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

ABSTRACT

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.

Neuropathology remains the “gold standard” in the diagnosis of degenerative parkinsonian disorders because of the lack of reliable biomarkers. Clinically differentiating these disorders remains challenging even to the movement disorders specialists1, 2 and it is common for patients to present with a vast array of symptoms to a diverse range of disciplines. ³ So much so that >60% of patients with atypical parkinsonism will have their clinical diagnosis revised during their subsequent disease course. ¹ Recent advances of imaging technologies including diffusion imaging, automated machine‐learning techniques to identify diagnostic patterns, dopamine imaging, and tau positron emission tomography imaging have provided optimism that a useful noninvasive biomarker will be imminently available. However, many of these techniques are not readily accessible to the clinician in day‐to‐day practice apart from dopamine transporter imaging, which is now available in many countries. Aside from clinical assessment, conventional magnetic resonance imaging (cMRI) is still one of the most powerful diagnostic tools in neurological practice. A number of radiological signs have been identified as useful in differentiating parkinsonian disorders. Some of the better recognized signs include the hummingbird sign (flat or concave midbrain tegmentum [beak] with preserved pontine volume [body] in the sagittal plane forming the silhouette of the head of a hummingbird or king penguin) and the morning glory flower sign (reduced anteroposterior midbrain diameter with concavity of the lateral margin of the midbrain tegmentum in the axial plane resembling a lateral view of the morning glory flower) supporting a diagnosis of progressive supranuclear palsy (PSP) and the hot cross bun sign (cruciform configuration of hyperintensity in the pons attributed to degeneration of transverse pontocerebellar fibers) supporting a diagnosis of multiple system atrophy (MSA). ⁴ These signs, in addition to a number of other cMRI structural parameters, have been put forward as being potentially useful in differentiating between parkinsonian disorders. However, studies had largely drawn conclusions based on tests of statistical significance without consideration of effect size. A useful discriminator applied in small samples may not necessarily reach statistical significance, whereas a marker demonstrating a small difference between large samples could be deemed significant but would be of little utility in individuals because of extensive overlap between groups. ⁵ This meta‐analysis is a pragmatic study designed to determine which cMRI markers are most useful for the diagnosis of parkinsonian disorders in day‐to‐day clinical practice that the clinician can simply apply through standard imaging viewing platforms. To do so, the effect size of cMRI markers previously found to separate parkinsonian disorders with statistical significance was quantified.

Methods

A PubMed search using combinations of terms outlined in Table 1 was performed, restricted to human studies in English, and published between January 2000 and March 2020. This search resulted in a total of 8848 publications. Study abstracts were screened, and those relevant to this analysis were reviewed in detail to determine suitability for inclusion. Studies were only included if direct comparison was being made between 2 or more parkinsonian disorders (idiopathic Parkinson's disease [PD], PSP, MSA, and corticobasal syndrome [CBS]) and standard routine structural magnetic resonance imaging (MRI) was being examined. Studies were excluded where comparisons were being made between 1 parkinsonian disorder and healthy controls as control groups may not necessarily be comparable across studies. Where specified, patients with MSA cerebellar subtype were also excluded. Another exclusion criterion was the use of MRI techniques requiring additional specialized sequences or software beyond what is expected to be part of a conventional, routine study. Therefore, studies examining the following techniques were excluded: magnetic field strength greater than 3 Tesla (T), volumetry, diffusion tensor imaging, morphometry (cross‐sectional area measurement), and quantitative diffusion imaging.

TABLE 1.

PubMed search criteria

	Parkinson OR Parkinson's OR parkinsonism
OR	progressive supranuclear OR Steele‐Richardson OR Richardson's OR PSP
OR	multiple system atrophy OR MSA
OR	corticobasal OR cortico‐basal OR CBD OR CBS
AND	diagnosis OR sensitivity OR specificity
AND	MRI OR magnetic resonance OR radiology OR radiological

Effect size was measured by Cohen's d for publications reporting means and standard deviations (SDs). Cohen's d measures the number of pooled standard deviations by which 2 distributions are separated, hence estimating the degree of overlap between 2 groups independent of sample size. Effect size is generally considered large when Cohen's d is greater than 0.8. ⁵ For cMRI parameters tested in more than 1 population of patients, weighted mean Cohen's d was calculated using the Hunter and Schmidt method. ⁶ For the purpose of this analysis, cMRI parameter was considered useful if Cohen's d was greater than 1.94 (<20% overlap).

For comparative studies expressing results as sensitivity and specificity, positive and negative likelihood (LR+/−) as well as diagnostic odds ratios (DORs) were calculated. LR is the ratio of the probability of attaining a specific test result in patients with the disease to the probability of attaining the same result in healthy individuals. LR is less dependent on disease prevalence than sensitivities or specificities and is of maximum impact with moderate (rather than very high or very low) pretest probabilities. Provided that the prior probability of a condition is intermediate (between 10% and 90%), LR+ of 2, 5, and 10 increase percentage probability of disease by approximately 15%, 30%, and 45%, respectively. ⁷ The DOR summarizes the effectiveness of a diagnostic test with dichotomous outcomes and is the ratio of LR+ to LR−. ⁸ Lack of available data prohibited the calculation of weighted mean when the effect size was measured by DOR. For the purpose of this study, cMRI parameter was considered useful if at least 1 of the following was satisfied: LR+ > 10, LR− < 0.1, or DOR > 20. ⁹

Results

A total of 36 publications satisfied criteria for inclusion ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ (Table S1). Most studies were either retrospective or cross‐sectional in design with the gold standard of diagnosis being clinical according to consensus criteria. ⁴⁶ , ⁴⁷ , ⁴⁸ Only 4 studies included neuropathological confirmation. Of the 25 publications examining MSA, 15 distinguished parkinsonism subtype specifically. Two of 29 publications examining PSP specified parkinsonism subtype as a patient group in addition to classic Richardson syndrome. The term CBS was used unless neuropathological confirmation of corticobasal degeneration was available. A total of 22 publications reported on 0.5 to 1.5 T MRI, whereas 13 used 3 T MRI.

A total of 867 patients with PSP, 629 with MSA, 2117 with PD, and 37 with CBS were examined in selected publications. The mean ages (years) of patients at MRI for PD, MSA, PSP, and CBS were 67 (53.6–72.5), 63.2 (55.9–67.8), 69.5 (62.5–76.8), and 67.9 (63–71), respectively. The mean disease durations (years) at assessment for PD, MSA, PSP, and CBS were 5.5 (2–10.7), 3.4 (1.7–5.6), 3.6 (1.5–7.3), and 3.6 (3.1–5.1), respectively. Data regarding disease severity and medication use were largely incomplete.

Effect sizes of cMRI parameters that differentiated parkinsonian disorders with statistical significance from selected studies are summarized in Table 2 and Table S2.

TABLE 2.

Weighted mean Cohen's d of cMRI parameters examined in more than 1 population of patients

Test Groups	MRI Parameter	Cohen's d (95% Confidence Interval)
PSP vs. MSA	MCP diameter, sagittal− a	0.87 (0.02–1.72)
	SCP diameter, oblique coronal− b	1.55 (0.59–2.51)
	MCPsag/SCPcor ratio+ b	2.42 (1.4–3.4)
	Midbrain diameter, axial− b	4.98 (0.12–9.84)
	Midbrain diameter, sagittal− b	2.33 (2.12–2.53)
	Pontine diameter, sagittal− b	0.94 (0.73–1.18)
	Midbrain/pontine diameter ratio, sagittal− b	2.1 (1.82–2.83)
	Cerebral interpeduncular angle+ b	1.39 (0.33–2.44)
PD vs. MSA	Pontine diameter, sagittal− a	1.12 (0.89–1.36)
	MCP diameter, sagittal− a	2.24 (1.9–2.58)
	MCP diameter, axial− a	1.88 (1.2–2.55)
	SCP diameter, oblique coronal− a	1.96 (0.33–3.59)
	Midbrain diameter, sagittal− a	1.52 (0.21–2.83)
	Midbrain/pontine diameter ratio, sagittal+ a	1.05 (0.42–1.69)
PSP vs. PD	Third ventricular/frontal horn width+ b	2.39 (1.98–2.81)
	Frontal horn width, axial+ b	0.41 (0–0.83)
	Third ventricular width, axial+ b	2.09 (1.62–2.56)
	Cerebral interpeduncular angle+ b	1.21 (0.41–2.01)
	MCPsag/SCPcor ratio+ b	1.46 (0.59–2.34)
	Pontine diameter, sagittal− b	1.18 (0.35–2.01)
	Midbrain/pontine diameter ratio, sagittal− b	2.42 (2.16–2.67)
	MCP diameter, sagittal− b	1.67 (1.16–2.19)
	SCP diameter, oblique coronal− b	2.47 (1.79–3.15)
	Midbrain diameter, axial− b	8.69 (−1.47–18.8)
	Midbrain diameter, sagittal− b	3.13 (2.67–3.59)

^aFavors PSP.

^bFavors MSA.

A reduction of cMRI parameter is indicated by −, whereas an increase of cMRI parameter is indicated by +.

Abbreviations: cMRI, conventional magnetic resonance imaging; MRI, magnetic resonance imaging; PSP, progressive supranuclear palsy; MSA, multiple system atrophy; MCP, middle cerebellar peduncle; SCP, superior cerebellar peduncle; MCP_sag, MCP sagittal diameter; SCP_cor, SCP oblique coronal diameter; PD, Parkinson's disease; CBS, corticobasal syndrome. MRI parameters with italicised Cohen's d reached the pre‐defined threshold of clinical usefulness.

MSA Versus PD

LR and DOR were calculated for 15 cMRI parameters, whereas Cohen's d was calculated for 12 (Table S2) with the weighted mean Cohen's d calculated for 6 (Table 2). cMRI parameters that were identified as useful in differentiating MSA from PD in more than 1 population of patients included reduced MCP sagittal diameter (MCP_sag) (weighted mean Cohen's d 2.24; DOR infinity ¹⁵ , ²⁶ , ³² , ³⁹ ), reduced superior cerebellar peduncle oblique coronal diameter (SCP_cor) (weight mean Cohen's d 1.96 ¹⁵ , ²⁶ ); hot cross bun sign (LR+ and DOR infinity ¹³ , ²⁹ , ³² , ³⁷ ), putaminal slit‐like hyperintensity (DOR 23–30 ²⁴ , ³² , ³⁷ ), putaminal T2 hypointensity (DOR 23–infinity ¹³ , ³² , ³⁷ , ⁴² ), putaminal atrophy (Cohen's d 2.09; DOR 32–infinity ¹³ , ¹⁸ , ²⁴ , ³⁷ ), MCP hyperintensity (LR+ and DOR infinity ¹³ , ²⁴ , ²⁹ , ³⁷ ), pontine atrophy (LR+ and DOR infinity ¹³ , ³⁷ ), and dilatation of fourth ventricle (DOR 22 and infinity ²⁴ , ²⁹ ). Other potentially useful parameters in differentiating MSA from PD examined in a single population of patients included reduced midbrain axial diameter (Cohen's d 4.42 ⁴⁵ ), midbrain and medullary atrophy,¹³ and increased score on a composite scale quantifying putaminal characteristics and pontocerebellar structures ⁴⁴ (DOR infinity).

PSP Versus PD

LR and DOR were calculated for 14 cMRI parameters, whereas Cohen's d was calculated for 21 (Table S2) with the weighted mean Cohen's d calculated for 11 (Table 2). cMRI parameters that were identified as useful in differentiating PSP from PD in more than 1 population of patients included increased third ventricle/frontal horn width (weighted mean Cohen's d 2.39 ³⁶ ), widened third ventricular axial width (weighted mean Cohen's d 2.09; DOR infinity ¹² , ³⁶ ), reduced midbrain/pontine sagittal diameter ratio (weighted mean Cohen's d 2.42 ¹⁴ , ²⁶ , ²⁷ , ²⁸ , ³⁵ ), reduced SCP_cor (weighted mean Cohen's d 2.47; DOR infinity ¹⁴ , ¹⁵ , ²⁰ , ²⁵ , ²⁶ , ²⁷ , ²⁹ , ³⁰ , ³³ , ³⁴ , ³⁷ , ³⁹ ), reduced sagittal midbrain diameter (weighted mean Cohen's d 3.13 ¹² , ¹⁴ , ¹⁵ , ¹⁹ , ²⁶ , ²⁷ , ²⁸ , ³⁵ , ³⁹ , ⁴³ ), reduced midbrain axial diameter (weighted mean Cohen's d 8.69 ¹⁴ , ⁴⁵ ), hummingbird sign (DOR 177.56 and infinity ²⁹ , ³¹ ), and morning glory flower sign (DOR infinity ¹⁰ , ²⁹ ). The product and ratio of midbrain, pontine, MCP_sag, and SCP_cor diameters in various combinations (Cohen's d 2–6 ¹⁴ ) along with increased interpeduncular fossa/midbrain sagittal length (Cohen's d 2.72 ¹⁹ ), reduced quadrigeminal plate thickness (Cohen's d 3.07 ¹² ), and dilatation of fourth ventricle (DOR infinity ²⁹ ) and a higher score on a composite scale examining the midbrain, cortical, putaminal, and pontocerebellar structures (DOR 23.58–infinity ⁴⁴ ) also yielded large effect sizes in differentiating PSP from PD in a single population of patients.

PSP Versus MSA

LR and DOR were calculated for 28 cMRI parameters, whereas Cohen's d was calculated for 20 (Table S2) with the weighted mean Cohen's d calculated for 8 (Table 2). cMRI parameters that were identified as useful in differentiating PSP from MSA in more than 1 population of patients included reduced midbrain sagittal diameter (weighted mean Cohen's d 2.42 ¹¹ , ¹⁴ , ¹⁵ , ²⁶ , ²⁷ , ²⁸ , ³⁵ , ³⁹ , ⁴³ ), reduced midbrain axial diameter (weighted mean Cohen's d 4.98 ¹⁴ , ⁴⁵ ), increased MCP_sag/SCP_cor ratio (weighted mean Cohen's d 2.42; DOR 127.7 ¹⁴ , ¹⁵ , ³⁷ ), reduced midbrain/pontine sagittal diameter ratio (weighted mean Cohen's d 2.1; DOR infinity ¹⁴ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁵ ), hummingbird sign (DOR infinity ¹⁶ , ²⁹ , ³¹ ), and morning glory flower sign (DOR 51 ¹⁰ and infinity ²⁹ ). The product and ratio of midbrain, pontine, MCP_sag, and SCP_cor diameters in various combinations (Cohen's d 2.99–4.26 ¹⁴ ) and a higher score on a composite scale quantifying frontoparietal atrophy (LR+, 11 ⁴⁴ ) also differentiated PSP with a large effect size in a single population of patients. cMRI features favoring MSA in more than 1 population of patients included the hot cross bun sign (DOR 32 ⁴¹ and infinity ²⁹ ) and putaminal T2 hyperintensity (Cohen's d 2.07; DOR infinity ¹⁸ , ⁴¹ ). Other features favoring MSA tested in a single population of patients included MCP hyperintensity (DOR infinity ²⁹ ), cerebellar and inferior olivary hyperintensity (DOR infinity ⁴¹ ), and dentate atrophy (DOR 29.6 ⁴¹ ).

CBS Versus PD, MSA, and PSP

LR and DOR were calculated for 8 cMRI parameters, whereas Cohen's d was calculated for 2 (Table S2). No parameter was tested in more than 1 population of patients. The hummingbird sign was found to be useful in differentiating PSP from CBS (DOR infinity ²⁹ ). The hot cross bun sign and MCP hyperintensity were useful in differentiating MSA from CBS (DOR Infinity ²⁹ ). Conversely, reduced midbrain sagittal diameter differentiated CBS from MSA (Cohen's d 2.04 ³⁹ ). A higher score on a composite scale quantifying midbrain (DOR infinity ⁴⁴ ) as well as cortical, putaminal, and pontocerebellar structures (DOR 113 ⁴⁴ ) favored CBS over PD.

Discussion

During the past 2 decades, numerous publications identified a large number of cMRI parameters that differentiated parkinsonian disorders with statistical significance. However, only a small proportion, which included the well‐described hummingbird, morning glory flower, and hot cross bun signs, was determined to be useful once effect size was considered. Putaminal, pontine, and MCP abnormalities were most useful in differentiating MSA from other parkinsonian disorders, whereas reduced midbrain and superior cerebellar peduncle diameters distinguished PSP from PD and MSA. Limited evidence suggests that CBS does not have any distinguishing cMRI features, but reduced midbrain diameter and frontoparietal atrophy may help separate CBS from PD (Table 3).

TABLE 3.

Summary of most useful cMRI parameters differentiating parkinsonian disorders

	PD	MSA	PSP	CBS
MSA	Putaminal abnormalities: slit‐like rim hyperintensity; T2 hypointensity, atrophy MCP abnormalities: hyperintensity and reduced sagittal diameter Pontine atrophy Fourth ventricle dilatation Hot cross bun sign		Putaminal T2 hyperintensity Hot cross bun sign Pontocerebellar abnormalities: MCP, inferior olivary, and cerebellar hyperintensity; dentate atrophy	Hot cross bun sign MCP hyperintensity
PSP	Midbrain abnormalities: reduced sagittal and axial diameter, reduced midbrain/pontine sagittal diameter ratio, hummingbird sign, morning glory flower sign Third and fourth ventricular dilatation Reduced SCP oblique coronal diameter Frontoparietal atrophy	Midbrain abnormalities: reduced sagittal and axial diameter, reduced midbrain/pontine sagittal diameter ratio, hummingbird sign, morning glory flower sign Increased MCPsag/SCPcor diameter ratio Frontoparietal atrophy		Hummingbird sign
CBS	Reduced midbrain sagittal diameter Frontoparietal atrophy	Reduced midbrain sagittal diameter

Abbreviations: cMRI, conventional magnetic resonance imaging; PD, Parkinson's disease; MSA, multiple system atrophy; PSP, progressive supranuclear palsy; CBS, corticobasal syndrome; MCP, middle cerebellar peduncle; SCP, superior cerebellar peduncle; MCP_sag, MCP sagittal diameter; SCP_cor, SCP oblique coronal diameter.

Findings from this analysis are consistent with the current consensus criteria, which include atrophy of the putamen, MCP, pons, or cerebellum as additional features to consider for a diagnosis of possible MSA. ⁴⁹ In contrary, radiological features had never featured as inclusion criteria in previous iterations of consensus diagnostic criteria for PSP. ⁴⁸ , ⁵⁰ Several studies on PSP included in this analysis examined midbrain relative to pontine diameter, but this was not found to be any more useful than considering midbrain diameter alone. Similarly, MCP_sag/SCP_cor diameter was actually less useful than SCP_cor diameter alone in differentiating PSP from PD. The same ratio, however, was more useful in differentiating MSA from PSP, probably because of MCP involvement in MSA. The sagittal midbrain diameter of normal controls ranged between 10.1 to 12.6 mm, ¹⁴ , ¹⁵ , ¹⁹ , ²⁷ , ²⁸ whereas a recent study found the normal mean axial midbrain diameter to be 17 mm. ⁵¹ A sagittal midbrain diameter of 9.35 mm or less has been shown to differentiate PSP from other parkinsonian disorders with a sensitivity of 83% and a specificity and positive predictive value (PPV) of 100%. ²⁸ Axial measurement appeared less useful with a diameter of less than 17 mm achieving a diagnostic specificity of 96% and a PPV of 80%, but only a sensitivity of 23%. ⁴¹ The diagnostic cutoff for SCP_cor diameter has not been determined. Healthy controls and patients with PD had measurements of >3.6 mm, whereas patients with MSA with parkinsonism subtype and patients with PSP had measurements of >3.4 mm and <3 mm, respectively. ¹⁴ , ¹⁵ , ²⁵ , ²⁷ , ²⁸ , ³⁰ , ³³ Therefore, it is reasonable to consider <9.35 mm as a cutoff for sagittal midbrain diameter and <3 mm as an indication of reduced SCP_cor diameter when contemplating a diagnosis of PSP.

A number of limitations must be taken into consideration when interpreting studies on cMRI in parkinsonian disorders. Selected publications included in this meta‐analysis spanned over 2 decades, and as such there was great heterogeneity in study designs, particularly with respect to the diagnostic criteria used and MRI technical specifications and interpretations. In the MSA group, although most patients were specified to have MSA with parkinsonism subtype, patients with cerebellar subtype may have been included in some studies. MRI magnetic field strength ranged between 0.5 to 3 T, with a wide range of slice thickness and gap. Furthermore, the interpretation of radiological features could be greatly affected by subjectivity even with well‐established signs because of the lack of consensus criteria and “normal” cut‐off values for structural measurements. For example, putaminal density relative to that of the globus pallidus is said to be dependent on the MRI magnetic field strength and is most reliable with 1.5 T scanners. The putamen can appear hyperintense at 0.5 T, whereas the hyperintense rim of the putamen can be seen in healthy individuals at 3 T. ⁴ These factors would undoubtedly affect the consistency of comparisons across studies and between parkinsonian disorders while contributing to the variability of determined usefulness of cMRI parameters.

The lack of neuropathological confirmation and the utilization of clinical criteria as the gold standard of diagnosis in the vast major of identified studies constitute another major pitfall. Although the PPV for a clinical diagnosis of PD is quite high (98.6%), particularly when made by a movement disorders specialist, PPV of a clinical diagnosis in atypical parkinsonism is markedly inferior (71%). ¹ The accuracy of a clinical diagnosis of corticobasal degeneration based on the presence of basal ganglia and cortical signs is even lower (25%), and considering the heterogeneity of neuropathologies in CBS, a large proportion is likely to have underlying PSP and Alzheimer's disease pathologies. ⁵² This may well account for the limited usefulness of any tests including cMRI in the diagnosis of CBS. As such, the utilization of clinical criteria as the gold standard of diagnosis is problematic, especially for publications examining atypical parkinsonian disorders with a cross‐sectional design. It is impossible to assess how much cMRI features contribute to the diagnosis of parkinsonian disorders when the gold standard of clinical diagnosis is applied contemporaneously as MRI assessment. The accuracy of clinical diagnosis increases with disease progression. ¹ Therefore, to satisfy the purpose of determining the predictive value of cMRI features in parkinsonian disorders, the gold standard clinical assignment should ideally only be made after a period of prospective follow‐up.

Major unmet needs exist in the diagnosis of neurodegenerative parkinsonian disorders as there is still no biomarker that is inexpensive, accessible, and reproducible but also accurate in reflecting not only the underlying distribution but also type of neuropathology. Although imaging techniques have evolved substantially during the past 2 decades, access to these novel modalities and methods remains mostly limited to the research setting and generally inaccessible in day‐to‐day clinical practice. Although there is a sizeable body of publications devoted to determining the usefulness of cMRI in differentiating parkinsonian disorders, it remains unclear how cMRI compares to clinical diagnosis and other novel imaging techniques and modalities as a potential biomarker. It is reasonable to conclude that cMRI is a relatively inexpensive, accessible and reproducible method of measuring the underlying distribution of neuropathology only and its utility is very limited in diseases with a broad spectrum of presentation beyond that of the classical form e.g. PSP, or in syndromes not specific to a single type of neuropathology e.g. CBS.

Despite the limitations outlined, a number of cMRI features consistently demonstrated large effect sizes in the differentiation of parkinsonian disorders. When calculation of LR− was possible, all of these features were found to carry a value >0.1, indicating that only the presence and not the absence of these features is useful in the differentiation of parkinsonian disorders. In patients with high pretest probability, the utility of cMRI is limited. However, the presence of certain cMRI features may be helpful in determining a diagnosis in patients with moderate pretest probability and narrow down the diagnostic possibilities in patients with low pretest probability. cMRI features should therefore be interpreted within the context of the individual's clinical scenario and should not be relied on as a major consideration in the diagnosis of parkinsonian disorders.

Disclosures

Ethical Compliance Statement

Patient consent was not required for this work as it is a meta‐analysis from existing literature. The approval of an institutional review board was not required. I confirm that I have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflict of Interest

This study is not industry sponsored. Dr. Lee reports no conflicts of interest.

Financial Disclosures for the Previous 12 Months

Dr. Lee is employed by Eastern Health. He received grants from the Eastern Health Research Foundation.

Supporting information

Table S1. Publications included for analysis.

Table S2. Effect sizes of conventional magnetic resonance imaging parameters from selected publications reported to differentiate parkinsonian disorders with statistical significance.