期刊
EUROPEAN JOURNAL OF NUCLEAR MEDICINE AND MOLECULAR IMAGING
卷 47, 期 11, 页码 2681-2690出版社
SPRINGER
DOI: 10.1007/s00259-020-04786-y
关键词
Amyotrophic lateral sclerosis; ALS mimics; Brain and spinal [F-18]-FDG PET-CT; Support vector machine; Automated spinal cord segmentation; Convolutional neural network
资金
- Fonds Wetenschappelijk Onderzoek [.., (1179620N), ...] Funding Source: Medline
- KU Leuven funds Een Hart voor ALS, Laeversfonds voor ALS Onderzoek [...] Funding Source: Medline
- als liga [...] Funding Source: Medline
- KOOR [...] Funding Source: Medline
- E. von Behring Chair for Neuromuscular and Neurodegenerative Disorders [...] Funding Source: Medline
- valerie perrier race [...] Funding Source: Medline
Purpose Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder with on average a 1-year delay between symptom onset and diagnosis. Studies have demonstrated the value of [F-18]-FDG PET as a sensitive diagnostic biomarker, but the discriminatory potential to differentiate ALS from patients with symptoms mimicking ALS has not been investigated. We investigated the combination of brain and spine [F-18]-FDG PET-CT for differential diagnosis between ALS and ALS mimics in a real-life clinical diagnostic setting. Methods Patients with a suspected diagnosis of ALS (n = 98; 64.8 +/- 11 years; 61 M) underwent brain and spine [F-18]-FDG PET-CT scans. In 62 patients, ALS diagnosis was confirmed (67.8 +/- 10 years; 35 M) after longitudinal follow-up (average 18.1 +/- 8.4 months). In 23 patients, another disease was diagnosed (ALS mimics, 60.9 +/- 12.9 years; 17 M) and 13 had a variant motor neuron disease, primary lateral sclerosis (PLS; n = 4; 53.6 +/- 2.5 years; 2 M) and progressive muscular atrophy (PMA; n = 9; 58.4 +/- 7.3 years; 7 M). Spine metabolism was determined after manual and automated segmentation. VOI- and voxel-based comparisons were performed. Moreover, a support vector machine (SVM) approach was applied to investigate the discriminative power of regional brain metabolism, spine metabolism and the combination of both. Results Brain metabolism was very similar between ALS mimics and ALS, whereas cervical and thoracic spine metabolism was significantly different (in standardised uptake values; cervical: ALS 2.1 +/- 0.5, ALS mimics 1.9 +/- 0.4; thoracic: ALS 1.8 +/- 0.3, ALS mimics 1.5 +/- 0.3). As both brain and spine metabolisms were very similar between ALS mimics and PLS/PMA, groups were pooled for accuracy analyses. Mean discrimination accuracy was 65.4%, 80.0% and 81.5%, using only brain metabolism, using spine metabolism and using both, respectively. Conclusion The combination of brain and spine FDG PET-CT with SVM classification is useful as discriminative biomarker between ALS and ALS mimics in a real-life clinical setting.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据