Retraction notice to “An easy-to-use deep-learning model for highly accurate diagnosis of Parkinson's disease using SPECT images” [Comput. Med. Imaging Graphics 87 (2021) 101810] (Computerized Medical Imaging and Graphics (2021) 87, (S0895611120301051), (10.1016/j.compmedimag.2020.101810))

This article has been retracted: please see Elsevier Policy on Article Withdrawal (http://www.elsevier.com/locate/withdrawalpolicy). This article has been retracted at the request of the Editor-in-Chief in agreement with the Authors. Following a concern raised by a third party, errors in the statistical and image analysis described in this paper were found. Issue 1: The definition of healthy control (HC) is incorrect. SWEDD cohort cannot be used as HC. This issue can be resolved by dividing the population into three groups, Parkinson's disease (PD), SWEDD, and HC, instead of binary groups of PD and HC. This would require a change of methodology and that may lead to a different conclusion. Issue 2: Inadequate explanation of the use of 1359 SPECT images stated in the manuscript. The assumption that one can treat repeated scans of the same patient as independent from each other in disease classification is incorrect. The reason that the bias was not seen in health control analysis as HC only has 5% repeated scans in their study, thus the impact is rather insignificant; This does not apply to the PD cases, in which the error is about 300 %. In addition, the number of subjects in Table 6 should be 449, not 1400+ subjects as stated. These errors, though unintentional, render the conclusions of the article no longer valid. We are therefore retracting our paper and apologize to the scientific community for any inconvenience caused.