A Visual Rating Scale for the Hummingbird Sign with Adjustable Diagnostic Validity

Comparative validation of AI and non-AI methods in MRI volumetry to diagnose Parkinsonian syndromes

Automated segmentation and volumetry of brain magnetic resonance imaging (MRI) scans are essential for the diagnosis of Parkinson's disease (PD) and Parkinson's plus syndromes (P-plus). To enhance the diagnostic performance, we adopt deep learning (DL) models in brain MRI segmentation and compared their performance with the gold-standard non-DL method. We collected brain MRI scans of healthy controls ([Formula: see text]) and patients with PD ([Formula: see text]), multiple systemic atrophy ([Formula: see text]), and progressive supranuclear palsy ([Formula: see text]) at Samsung Medical Center from January 2017 to December 2020. Using the gold-standard non-DL model, FreeSurfer (FS), we segmented six brain structures: midbrain, pons, caudate, putamen, pallidum, and third ventricle, and considered them as annotated data for DL models, the representative convolutional neural network (CNN) and vision transformer (ViT)-based models. Dice scores and the area under the curve (AUC) for differentiating normal, PD, and P-plus cases were calculated to determine the measure to which FS performance can be reproduced as-is while increasing speed by the DL approaches. The segmentation times of CNN and ViT for the six brain structures per patient were 51.26 ± 2.50 and 1101.82 ± 22.31 s, respectively, being 14 to 300 times faster than FS (15,735 ± 1.07 s). Dice scores of both DL models were sufficiently high (> 0.85) so their AUCs for disease classification were not inferior to that of FS. For classification of normal vs. P-plus and PD vs. P-plus (except multiple systemic atrophy - Parkinsonian type) based on all brain parts, the DL models and FS showed AUCs above 0.8, demonstrating the clinical value of DL models in addition to FS. DL significantly reduces the analysis time without compromising the performance of brain segmentation and differential diagnosis. Our findings may contribute to the adoption of DL brain MRI segmentation in clinical settings and advance brain research.

Midbrain area and the hummingbird sign from brain MRI in progressive supranuclear palsy and idiopathic normal pressure hydrocephalus

BACKGROUND AND PURPOSE

The main radiological finding in progressive supranuclear palsy (PSP) is reduced midbrain volume. Both qualitative (e.g., hummingbird sign) and quantitative (e.g., area measurements) markers have been noted. Recent studies have shown a similar reduction also in idiopathic normal pressure hydrocephalus (iNPH). The purpose was to investigate the reliability and accuracy of these markers in discriminating PSP from iNPH and controls.

METHODS

Eight neuroradiologists viewed sagittal MR images of the midbrain from 104 subjects: 26 PSP patients, 40 iNPH patients, and 38 healthy controls. They visually assessed whether the hummingbird sign was present or not, grading their confidence from 1 to 5. Assessments were translated into a score between +5 and -5: from maximum confidence of presence to maximum confidence of absence. A positive median score was considered to indicate hummingbird sign. Sagittal midbrain area was manually measured in each subject.

RESULTS

Seventy-seven percent of PSP patients, 65% of iNPH, and 3% of controls were visually assessed as having the hummingbird sign. Manually measured midbrain area also showed overlap between PSP and iNPH. Regarding discrimination of PSP patients, midbrain area measurements, using a cutoff of 90 mm² , yielded a higher area under the curve (AUC = 0.86) than visual assessment scores (AUC = 0.83), and higher reliability.

CONCLUSIONS

Measuring sagittal midbrain area is more accurate and reliable than visual assessment. Due to significant overlap in appearance, a midbrain with a hummingbird sign or reduced sagittal area should raise the suspicion of PSP only after other signs of iNPH have been considered.

Conventional Magnetic Resonance Imaging in the Diagnosis of Parkinsonian Disorders: A Meta-Analysis

Background

Numerous conventional magnetic resonance imaging (cMRI) parameters were previously found to differentiate parkinsonian disorders with statistical significance, but effect size has not been considered.

Objectives

To quantify effect size of previously identified cMRI parameters that differentiated parkinsonian disorders with statistical significance.

Method

A PubMed search limited to studies assessing cMRI parameters in at least 2 of Parkinson's disease, progressive supranuclear palsy, multiple system atrophy, and corticobasal degeneration/syndrome were selected. Either Cohen's d or positive and negative likelihood (LR+/-) as well as diagnostic odds ratios (DORs) were calculated as appropriate. cMRI parameter was considered useful if Cohen's d > 1.94 (<20% overlap) or if LR+ > 10, LR- < 0.1, or DOR > 20.

Results

Literature search identified 8848 publications and 36 were included for analysis. Putaminal (Cohen's d 2.07; DOR 23-infinity), pontine (DOR 32-infinity), and middle cerebellar peduncle (Cohen's d 2.24; DOR infinity) abnormalities were most useful in differentiating multiple system atrophy while reduced midbrain (Cohen's d 2.33-8.69; DOR infinity) and superior cerebellar peduncle (Cohen's d 2.47; DOR 51-infinity) diameters separated progressive supranuclear palsy. Corticobasal degeneration/syndrome does not have any distinguishing cMRI features, but reduced midbrain diameter may help differentiate corticobasal degeneration/syndrome from Parkinson's disease (DOR infinity). When LR- was calculated, all of these features carried a value of <0.1.

Conclusion

A number of cMRI features consistently demonstrated large effect size in separating parkinsonian disorders. However, it is the presence and not absence of these cMRI features that is most useful in patients with low to moderate pretest probability.

Multiparametric MR imaging of Parkinsonisms at 3 tesla: Its role in the differentiation of idiopathic Parkinson's disease versus atypical Parkinsonian disorders

OBJECTIVES

The Nigrosome-1 and putaminal hypointensity depicted on susceptibility-weighted imaging (SWI), and midbrain atrophy assessed on T1-weighted are some of the most common radiological parameters to diagnose Parkinsonism at Magnetic Resonance (MR) imaging. Our aim is to assess the feasibility of these signs in the differentiation of Idiopathic Parkinson's disease (IPD) patients versus disease (DC) and healthy controls (HC) and in the assessment of the Atypical Progressive Parkinsonisms (APPs).

METHODS

Presence or loss of the Nigrosome-1 was assessed retrospectively on multiple-echo SWI obtained on a 3 T scan by two neuroradiologists. Results were compared with the 123I-FP-CIT SPECT images. Morphologic diagnostic features suggestive of APPs such as midbrain atrophy and putaminal hypointensity were evaluated by qualitative scores. The midbrain and putaminal scores were summed (combined score) and then added to the Nigrosome-1 score (global score).

RESULTS

The study included 126 patients with IPD (n = 56), APPs patients (n = 30; 18 PSP, 3 MSA-C, 9 MSA-P), 16 DC and 24 HC. Sensitivity and specificity of the Nigrosome-1 in discriminating IPD from controls were 96,43% and 85.00%, APPs from controls were 100% and 85%, IPD from APPs were 96,43% and 0% respectively. Combined score for midbrain atrophy and putaminal hypointensity resulted in the most accurate for distinguishing APPs from IPD with a value of ≥ 2 (AUC = 0.98).

CONCLUSION

Nigrosome-1 is a valid tool to differentiate IPD-APPs from controls. The combined score of midbrain atrophy and putaminal hypointensity represents a valid diagnostic pointer in the differential diagnosis of APPs from IPD.

Neuroimaging in Dementia

Although the diagnosis of dementia still is primarily based on clinical criteria, neuroimaging is playing an increasingly important role. This is in large part due to advances in techniques that can assist with discriminating between different syndromes. Magnetic resonance imaging remains at the core of differential diagnosis, with specific patterns of cortical and subcortical changes having diagnostic significance. Recent developments in molecular PET imaging techniques have opened the door for not only antemortem but early, even preclinical, diagnosis of underlying pathology. This is vital, as treatment trials are underway for pharmacological agents with specific molecular targets, and numerous failed trials suggest that earlier treatment is needed. This article provides an overview of classic neuroimaging findings as well as new and cutting-edge research techniques that assist with clinical diagnosis of a range of dementia syndromes, with an emphasis on studies using pathologically proven cases.

Radiological biomarkers for diagnosis in PSP: Where are we and where do we need to be?

PSP is a pathologically defined neurodegenerative tauopathy with a variety of clinical presentations including typical Richardson's syndrome and other variant PSP syndromes. A large body of neuroimaging research has been conducted over the past two decades, with many studies proposing different structural MRI and molecular PET/SPECT biomarkers for PSP. These include measures of brainstem, cortical and striatal atrophy, diffusion weighted and diffusion tensor imaging abnormalities, [18F] fluorodeoxyglucose PET hypometabolism, reductions in striatal dopamine imaging and, most recently, PET imaging with ligands that bind to tau. Our aim was to critically evaluate the degree to which structural and molecular neuroimaging metrics fulfill criteria for diagnostic biomarkers of PSP. We queried the PubMed, Cochrane, Medline, and PSYCInfo databases for original research articles published in English over the past 20 years using postmortem diagnosis or the NINDS-SPSP criteria as the diagnostic standard from 1996 to 2016. We define a five-level theoretical construct for the utility of neuroimaging biomarkers in PSP, with level 1 representing group-level findings, level 2 representing biomarkers with demonstrable individual-level diagnostic utility, level 3 representing biomarkers for early disease, level 4 representing surrogate biomarkers of PSP pathology, and level 5 representing definitive PSP biomarkers of PSP pathology. We discuss the degree to which each of the currently available biomarkers fit into this theoretical construct, consider the role of biomarkers in the diagnosis of Richardson's syndrome, variant PSP syndromes and autopsy confirmed PSP, and emphasize current shortfalls in the field. © 2017 The Authors. Movement Disorders published by Wiley Periodicals, Inc. on behalf of International Parkinson and Movement Disorder Society.

Frontotemporal Dementia

Frontotemporal dementia (FTD) is a heterogeneous disorder with distinct clinical phenotypes associated with multiple neuropathologic entities. Presently, the term FTD encompasses clinical disorders that include changes in behavior, language, executive control, and often motor symptoms. The core FTD spectrum disorders include behavioral variant FTD, nonfluent/agrammatic variant primary progressive aphasia, and semantic variant PPA. Related FTD disorders include frontotemporal dementia with motor neuron disease, progressive supranuclear palsy syndrome, and corticobasal syndrome. In this article, the authors discuss the clinical presentation, diagnostic criteria, neuropathology, genetics, and treatments of these disorders.

A Clinical Guide to Frontotemporal Dementias

The term frontotemporal dementia (FTD) describes a diverse group of clinical syndromes, including behavioral-variant FTD (bvFTD), nonfluent/agrammatic-variant primary progressive aphasia (nfvPPA), semantic-variant primary progressive aphasia (svPPA), FTD motor neuron disease (FTD-MND), progressive supranuclear palsy syndrome (PSP-S), and corticobasal syndrome (CBS). Although each of these syndromes may be distinguished by their respective disturbances in behavior, language, or motor function and characteristic imaging findings, they may present a diagnostic dilemma when encountered clinically. In this article, we review the clinical features, diagnostic criteria, pathology, genetics, and therapeutic interventions for FTD spectrum disorders.