Optimized data preprocessing for multivariate analysis applied to 99mTc-ECD SPECT data sets of Alzheimer's patients and asymptomatic controls

Arterial spin labeling-MR may be an alternative to SPECT for evaluating cerebral perfusion in patients with unilateral middle cerebral artery stenosis

OBJECTIVE

Cerebral blood flow (CBF) mapping of single-photon emission tomography (SPECT) is considered a gold standard for evaluating cerebral perfusion. However, invasiveness, high costs and strict technical requirements can limit its clinical use. We aimed to evaluate the concordance of CBF maps obtained from SPECT and pseudo-continuous arterial spin labeling magnetic resonance (PCASL-MR) imaging for evaluating cerebral perfusion.

METHODS

PCASL-MR/SPECT-CBF maps were obtained from 16 eligible patients with unilateral middle cerebral artery stenosis (MCAS). Three slices (basal ganglia, semi-oval center and cerebellum) on both PCASL-MR and SPECT maps were divided into different regions of interest (ROIs) according to the ASPECT criterion, arterial territories, and cerebral hemispheres, respectively. The concordance of the two types of CBF maps and the specificity and sensitivity of PCASL-MR imaging on predicting regional hypoperfusion were calculated.

RESULTS

A total of 448 ROIs were divided according to the ASPECT criterion, 192 ROIs partitioned in accordance with arterial territories, and 96 ROIs delineated based on cerebral hemispheres were analyzed. PCASL-MR imaging exhibited 83.78% to 100% sensitivity, 90.19% to 95.83% specificity for detection of hypoperfusion. Qualitative analyses revealed a strong concordance between PCASL-MR and SPECT on reflecting regional cerebral hypoperfusion (Kappa coefficient = 0.662-0.920, p < 0.01). Semi-quantitative analysis by ΔCBF revealed moderate consistency (Spearman correlation coefficient = 0.610-0.571).

CONCLUSIONS

Our findings suggest that PCASL-MR may be a promising non-invasive, inexpensive alternative to SPECT for evaluating cerebral perfusion accurately in patients with symptomatic MCAS.

Clinical and neuroimaging correlates among cohorts of cerebral arteriostenosis, venostenosis and arterio-venous stenosis

The purpose of this study was to discriminate the clinical and imaging correlates of cerebral arterial stenosis (CAS), venous stenosis (CVS) and arterio-venous stenosis (CAVS) in the clinical setting. Patients were classified into three groups: CAS (n = 75), CVS (n=74) and CAVS (n=67). Focal neurological deficits were the prominent presenting symptoms in CAS group, while venous turbulence related symptoms were common in both CVS and CAVS group. Risk factor analysis showed the OR (95%CI) for diabetes, male gender and age in CAS vs. CVS group were 13.67(2.71, 68.85), 6.69(2.39, 18.67) and 1.07(1.03, 1.12) respectively. Male gender, diabetes and age in CAVS vs. CAS groups were 0.27(0.11, 0.63), 0.26(0.10, 0.67) and 1.09(1.04, 1.14) respectively, while age in CAVS vs. CVS group was 1.11(1.07, 1.15). The white matter lesions (WMLs) in CAS group varied in size, with clear boundaries asymmetrically distributed in bilateral hemispheres. CVS-induced WMLs revealed a bilaterally symmetric, cloudy-like appearance. The cerebral perfusion was asymmetrically reduced in CAS but symmetrically reduced in CVS group. The clinical characteristics and neuroimaging presentations were different among patients with CAS, CVS and CAVS. We recommended for aged patients, both arterial and venous imaging should be considered in diagnosis of cerebral stenotic vascular disorders.

Review of Medical Decision Support and Machine-Learning Methods

Machine-learning methods can assist with the medical decision-making processes at the both the clinical and diagnostic levels. In this article, we first review historical milestones and specific applications of computer-based medical decision support tools in both veterinary and human medicine. Next, we take a mechanistic look at 3 archetypal learning algorithms—naive Bayes, decision trees, and neural network—commonly used to power these medical decision support tools. Last, we focus our discussion on the data sets used to train these algorithms and examine methods for validation, data representation, transformation, and feature selection. From this review, the reader should gain some appreciation for how these decision support tools have and can be used in medicine along with insight on their inner workings.

γ-Aminobutyric acid type A receptor binding affinity in the right inferior frontal gyrus at resting state predicts the performance of healthy elderly people in the visual sustained attention test

ABSTRACTBackground:Although recent studies have suggested that the γ-aminobutyric acid type A (GABAA) receptor binding affinity can be a more sensitive marker of age-related neuronal loss than regional gray matter (GM) volume, knowledge about the relationship between decreased GABAA receptor binding affinity and cognitive decline during normal aging is still limited.

METHODS

Thirty-seven healthy elderly individuals (aged 50-77 years (mean, 64.5 ± 7.3 years); 15 males and 22 females) were enrolled in this study. We investigated the association of the performance of the healthy elderly in the attentional function test with regional GM volume, regional cerebral bold flow (rCBF), and GABAA receptor binding affinity in the resting state by structural magnetic resonance imaging (MRI), arterial spin labeling (ASL), and 123I-iomazenil (IMZ) SPECT, with the analysis focusing on the bilateral inferior frontal gyri.

RESULTS

The score of the rapid visual information processing (RVP) test, which is used to assess visual sustained attention, showed a positive correlation with GABAA receptor binding affinity in the right inferior frontal gyrus. No significant correlation was found between RVP test score and regional GM volume or rCBF.

CONCLUSION

The findings of 123I-IMZ SPECT, but not those of structural MRI or ASL, suggest that a decreased GABAA receptor binding affinity can be a sensitive marker of cognitive impairment.

Safety and efficacy of remote ischemic conditioning in pediatric moyamoya disease patients treated with revascularization therapy

BACKGROUND

Revascularization surgery has been the standard treatment to prevent ischemic stroke in pediatric Moyamoya disease (MMD) patients with ischemic symptoms. However, perioperative complications, such as hyperperfusion syndrome, new infarct on imaging, or ischemic stroke, are inevitable. Remote ischemic conditioning (RIC) is a noninvasive and easy-to-use neuroprotective strategy, and it has potential effects on preventing hyperperfusion syndrome and ischemic infarction.

AIMS

The aim of this study is to investigate the safety and efficacy of RIC in pediatric MMD patients undergoing revascularization surgery.

METHOD

A total of 60 pediatric MMD patients with one or more ischemic symptoms will be recruited and allocated in 1:1 ratio to the RIC group and sham group, respectively. Both RIC and sham RIC will be performed twice daily for 7 consecutive days before revascularization surgery with different cuff pressures during the ischemia period (50 mmHg over-systolic blood pressure and 30 mmHg). Single photon emission computed tomography will be performed within 7 days preoperatively and 3 months postoperatively, respectively, to evaluate the cerebral perfusion status. Other outcomes, including safety, plasma biomarker, functional outcome, and the incidence of infarction and its size, will also be evaluated.

CONCLUSION

This study will provide insights into the preliminary proof of principle, safety, and efficacy of RIC in pediatric MMD patients undergoing revascularization surgery therapy, and this data will provide parameters for future larger scale clinical trials if efficacious.

CT-Based Attenuation Correction in Brain SPECT/CT Can Improve the Lesion Detectability of Voxel-Based Statistical Analyses

Background

Integrated SPECT/CT enables non-uniform attenuation correction (AC) using built-in CT instead of the conventional uniform AC. The effect of CT-based AC on voxel-based statistical analyses of brain SPECT findings has not yet been clarified. Here, we assessed differences in the detectability of regional cerebral blood flow (CBF) reduction using SPECT voxel-based statistical analyses based on the two types of AC methods.

Subjects and Methods

N-isopropyl-p-[123I]iodoamphetamine (IMP) CBF SPECT images were acquired for all the subjects and were reconstructed using 3D-OSEM with two different AC methods: Chang’s method (Chang’s AC) and the CT-based AC method. A normal database was constructed for the analysis using SPECT findings obtained for 25 healthy normal volunteers. Voxel-based Z-statistics were also calculated for SPECT findings obtained for 15 patients with chronic cerebral infarctions and 10 normal subjects. We assumed that an analysis with a higher specificity would likely produce a lower mean absolute Z-score for normal brain tissue, and a more sensitive voxel-based statistical analysis would likely produce a higher absolute Z-score for in old infarct lesions, where the CBF was severely decreased.

Results

The inter-subject variation in the voxel values in the normal database was lower using CT-based AC, compared with Chang’s AC, for most of the brain regions. The absolute Z-score indicating a SPECT count reduction in infarct lesions was also significantly higher in the images reconstructed using CT-based AC, compared with Chang’s AC (P = 0.003). The mean absolute value of the Z-score in the 10 intact brains was significantly lower in the images reconstructed using CT-based AC than in those reconstructed using Chang’s AC (P = 0.005).

Conclusions

Non-uniform CT-based AC by integrated SPECT/CT significantly improved sensitivity and the specificity of the voxel-based statistical analyses for regional SPECT count reductions, compared with conventional uniform Chang's AC.

MALDI imaging MS reveals candidate lipid markers of polycystic kidney disease

Autosomal recessive polycystic kidney disease (ARPKD) is a severe, monogenetically inherited kidney and liver disease. PCK rats carrying the orthologous mutant gene serve as a model of human disease, and alterations in lipid profiles in PCK rats suggest that defined subsets of lipids may be useful as molecular disease markers. Whereas MALDI protein imaging mass spectrometry (IMS) has become a promising tool for disease classification, widely applicable workflows that link MALDI lipid imaging and identification as well as structural characterization of candidate disease-classifying marker lipids are lacking. Here, we combine selective MALDI imaging of sulfated kidney lipids and Fisher discriminant analysis (FDA) of imaging data sets for identification of candidate markers of progressive disease in PCK rats. Our study highlights strong increases in lower mass lipids as main classifiers of cystic disease. Structure determination by high-resolution mass spectrometry identifies these altered lipids as taurine-conjugated bile acids. These sulfated lipids are selectively elevated in the PCK rat model but not in models of related hepatorenal fibrocystic diseases, suggesting that they be molecular markers of the disease and that a combination of MALDI imaging with high-resolution MS methods and Fisher discriminant data analysis may be applicable for lipid marker discovery.

Multiclass classification of FDG PET scans for the distinction between Parkinson's disease and atypical parkinsonian syndromes

Most available pattern recognition methods in neuroimaging address binary classification problems. Here, we used relevance vector machine (RVM) in combination with booststrap resampling (‘bagging’) for non-hierarchical multiclass classification. The method was tested on 120 cerebral ¹⁸fluorodeoxyglucose (FDG) positron emission tomography (PET) scans performed in patients who exhibited parkinsonian clinical features for 3.5 years on average but that were outside the prevailing perception for Parkinson's disease (PD). A radiological diagnosis of PD was suggested for 30 patients at the time of PET imaging. However, at follow-up several years after PET imaging, 42 of them finally received a clinical diagnosis of PD. The remaining 78 APS patients were diagnosed with multiple system atrophy (MSA, N = 31), progressive supranuclear palsy (PSP, N = 26) and corticobasal syndrome (CBS, N = 21), respectively. With respect to this standard of truth, classification sensitivity, specificity, positive and negative predictive values for PD were 93% 83% 75% and 96%, respectively using binary RVM (PD vs. APS) and 90%, 87%, 79% and 94%, respectively, using multiclass RVM (PD vs. MSA vs. PSP vs. CBS). Multiclass RVM achieved 45%, 55% and 62% classification accuracy for, MSA, PSP and CBS, respectively. Finally, a majority confidence ratio was computed for each scan on the basis of class pairs that were the most frequently assigned by RVM. Altogether, the results suggest that automatic multiclass RVM classification of FDG PET scans achieves adequate performance for the early differentiation between PD and APS on the basis of cerebral FDG uptake patterns when the clinical diagnosis is felt uncertain. This approach cannot be recommended yet as an aid for distinction between the three APS classes under consideration.

•

Multiclass classification is one of the challenges of computer-aided diagnosis.

•

This was addressed here using relevance vector machine and bootstrap aggregation.

•

Performance was tested on FDG-PET scans from 120 parkinsonian patients.

•

Four diagnostic classes under consideration as defined on average 3.5 years after PET.

•

Confusion matrices, majority confidence ratio and discriminant maps were computed.

Perfusion SPECT and FDG-PET

Both perfusion SPECT and FDG-PET provide images that closely reflect neuronal activity. There is a characteristic regional impairment in Alzheimer’s disease (AD) that involves mainly the temporo-parietal association cortices, mesial temporal structures and to a more variable degree also the frontal association cortex. This pattern of functional impairment can provide a biomarker for diagnosis of AD and other neurodegenerative dementias at the clinical stage of mild cognitive impairment, and for monitoring of progression. FDG-PET is quantitatively more accurate and thus better suited to multicenter studies than perfusion SPECT. Regional metabolic and blood flow changes are closely related to clinical symptoms, and most areas involved in these changes will also develop significant cortical atrophy. FDG-PET is complementary to amyloid PET, which targets a molecular marker that does not have a close relation to current symptoms. Current restrictions in the availability and cost of FDG-PET are being reduced, as oncological FDG-PET is being adopted as a standard clinical service in most countries. Limitations in the availability of trained staff should be overcome by training programs set up by professional organizations. Against the background of the development of new criteria for diagnosing AD before the onset of dementia, FDG-PET is expected to play an increasing role in diagnosing patients at an early stage of AD and in clinical trials of drugs aimed at preventing or delaying the onset of dementia.

Multimodality approach in cryptogenic epilepsy with focus on morphometric 3T MRI

PURPOSE

This study aimed to investigate the potential contribution of morphometric MRI analysis in comparison to other modalities, such as MEG, SPECT and PET, in identifying the epileptogenic focus in patients with cryptogenic epilepsy.

PATIENTS AND METHODS

Study inclusion was limited to epilepsy patients with a monolobar focus hypothesis, as concluded from EEG/seizure semiology and the best individual concordance rate. Feature maps, generated by the MATLAB(®) "morphometric analysis program" (MAP), were evaluated by a neuroradiologist blinded to conventional MRI and the focus hypothesis (MAP(1)). In addition, the feature maps were also interpreted by simultaneous matching conventional MRI but, again, with the reader having no knowledge of the focus hypothesis (MAP(2)).

RESULTS

In 12 out of 51 patients, true-positive findings were achieved (MAP(1): sensitivity 24%; specificity 96%). The sensitivity of the MAP(1) results was superior extratemporally. After matching conventional MRI, FCD was traced in six of the 12 patients (MAP(2): sensitivity 12%; specificity 100%). MEG sensitivity was 62%. Sensitivity of interictal and ictal SPECT was 20% and 50%, respectively. PET was not as sensitive extratemporally (19%) as temporally (82%). The greatest correspondence with the best individual concordance rate was noted with PET (14/16; 88%) and MEG (8/10; 80%), followed by interictal (5/8; 63%) and ictal (9/15; 60%) SPECT. Results for MAP(1) were 53% (10/19), and 100% for MAP(2) (6/6).

CONCLUSION

Although MAP sensitivity and specificity results are lower in comparison to other modalities, implementation of the technique should be considered first, before arranging any further investigations. The present study results offer guidelines for the implementation, interpretation and concordance of diagnostic procedures.

Verification of predicted robustness and accuracy of multivariate analysis

The assessment of accuracy and robustness of multivariate analysis of FDG-PET brain images as presented in [Markiewicz, P.J., Matthews, J.C., Declerck, J., Herholz, K., 2009. Robustness of multivariate image analysis assessed by resampling techniques and applied to FDG-PET scans of patients with Alzheimer's disease. Neuroimage 46, 472-485.] using a homogeneous sample (from one centre) of small size is here verified using a heterogeneous sample (from multiple centres) of much larger size. Originally the analysis, which included principal component analysis (PCA) and Fisher discriminant analysis (FDA), was established using a sample of 42 subjects (19 Normal Controls (NCs) and 23 Alzheimer's disease (AD) patients) and here the analysis is verified using an independent sample of 166 subjects (86 NCs and 80 ADs) obtained from the ADNI database. It is shown that bootstrap resampling combined with the metric of the largest principal angle between PCA subspaces as well as the deliberate clinical misdiagnosis simulation can predict robustness of the multivariate analysis when used with new datasets. Cross-validation (CV) and the .632 bootstrap overestimated the predictive accuracy encouraging less robust solutions. Also, it is shown that the type of PET scanner and image reconstruction method has an impact on such analysis and affects the accuracy of the verification sample.

Perfusion Imaging with SPECT in the Era of Pathophysiology-Based Biomarkers for Alzheimer's Disease

SPECT allows registration of regional cerebral blood flow (rCBF) which is altered in a characteristic temporoparietal pattern in Alzheimer's Dementia. Numerous studies have shown the diagnostic value of reduced cerebral blood flow and metabolic changes using perfusion SPECT and FDG-PEPT in AD diagnosis as well as in differential diagnosis against frontotemporal dementia, dementia with Lewy bodies and vascular disease. Recently more pathophysiology-based biomarkers in CSF and Amyloid-PET tracers have been developed that probably have a higher diagnostic accuracy than the more indirect rCBF changes seen in perfusion SPECT. In the paper review, we describe recent advances in AD biomarkers as well as improvements in the SPECT technique.

Confirmation rate of blinded (99m)Tc-SPECT compared to neurochemical dementia biomarkers in CSF in patients with Alzheimer disease

In Alzheimer disease, CSF biomarkers and nuclear imaging are of particular interest. Many studies investigated only one technique, limiting comparison. Here, in 76 patients blinded 99mTc-SPECT was compared to CSF. Sensitivity of CSF was 92%; and 51% for SPECT. Specificity favored SPECT (90 vs. 80%). Both techniques showed no coherence (p = 0.17-0.47). Our results confirm that CSF biomarkers show higher sensitivity. SPECT has higher specificity and can also be used for other dementias without established CSF biomarkers.