicine & biology Vol. 66; no. 1; pp. 15006 - 15018
• Main Authors: Hashimoto, Fumio, Ohba, Hiroyuki, Ote, Kibo, Kakimoto, Akihiro, Tsukada, Hideo, Ouchi, Yasuomi
• Format: Journal Article
• Language: English
• Published: England IOP Publishing 07.01.2021
• Subjects: Animals; Brain - diagnostic imaging; Brain - metabolism; convolutional neural network (CNN); deep image prior (DIP); deep learning; denoising; dynamic positron emission tomography (PET); Fluorodeoxyglucose F18 - metabolism; Haplorhini; Humans; Image Processing, Computer-Assisted - methods; Imaging, Three-Dimensional - methods; Neural Networks, Computer; Positron-Emission Tomography - methods; Radiopharmaceuticals - metabolism

Although convolutional neural networks (CNNs) demonstrate the superior performance in denoising positron emission tomography (PET) images, a supervised training of the CNN requires a pair of large, high-quality PET image datasets. As an unsupervised learning method, a deep image prior (DIP) has rece...