AI approach of cycle-consistent generative adversarial networks to synthesize PET images to train computer-aided diagnosis algorithm for dementia

• Published in: Annals of nuclear medicine Vol. 34; no. 7; pp. 512 - 515
• Main Authors: Kimura, Yuichi, Watanabe, Aya, Yamada, Takahiro, Watanabe, Shogo, Nagaoka, Takashi, Nemoto, Mitsutaka, Miyazaki, Koichi, Hanaoka, Kohei, Kaida, Hayato, Ishii, Kazunari
• Format: Journal Article
• Language: English
• Published: Singapore Springer Singapore 01.07.2020; Springer Nature B.V
• Subjects: Algorithms; Amyloid; Artificial intelligence; Degeneration; Dementia; Dementia - diagnostic imaging; Dementia disorders; Diagnosis; Diagnosis, Computer-Assisted; Frontotemporal dementia; Generative adversarial networks; Humans; Image Processing, Computer-Assisted - methods; Imaging; Lewy bodies; Medical diagnosis; Medicine; Medicine & Public Health; Neural Networks, Computer; Nuclear Medicine; Positron emission; Positron-Emission Tomography; Radiology; Short Communication; Synthesis; Tomography; Training; AI (artificial intelligence); CAD (computer-aided diagnosis); Amyloid imaging
• ISSN: 0914-7187; 1864-6433; 1864-6433
• DOI: 10.1007/s12149-020-01468-5

Objective An artificial intelligence (AI)-based algorithm typically requires a considerable amount of training data; however, few training images are available for dementia with Lewy bodies and frontotemporal lobar degeneration. Therefore, this study aims to present the potential of cycle-consistent generative adversarial networks (CycleGAN) to obtain enough number of training images for AI-based computer-aided diagnosis (CAD) algorithms for diagnosing dementia. Methods We trained CycleGAN using 43 amyloid-negative and 45 positive images in slice-by-slice. Results The CycleGAN can be used to synthesize reasonable amyloid-positive images, and the continuity of slices was preserved. Discussion Our results show that CycleGAN has the potential to generate a sufficient number of training images for CAD of dementia.