Search Results
Department of Radiology, Jichi Medical University, School of Medicine, Shimotsuke, Tochigi, Japan
Search for other papers by Tomohiro Kikuchi in
Google Scholar
PubMed
Search for other papers by Shouhei Hanaoka in
Google Scholar
PubMed
Search for other papers by Takahiro Nakao in
Google Scholar
PubMed
Center for Frontier Medical Engineering, Chiba University, Yayoicho, Inage–ku, Chiba, Japan
Search for other papers by Yukihiro Nomura in
Google Scholar
PubMed
Search for other papers by Takeharu Yoshikawa in
Google Scholar
PubMed
Search for other papers by Ashraful Alam in
Google Scholar
PubMed
Search for other papers by Harushi Mori in
Google Scholar
PubMed
Search for other papers by Naoto Hayashi in
Google Scholar
PubMed
Objective
This study aimed to determine a standardized cut-off value for abnormal 18F-fluorodeoxyglucose (FDG) accumulation in the thyroid gland.
Methods
Herein, 7013 FDG–PET/CT scans were included. An automatic thyroid segmentation method using two U-nets (2D- and 3D-U-net) was constructed; mean FDG standardized uptake value (SUV), CT value, and volume of the thyroid gland were obtained from each participant. The values were categorized by thyroid function into three groups based on serum thyroid-stimulating hormone levels. Thyroid function and mean SUV with increments of 1 were analyzed, and risk for thyroid dysfunction was calculated. Thyroid dysfunction detection ability was examined using a machine learning method (LightGBM, Microsoft) with age, sex, height, weight, CT value, volume, and mean SUV as explanatory variables.
Results
Mean SUV was significantly higher in females with hypothyroidism. Almost 98.9% of participants in the normal group had mean SUV < 2 and 93.8% participants with mean SUV < 2 had normal thyroid function. The hypothyroidism group had more cases with mean SUV ≥ 2. The relative risk of having abnormal thyroid function was 4.6 with mean SUV ≥ 2. The sensitivity and specificity for detecting thyroid dysfunction using LightGBM (Microsoft) were 14.5 and 99%, respectively.
Conclusions
Mean SUV ≥ 2 was strongly associated with abnormal thyroid function in this large cohort, indicating that mean SUV with FDG–PET/CT can be used as a criterion for thyroid evaluation. Preliminarily, this study shows the potential utility of detecting thyroid dysfunction based on imaging findings.
Search for other papers by Kristine Z Swan in
Google Scholar
PubMed
Search for other papers by Johnson Thomas in
Google Scholar
PubMed
Search for other papers by Viveque E Nielsen in
Google Scholar
PubMed
Search for other papers by Marie Louise Jespersen in
Google Scholar
PubMed
Search for other papers by Steen J Bonnema in
Google Scholar
PubMed
Fairness, Accountability and Transparency , pp. 77 – 91 . Eds Sorelle AF Christo W . Proceedings of Machine Learning Research: PMLR , 2018 . 21 Thomas J Ledger GA Mamillapalli CK . Use of artificial intelligence and machine learning for
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by J L Reverter in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by L Ferrer-Estopiñan in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by F Vázquez in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by S Ballesta in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by S Batule in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by A Perez-Montes de Oca in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by C Puig-Jové in
Google Scholar
PubMed
Department of Medicine, Autonomous University of Barcelona, Barcelona, Spain
Search for other papers by M Puig-Domingo in
Google Scholar
PubMed
inevitable and the diagnostic performance of US is highly related with the experience of the clinician who performs the image acquisition ( 6 ). To improve the diagnostic accuracy and efficiency, machine learning-based computer-aided diagnosis (CAD) systems
