Diagnostic potential of automated subcortical volume segmentation in atypical parkinsonism

Scherfler, Christoph; Göbel, Georg; Müller, Christoph; Nocker, Michael; Wenning, Gregor K.; Schocke, Michael; Poewe, Werner; Seppi, Klaus

doi:10.1212/wnl.0000000000002518

Cited by 92 publications

(121 citation statements)

References 31 publications

Supporting

Mentioning

112

Contrasting

Order By: Relevance

“…Finally, this study did not compare our models with conventional MRI clues used by radiologists in these disorders, such as the “hummingbird” and “hot cross bun” signs, midbrain atrophy, and putaminal T2-weighted hypointensity. 50–52 Notably, Reiter et al, using visual rating of dorsolateral nigral hyperintensity in susceptibility weighted images, showed promising discriminability in differentiating parkinsonian syndromes from controls. 51 It will be important to discern the additional value a quantitative MRI marker derived from combining DTI and R2* provides compared to the best medical knowledge.…”

Section: Discussionmentioning

confidence: 99%

Combined Diffusion Tensor Imaging and Apparent Transverse Relaxation Rate Differentiate Parkinson Disease and Atypical Parkinsonism

Du¹,

Lewis²,

Kanekar³

et al. 2017

AJNR Am J Neuroradiol

View full text Add to dashboard Cite

Background and Purpose Both diffusion tensor imaging (DTI) and R2* have shown promise in differentiating Parkinson’s disease (PD) from atypical parkinsonism [particularly multiple system atrophy (MSA) and progressive supranuclear palsy (PSP)]. We assessed DTI, R2*, and their combination for differentiating PD, MSA, PSP, and controls. Materials and Methods A total of 106 subjects (36 controls, 35 PD, 16 MSA, and 19 PSP) were included. DTI and R2* measures from striatal, midbrain, limbic, and cerebellar regions were obtained and compared between groups. The discrimination performance of DTI and R2* among groups was assessed using elastic-net machine learning and receiver operating characteristic analysis. Results Compared to controls, PD patients showed significant R2* differences in the red nucleus. Compared to PD, both MSA and PSP patients showed more widespread changes, extending from the midbrain to striatal and cerebellar structures. The pattern of changes, however, was different between the two groups. For instance, MSA patients showed decreased FA and increased R2* in the subthalamic nucleus, whereas PSP patients showed increased MD in the hippocampus. Combined DTI and R2* were significantly better than DTI or R2* alone in separating controls from PD/MSA-P/PSP, controls from PD, PD from MSA-P/PSP, and PD from MSA-P, but not PD from PSP or MSA-P from PSP. Conclusion DTI and R2* provide different but complementary information for different parkinsonisms. Combined DTI and R2* may be a superior marker for PD differential diagnosis.

show abstract

Section: Discussionmentioning

confidence: 99%

Combined Diffusion Tensor Imaging and Apparent Transverse Relaxation Rate Differentiate Parkinson Disease and Atypical Parkinsonism

Du¹,

Lewis²,

Kanekar³

et al. 2017

AJNR Am J Neuroradiol

View full text Add to dashboard Cite

show abstract

“…These fully automated methods use support vector machine (SVM) classification and other machine-learning method-derived classification algorithms for quantitative MRI analysis including volumetric datasets (Huppertz et al 2016; Scherfler et al 2016), neuromelanin-sensitive MRI (NM-MRI) (Castellanos et al 2015) and resting-state functional MRI (rs-fMRI) (Chen et al 2015). …”

Section: Techniquesmentioning

confidence: 99%

Magnetic resonance imaging for the diagnosis of Parkinson’s disease

et al. 2017

Self Cite

View full text Add to dashboard Cite

The differential diagnosis of parkinsonian syndromes is considered one of the most challenging in neurology and error rates in the clinical diagnosis can be high even at specialized centres. Despite several limitations, magnetic resonance imaging (MRI) has undoubtedly enhanced the diagnostic accuracy in the differential diagnosis of neurodegenerative parkinsonism over the last three decades. This review aims to summarize research findings regarding the value of the different MRI techniques, including advanced sequences at high- and ultra-high-field MRI and modern image analysis algorithms, in the diagnostic work-up of Parkinson’s disease. This includes not only the exclusion of alternative diagnoses for Parkinson’s disease such as symptomatic parkinsonism and atypical parkinsonism, but also the diagnosis of early, new onset, and even prodromal Parkinson’s disease.

show abstract

“…Kale et al (2013) employed feedforward neural network (FNN). Scherfler et al (2016) used decision tree (DT) as the classifier. Vasta et al (2016) utilized naive Bayesian classifier (NBC).…”

Section: Resultsmentioning

confidence: 99%

“…Vasta et al (2016) utilized naive Bayesian classifier (NBC). In this study, we compared the proposed DAG-SVM with FNN (Kale et al, 2013), DT (Scherfler et al, 2016), and NBC (Vasta et al, 2016). The features were the same as three-level WEs, the statistical analysis is all set to 10 × 10-fold cross validation, and the optimal parameters of classifiers were obtained by grid searching.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Wavelet Entropy and Directed Acyclic Graph Support Vector Machine for Detection of Patients with Unilateral Hearing Loss in MRI Scanning

Wang

Yang

et al. 2016

Front. Comput. Neurosci.

View full text Add to dashboard Cite

Highlights We develop computer-aided diagnosis system for unilateral hearing loss detection in structural magnetic resonance imaging.Wavelet entropy is introduced to extract image global features from brain images. Directed acyclic graph is employed to endow support vector machine an ability to handle multi-class problems.The developed computer-aided diagnosis system achieves an overall accuracy of 95.1% for this three-class problem of differentiating left-sided and right-sided hearing loss from healthy controls. Aim: Sensorineural hearing loss (SNHL) is correlated to many neurodegenerative disease. Now more and more computer vision based methods are using to detect it in an automatic way.Materials: We have in total 49 subjects, scanned by 3.0T MRI (Siemens Medical Solutions, Erlangen, Germany). The subjects contain 14 patients with right-sided hearing loss (RHL), 15 patients with left-sided hearing loss (LHL), and 20 healthy controls (HC).Method: We treat this as a three-class classification problem: RHL, LHL, and HC. Wavelet entropy (WE) was selected from the magnetic resonance images of each subjects, and then submitted to a directed acyclic graph support vector machine (DAG-SVM).Results: The 10 repetition results of 10-fold cross validation shows 3-level decomposition will yield an overall accuracy of 95.10% for this three-class classification problem, higher than feedforward neural network, decision tree, and naive Bayesian classifier.Conclusions: This computer-aided diagnosis system is promising. We hope this study can attract more computer vision method for detecting hearing loss.

show abstract

Diagnostic potential of automated subcortical volume segmentation in atypical parkinsonism

Abstract: This study provides Class III evidence that automated MRI analysis accurately discriminates among early-stage PD, MSA, and PSP.

Cited by 92 publications

References 31 publications

Combined Diffusion Tensor Imaging and Apparent Transverse Relaxation Rate Differentiate Parkinson Disease and Atypical Parkinsonism

Combined Diffusion Tensor Imaging and Apparent Transverse Relaxation Rate Differentiate Parkinson Disease and Atypical Parkinsonism

Magnetic resonance imaging for the diagnosis of Parkinson’s disease

Wavelet Entropy and Directed Acyclic Graph Support Vector Machine for Detection of Patients with Unilateral Hearing Loss in MRI Scanning

Contact Info

Product

Resources

About