A Deep Learning Model to Predict a Diagnosis of Alzheimer Disease by Using 18 F-FDG PET of the Brain
Problem
Predicting the final diagnosis of Alzheimer disease, mild cognitive impairment, or neither from brain 18F-FDG PET images is difficult for radiologic readers.
Why it matters
Accurate early diagnosis of Alzheimer disease affects patient care and treatment decisions.
What the AI did
A convolutional neural network (InceptionV3) was trained on 90% of the ADNI PET dataset (2109 studies, 1002 patients) and tested on held-out ADNI data plus an independent 40-study test set, with performance compared to radiologic readers.
Results
not stated in source (abstract truncated at 'The algorithm')
Impact
Validated on an independent test set against radiologic readers, but performance numbers are cut off in the abstract.
Record
| Venue | Radiology |
|---|---|
| Published | 2018 |
| Work type | article |
| Authors | Yiming Ding, Jae Ho Sohn, Michael G. Kawczynski, Hari Trivedi, Roy J. Harnish, Nathaniel W. Jenkins, Dmytro Lituiev, Timothy P. Copeland et al. |
| Citations | 574 (OpenAlex, at import time) |
| References | 24 |
| Model or method | InceptionV3 convolutional neural network |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1148/radiol.2018180958 |
| Open access | green |
| Keywords | deep learning, Alzheimer disease, FDG PET, InceptionV3, diagnosis |
Importance rating
major — millions of lives, an industry or a field
Alzheimer's diagnosis affects treatment decisions for a major neurodegenerative disease; abstract gives no disease-scale figures but problem is clinically central.