Diffusion and structural MRI as potential biomarkers in people with Parkinson's disease and cognitive impairment.

OBJECTIVE: To explore the neuroimage change in Parkinson's disease (PD) patients with cognitive impairments, this study investigated the correlation between plasma biomarkers and morphological brain changes in patients with normal cognition and mild cognitive impairment. The objective was to identify the potential target deposition regions of the plasma biomarkers and to search for the relevant early neuroimaging biomarkers on the basis of different cognitive domains. METHODS: Structural brain MRI and diffusion weighted images were analyzed from 49 eligible PD participants (male/female: 27/22; mean age: 73.4 ± 8.5 years) from a retrospective analysis. Plasma levels of α-synuclein, amyloid beta peptide, and total tau were collected. A comprehensive neuropsychological assessment of the general and specific cognitive domains was performed. Difference between PD patients with normal cognition and impairment was examined. Regression analysis was performed to evaluate the correlation between image-derived index and plasma biomarkers or neuropsychological assessments. RESULTS: Significant correlation was found between plasma Aß-42 level and fractional anisotropy of the middle occipital, angular, and middle temporal gyri of the left brain, as well as plasma T-tau level and the surface area of the isthmus or the average thickness of the posterior part of right cingulate gyrus. Visuospatial and executive function is positively correlated with axial diffusivity in bilateral cingulate gyri. CONCLUSION: In nondemented PD patients, the target regions for plasma deposition might be located in the cingulate, middle occipital, angular, and middle temporal gyri. Changes from multiple brain regions can be correlated to the performance of different cognitive domains. CLINICAL RELEVANCE STATEMENT: Cognitive impairment in Parkinson's disease is primarily linked to biomarkers associated with Alzheimer's disease rather than those related to Parkinson's disease and resembles the frontal variant of Alzheimer's disease, which may guide management strategies for cognitive impairment in Parkinson's disease. KEY POINTS: • Fractional anisotropy, surface area, and thickness in the cingulate, middle occipital, angular, and middle temporal gyri can be significantly correlated with plasma Aß-42 and T-tau level. • Axial diffusivity in the cingulate gyri was correlated with visuospatial and executive function. • The pattern of cognitive impairment in Parkinson's disease can be similar to the frontal variant than typical Alzheimer's disease.

Utilizing collimated aperture with proton pencil beam scanning (PBS) for stereotactic radiotherapy.

PURPOSE: Proton stereotactic radiosurgery (PSRS) has emerged as an innovative proton therapy modality aimed at achieving precise dose delivery with minimal impact on healthy tissues. This study explores the dosimetric outcomes of PSRS in comparison to traditional intensity-modulated proton therapy (IMPT) by focusing on cases with small target volumes. A custom-made aperture system designed for proton therapy, specifically tailored to small target volumes, was developed and implemented for this investigation. METHODS: A prerequisite mechanical validation through an isocentricity test precedes dosimetric assessments, ensuring the seamless integration of mechanical and dosimetry analyses. Five patients were enrolled in the study, including two with choroid melanoma and three with arteriovenous malformations (AVM). Two treatment plans were meticulously executed for each patient, one utilizing a collimated aperture and the other without. Both plans were subjected to robust optimization, maintaining identical beam arrangements and consistent optimization parameters to account for setup errors of 2 mm and range uncertainties of 3.5%. Plan evaluation metrics encompassing the Heterogeneity Index (HI), Paddick Conformity Index (CIPaddick), Gradient Index (GI), and the R50% index to evaluate alterations in low-dose volume distribution. RESULTS: The comparative analysis between PSRS and traditional PBS treatment revealed no significant differences in plan outcomes, with both modalities demonstrating comparable target coverage. However, collimated apertures resulted in discernible improvements in dose conformity, dose fall-off, and reduced low-dose volume. CONCLUSIONS: This study underscores the advantageous impact of the aperture system on proton therapy, particularly in cases involving small target volumes.

The neural basis of pain during labor.

Characterizing a labor pain-related neural signature is a key prerequisite for devising optimized pharmacologic and nonpharmacologic labor pain relief methods. The aim of this study was to describe the neural basis of labor pain and to provide a brief summary of how epidural anesthesia may affect pain-related neuronal activity during labor. Possible future directions are also highlighted. By taking advantage of functional magnetic resonance imaging, brain activation maps and functional neural networks of women during labor that have been recently characterized were compared between pregnant women who received epidural anesthesia and those who did not. In the subgroup of women who did not receive epidural anesthesia, labor-related pain elicited activations in a distributed brain network that included regions within the primary somatosensory cortex (postcentral gyrus and left parietal operculum cortex) and within the traditional pain network (lentiform nucleus, insula, and anterior cingulate gyrus). The activation maps of women who had been administered epidural anesthesia were found to be different-especially with respect to the postcentral gyrus, the insula, and the anterior cingulate gyrus. Parturients who received epidural anesthesia were also compared with those who did not in terms of functional connectivity from selected sensory and affective regions. When analyzing women who did not receive epidural anesthesia, marked bilateral connections from the postcentral gyrus to the superior parietal lobule, supplementary motor area, precentral gyrus, and the right anterior supramarginal gyrus were observed. In contrast, women who received epidural anesthesia showed fewer connections from the postcentral gyrus-being limited to the superior parietal lobule and supplementary motor area. Importantly, one of the most noticeable effects of epidural anesthesia was observed in the anterior cingulate cortex-a primary region that modulates pain perception. The increased outgoing connectivity from the anterior cingulate cortex in women who received epidural anesthesia indicates that the cognitive control exerted by this area might play a major role in the relief from labor pain. These findings not only affirmed the existence of a brain signature for pain experienced during labor, but they also showed that this signature can be altered by the administration of epidural anesthesia. This finding raises a question about the extent to which the cingulo-frontal cortex may exert top-down influences to gate women's experiences of labor-related pain. Because the anterior cingulate cortex is also involved in the processing and modulation of emotional content, such as fear and anxiety, a related question is about the extent to which the use of epidural anesthesia can affect different components of pain perception. Finally, inhibition of anterior cingulate cortex neurons may represent a potential new therapeutic target for alleviating labor-associated pain.

Fully automated segmentation and radiomics feature extraction of hypopharyngeal cancer on MRI using deep learning.

OBJECTIVES: To use convolutional neural network for fully automated segmentation and radiomics features extraction of hypopharyngeal cancer (HPC) tumor in MRI. METHODS: MR images were collected from 222 HPC patients, among them 178 patients were used for training, and another 44 patients were recruited for testing. U-Net and DeepLab V3 + architectures were used for training the models. The model performance was evaluated using the dice similarity coefficient (DSC), Jaccard index, and average surface distance. The reliability of radiomics parameters of the tumor extracted by the models was assessed using intraclass correlation coefficient (ICC). RESULTS: The predicted tumor volumes by DeepLab V3 + model and U-Net model were highly correlated with those delineated manually (p < 0.001). The DSC of DeepLab V3 + model was significantly higher than that of U-Net model (0.77 vs 0.75, p < 0.05), particularly in those small tumor volumes of < 10 cm3 (0.74 vs 0.70, p < 0.001). For radiomics extraction of the first-order features, both models exhibited high agreement (ICC: 0.71-0.91) with manual delineation. The radiomics extracted by DeepLab V3 + model had significantly higher ICCs than those extracted by U-Net model for 7 of 19 first-order features and for 8 of 17 shape-based features (p < 0.05). CONCLUSION: Both DeepLab V3 + and U-Net models produced reasonable results in automated segmentation and radiomic features extraction of HPC on MR images, whereas DeepLab V3 + had a better performance than U-Net. CLINICAL RELEVANCE STATEMENT: The deep learning model, DeepLab V3 + , exhibited promising performance in automated tumor segmentation and radiomics extraction for hypopharyngeal cancer on MRI. This approach holds great potential for enhancing the radiotherapy workflow and facilitating prediction of treatment outcomes. KEY POINTS: • DeepLab V3 + and U-Net models produced reasonable results in automated segmentation and radiomic features extraction of HPC on MR images. • DeepLab V3 + model was more accurate than U-Net in automated segmentation, especially on small tumors. • DeepLab V3 + exhibited higher agreement for about half of the first-order and shape-based radiomics features than U-Net.

International Multicenter Analysis of Brain Structure Across Clinical Stages of Parkinson's Disease.

BACKGROUND: Brain structure abnormalities throughout the course of Parkinson's disease have yet to be fully elucidated. OBJECTIVE: Using a multicenter approach and harmonized analysis methods, we aimed to shed light on Parkinson's disease stage-specific profiles of pathology, as suggested by in vivo neuroimaging. METHODS: Individual brain MRI and clinical data from 2357 Parkinson's disease patients and 1182 healthy controls were collected from 19 sources. We analyzed regional cortical thickness, cortical surface area, and subcortical volume using mixed-effects models. Patients grouped according to Hoehn and Yahr stage were compared with age- and sex-matched controls. Within the patient sample, we investigated associations with Montreal Cognitive Assessment score. RESULTS: Overall, patients showed a thinner cortex in 38 of 68 regions compared with controls (dmax  = -0.20, dmin  = -0.09). The bilateral putamen (dleft  = -0.14, dright  = -0.14) and left amygdala (d = -0.13) were smaller in patients, whereas the left thalamus was larger (d = 0.13). Analysis of staging demonstrated an initial presentation of thinner occipital, parietal, and temporal cortices, extending toward rostrally located cortical regions with increased disease severity. From stage 2 and onward, the bilateral putamen and amygdala were consistently smaller with larger differences denoting each increment. Poorer cognition was associated with widespread cortical thinning and lower volumes of core limbic structures. CONCLUSIONS: Our findings offer robust and novel imaging signatures that are generally incremental across but in certain regions specific to disease stages. Our findings highlight the importance of adequately powered multicenter collaborations. © 2021 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Exploring the Spectrum of Subcortical Hyperintensities and Cognitive Decline.

White matter hyperintensities (WMHs) include periventricular WMH (pvWMH) and deep WMH. When hyperintensities in the basal ganglia or brainstem are included, the collective term is subcortical hyperintensities. Both WMH and medial temporal lobe atrophy (MTA) are risk factors for cognitive decline. This prospective study enrolled participants aged 50-85 years and followed their neuropsychological assessments annually for 2 years to explore the interactive effects of WMH and MTA on longitudinal clinical decline. Brain MRI was performed at the beginning of enrollment. Of the 200 participants, 57 were "normal" individuals, 40 had dysexecutive mild cognitive impairment, 53 had amnestic mild cognitive impairment, and 50 had Alzheimer's disease (AD). Overall, MTA significantly correlated with pvWMH (p=0.0004) but not with deep WMH, as defined by criteria using the Scheltens' Scale. Total Scheltens' score was specifically associated with the domain of semantic fluency (beta=-0.4, 95% CI=-0.7 to -0.2, p=0.002), which remained significant when adjusting for MTA (beta=-0.3, 95% CI=-0.5 to -0.1, p=0.017). The pvWMH was significantly higher in AD subjects than in normal control subjects (beta=0.3, 95% CI=0.1 to 0.4, p=0.001), especially the periventricular occipital caps (beta=0.2, 95% CI=0.1 to 0.3, p=0.0003). Cox proportional hazards model showed that the periventricular bands (PVB) predicted 1-year clinical decline (hazard ratio [HR]=5.3, 95% CI=1.8 to 15.7, p=0.002), which remained significant when further adjusting for MTA (HR=4.0, 95% CI=1.3 to 12.1, p=0.013). In summary, pvWMH, especially the occipital caps, was correlated with MTA and the AD subgroup. Assessment of semantic fluency may be useful for the clinical evaluation of the degree of subcortical hyperintensity burden. Visual rating of PVB could be an independent predictor for 1-year clinical decline.

Alterations of diffusion tensor MRI parameters in the brains of patients with Parkinson's disease compared with normal brains: possible diagnostic use.

OBJECTIVES: To investigate the diagnostic performance of diffusion tensor imaging in patients with Parkinson's disease (PD). METHODS: We examined a total of 126 PD patients (68 males/58 females, mean age: 62.0 ±7.6 years) and 91 healthy controls (43 males/48 females, mean age: 59.8 ±7.2 years). Images were acquired on a 3 Tesla magnetic resonance scanner. The Camino software was used to normalize and parcellate diffusion-weighted images into 90 cerebral regions based on the automatic anatomical labelling template. The minimum, median, and maximum values of the mean/radial/axial diffusivity/fractional anisotropy were determined. The diagnostic performance was assessed by receiver operating characteristic analysis. The associations of imaging parameters with disease severity were tested using Pearson's correlation coefficients after adjustment for disease duration. RESULTS: Compared with healthy controls, PD patients showed increased diffusivity in multiple cortical regions that extended beyond the basal ganglia. An area under curve of 85 % was identified for the maximum values of mean diffusivity in the ipsilateral middle temporal gyrus. The most significant intergroup difference was 26.8 % for the ipsilateral inferior parietal gyrus. CONCLUSION: The measurement of water diffusion from the parcellated cortex may be clinically useful for the assessment of PD patients. KEY POINTS: • Increased diffusivity was identified in multiple cortical regions of Parkinson's disease patients. • The area under the receiver operating curve was 85 % in the middle temporal gyrus. • The ipsilateral inferior parietal gyrus showed the most significant change.

Dynamic contrast-enhanced MRI, diffusion-weighted MRI and 18F-FDG PET/CT for the prediction of survival in oropharyngeal or hypopharyngeal squamous cell carcinoma treated with chemoradiation.

OBJECTIVES: We prospectively investigated the roles of pretreatment dynamic contrast-enhanced MR imaging (DCE-MRI), diffusion-weighted MR imaging (DWI) and 18F-fluorodeoxyglucose-positron emission tomography (18F-FDG PET)/CT for predicting survival of oropharyngeal or hypopharyngeal squamous cell carcinoma (OHSCC) patients treated with chemoradiation. METHODS: Patients with histologically proven OHSCC and neck nodal metastases scheduled for chemoradiation were eligible. Clinical variables as well as DCE-MRI-, DWI- and 18F-FDG PET/CT-derived parameters of the primary tumours and metastatic neck nodes were analysed in relation to 3-year progression-free survival (PFS) and overall survival (OS) rates. RESULTS: Eighty-six patients were available for analysis. Multivariate analysis identified the efflux rate constant (K ep)-tumour < 3.79 min-1 (P = 0.001), relative volume of extracellular extravascular space (V e)-node < 0.23 (P = 0.004) and SUVmax-tumour > 19.44 (P = 0.025) as independent risk factors for both PFS and OS. A scoring system based upon the sum of each of the three imaging parameters allowed stratification of our patients into three groups (patients with 0/1 factor, patients with 2 factors and patients with 3 factors, respectively) with distinct PFS (3-year rates = 72 %, 38 % and 0 %, P < 0.0001) and OS (3-year rates = 81 %, 46 % and 20 %, P < 0.0001). CONCLUSIONS: K ep-tumour, V e-node and SUVmax-tumour were independent prognosticators for OHSCC treated with chemoradiation. Their combination helped survival stratification. KEY POINTS: • K ep -tumour, V e -node and SUV max -tumour are independent predictors of survival rates. • The combination of these three prognosticators may help stratification of survival. • MRI and FDG-PET/CT play complementary roles in prognostication of head and neck cancer.

Myocardial triglyceride content at 3 T cardiovascular magnetic resonance and left ventricular systolic function: a cross-sectional study in patients hospitalized with acute heart failure.

BACKGROUND: Increased myocardial triglyceride (TG) content has been recognized as a risk factor for cardiovascular disease. However, its relation with cardiac function in patients on recovery from acute heart failure (HF) remains unclear. In this cross-sectional study, we sought to investigate the association between myocardial TG content measured on magnetic resonance spectroscopy ((1)H-MRS) and left ventricular (LV) function assessed on cardiovascular magnetic resonance (CMR) in patients who were hospitalized with HF. METHODS: A total of 50 patients who were discharged after hospitalization for acute HF and 21 age- and sex-matched controls were included in the study. Myocardial TG content and LV parameters (function and mass) were measured on a 3.0 T MR scanner. Fatty acid (FA) and unsaturated fatty acid (UFA) content was normalized against water (W) using the LC-Model algorithm. The patient population was dichotomized according to the left ventricular ejection fraction (LVEF, <50% or ≥ 50%). RESULTS: H-MRS data were available for 48 patients and 21 controls. Of the 48 patients, 25 had a LVEF <50% (mean, 31.2%), whereas the remaining 23 had a normal LVEF (mean, 60.2%). Myocardial UFA/W ratio was found to differ significantly in patients with low LVEF, normal LVEF, and controls (0.79% vs. 0.21% vs. 0.14%, respectively, p = 0.02). The myocardial UFA/TG ratio was associated with LV mass (r = 0.39, p < 0.001) and modestly related to LV end-diastolic volume (LVEDV; r = 0.24, p = 0.039). We also identified negative correlations of the myocardial FA/TG ratio with both LV mass (r = -0.39, p < 0.001) and LVEDV (r = -0.24, p = 0.039). CONCLUSIONS: As compared with controls, patients who were discharged after hospitalization for acute HF had increased myocardial UFA content; furthermore, UFA was inversely related with LVEF, LV mass and, to a lesser extent, LVEDV. Our study may stimulate further research on the measure of myocardial UFA content by (1)H-MRS for outcome prediction. TRIAL REGISTRATION: ClinicalTrial.gov: NCT02378402 . Registered 27/02/2015.

A simple method to improve the quality of diffusion-weighted magnetic resonance imaging with rapid histologic correlation in a murine model.

Diffusion-weighted magnetic resonance imaging (DW-MRI) has been used extensively in biomedical research. However, this technique has often suffered from distortion artifacts because of the magnetic field inhomogeneity surrounding the tissues. Histology is important for validating MRI interpretations, but correlating MRIs with tissue samples is challenging. Here we propose a method to improve DW-MRI and facilitate the matching between MRIs and tissue samples. A cryostat embedding medium, optimal cutting temperature (OCT) compound, was used to cover the examined target during the MRI studies. Frozen OCT compound could aid the examined target to be sectioned in parallel with the imaging plane. Phantom experiments demonstrated that embedding in OCT compound improved the magnetic field inhomogeneity while maintaining the apparent diffusion coefficient. Animal experiments revealed significantly reduced distortions in DW images in both the axial and coronal planes. The in vivo MRIs were easily matched with histologic specimens in a slice-to-slice fashion to examine the corresponding tissue microenvironment. This simple method might improve the quality of DW-MRI and provide histologic information for MRI to serve as an image biomarker.

Magnetic resonance monitoring of focused ultrasound/magnetic nanoparticle targeting delivery of therapeutic agents to the brain.

The superparamagnetic properties of magnetic nanoparticles (MNPs) allow them to be guided by an externally positioned magnet and also provide contrast for MRI. However, their therapeutic use in treating CNS pathologies in vivo is limited by insufficient local accumulation and retention resulting from their inability to traverse biological barriers. The combined use of focused ultrasound and magnetic targeting synergistically delivers therapeutic MNPs across the blood-brain barrier to enter the brain both passively and actively. Therapeutic MNPs were characterized and evaluated both in vitro and in vivo, and MRI was used to monitor and quantify their distribution in vivo. The technique could be used in normal brains or in those with tumors, and significantly increased the deposition of therapeutic MNPs in brains with intact or compromised blood-brain barriers. Synergistic targeting and image monitoring are powerful techniques for the delivery of macromolecular chemotherapeutic agents into the CNS under the guidance of MRI.

Neuroplastic Changes Associated With Hybrid Exercise-Cognitive Training in Stroke Survivors With Mild Cognitive Decline: A Randomized Controlled Trial.

BACKGROUND: Exercise and cognitive training have been shown to induce neuroplastic changes and modulate cognitive function following stroke. However, it remains unclear whether hybridized exercise-cognitive training facilitates cortical activity and further influences cognitive function after stroke. OBJECTIVE: The study aimed to investigate the effects of 2 hybridized exercise-cognitive trainings on neuroplastic changes and behavioral outcomes in stroke survivors with mild cognitive decline. METHODS: This study was a single-blind randomized controlled trial. Stroke survivors were randomly assigned to 1 of 3 groups: (1) sequential exercise-cognitive training (SEQ), (2) dual-task exercise-cognitive training (DUAL), or (3) control group (CON). All groups underwent training 60 min per day, 3 days per week, for a total of 12 weeks. The primary outcome was the resting-state (RS) functional connectivity (FC) in functional magnetic resonance imaging. Secondary behavioral outcomes included cognitive and physical functions. RESULTS: After 12 weeks of training, patients in the SEQ group (n = 21) exhibited increased RS FC between the left occipital lobe and posterior cingulate gyrus with right parietal lobe, compared to the DUAL (n = 22) and CON (n = 20) groups. Additionally, patients in the DUAL group showed increased FC of the left temporal lobe. However, changes in behavioral outcome measures were non-significant among the 3 groups (all P's > .05). CONCLUSIONS: This study highlights the distinct neuroplastic mechanisms associated with 2 types of exercise-cognitive hybridized trainings. The pre-post functional magnetic resonance imaging measurements illustrated the time course of neural mechanisms for cognitive recovery in stroke survivors following different exercise-cognitive training approaches. Trial registration. NCT03230253.

Diffusion tensor imaging for the differential diagnosis of Parkinsonism by machine learning.

BACKGROUND: There are currently no specific tests for either idiopathic Parkinson's disease or Parkinson-plus syndromes. The study aimed to investigate the diagnostic performance of features extracted from the whole brain using diffusion tensor imaging concerning parkinsonian disorders. METHODS: The retrospective data yielded 625 participants (average age: 61.4 ± 8.2, men/women: 313/312; healthy controls/idiopathic Parkinson's disease/multiple system atrophy/progressive supranuclear palsy: 219/286/51/69) between 2008 and 2017. Diffusion-weighted images were obtained using a 3T MR scanner. The 90th, 50th, and 10th percentiles of fractional anisotropy and mean/axial/radial diffusivity from each parcellated brain area were recorded. Statistical analysis was evaluated based on the features extracted from the whole brain, as determined using discriminant function analysis and support vector machine. 20% of the participants were used as an independent blind dataset with 5 times cross-verification. Diagnostic performance was evaluated by the sensitivity and the F1 score. RESULTS: Diagnoses were accurate for distinguishing idiopathic Parkinson's disease from healthy control and Parkinson-plus syndromes (87.4 ± 2.1% and 82.5 ± 3.9%, respectively). Diagnostic F1 scores varied for Parkinson-plus syndromes with 67.2 ± 3.8% for multiple system atrophy and 71.6 ± 3.5% for progressive supranuclear palsy. For early and late detection of idiopathic Parkinson's disease, the diagnostic performance was 79.2 ± 7.4% and 84.4 ± 6.9%, respectively. The diagnostic performance was 68.8 ± 11.0% and 52.5 ± 8.9% in early and late detection to distinguish different Parkinson-plus syndromes. CONCLUSIONS: Features extracted from diffusion tensor imaging of the whole brain can provide objective evidence for the diagnosis of healthy control, idiopathic Parkinson's disease, and Parkinson-plus syndromes with fair to very good diagnostic performance.

The Image-Histology Correlation of Subcutaneous mPEG-poly(Ala) Hydrogel-Embedded MIN6 Cell Grafts in Nude Mice.

Previously, we have successfully used noninvasive magnetic resonance (MR) and bioluminescence imaging to detect and monitor mPEG-poly(Ala) hydrogel-embedded MIN6 cells at the subcutaneous space for up to 64 days. In this study, we further explored the histological evolution of MIN6 cell grafts and correlated it with image findings. MIN6 cells were incubated overnight with chitosan-coated superparamagnetic iron oxide (CSPIO) and then 5 × 106 cells in the 100 µL hydrogel solution were injected subcutaneously into each nude mouse. Grafts were removed and examined the vascularization, cell growth and proliferation with anti-CD31, SMA, insulin and ki67 antibodies, respectively, at 8, 14, 21, 29 and 36 days after transplantation. All grafts were well-vascularized with prominent CD31 and SMA staining at all time points. Interestingly, insulin-positive cells and iron-positive cells were scattered in the graft at 8 and 14 days; while clusters of insulin-positive cells without iron-positive cells appeared in the grafts at 21 days and persisted thereafter, indicating neogrowth of MIN6 cells. Moreover, proliferating MIN6 cells with strong ki67 staining was observed in 21-, 29- and 36-day grafts. Our results indicate that the originally transplanted MIN6 cells proliferated from 21 days that presented distinctive bioluminescence and MR images.

A fixel-based analysis of white matter reductions early detects Parkinson disease with mild cognitive impairment.

BACKGROUND: White matter (WM) tract alterations are early signs of cognitive impairment in Parkinson disease (PD) patients. Fixel-based analysis (FBA) has advantages over traditional diffusion tensor imaging in managing complex and crossing fibers. We used FBA to measure fiber-specific changes in patients with PD mild cognitive impairment (PD-MCI) and PD normal cognition (PD-NC). METHODS: Seventy-one patients with PD without dementia were included: 39 PD-MCI and 32 PD-NC. All underwent diffusion-weighted imaging, clinical examinations, and tests to evaluate their cognitive function globally and in five cognitive domains. FBA was used to investigate fiber-tract alterations and compare PD-MCI with PD-NC subjects. Correlations with each cognitive test were analyzed. RESULTS: Patients with PD-MCI were significantly older (P = 0.044), had a higher male-to-female ratio (P = 0.006) and total Unified Parkinson's Disease Rating Scale score (P = 0.001). All fixel-based metrics were significantly reduced within the body of the corpus callosum and superior corona radiata in PD-MCI patients (family-wise error-corrected P value < 0.05) compared with PD-NC patients. The cingulum, superior longitudinal fasciculi, and thalamocortical circuit exhibited predominantly fiber-bundle cross-section (FC) changes. In regression analysis, reduced FC values in cerebellar circuits were associated with poor motor function in PD-MCI patients and poor picture-naming ability in PD-NC patients. CONCLUSIONS: PD-MCI patients have significant WM alterations compared with PD-NC patients. FBA revealed these changes in various bundle tracts, helping us to better understand specific WM changes that are functionally implicated in PD cognitive decline. FBA is potentially useful in detecting early cognitive decline in PD.

Generalizable transfer learning of automated tumor segmentation from cervical cancers toward a universal model for uterine malignancies in diffusion-weighted MRI.

PURPOSE: To investigate the generalizability of transfer learning (TL) of automated tumor segmentation from cervical cancers toward a universal model for cervical and uterine malignancies in diffusion-weighted magnetic resonance imaging (DWI). METHODS: In this retrospective multicenter study, we analyzed pelvic DWI data from 169 and 320 patients with cervical and uterine malignancies and divided them into the training (144 and 256) and testing (25 and 64) datasets, respectively. A pretrained model was established using DeepLab V3 + from the cervical cancer dataset, followed by TL experiments adjusting the training data sizes and fine-tuning layers. The model performance was evaluated using the dice similarity coefficient (DSC). RESULTS: In predicting tumor segmentation for all cervical and uterine malignancies, TL models improved the DSCs from the pretrained cervical model (DSC 0.43) when adding 5, 13, 26, and 51 uterine cases for training (DSC improved from 0.57, 0.62, 0.68, 0.70, p < 0.001). Following the crossover at adding 128 cases (DSC 0.71), the model trained by combining data from adding all the 256 patients exhibited the highest DSCs for the combined cervical and uterine datasets (DSC 0.81) and cervical only dataset (DSC 0.91). CONCLUSIONS: TL may improve the generalizability of automated tumor segmentation of DWI from a specific cancer type toward multiple types of uterine malignancies especially in limited case numbers.

Blind estimation of the arterial input function in dynamic contrast-enhanced MRI using purity maximization.

Uncertainty in arterial input function (AIF) estimation is one of the major errors in the quantification of dynamic contrast-enhanced MRI. A blind source separation algorithm was proposed to determine the AIF by selecting the voxel time course with maximum purity, which represents a minimal contamination from partial volume effects. Simulations were performed to assess the partial volume effect on the purity of AIF, the estimation accuracy of the AIF, and the influence of purity on the derived kinetic parameters. In vivo data were acquired from six patients with hypopharyngeal cancer and eight rats with brain tumor. Results showed that in simulation the AIF with the highest purity is closest to the true AIF. In patients, the manually selection had reduced purity, which could lead to underestimations of K(trans) and V(e) and an overestimation of V(p) when compared with those obtained by the proposed blind source separation algorithm. The derived kinetic parameters in the tumor were more susceptible to the changes in purity when compared with those in the muscle. The animal experiment demonstrated good reproducibility in blind source separation-AIF derived parameters. In conclusion, the blind source separation method is feasible and reproducible to identify the voxel with the tracer concentration time course closest to the true AIF.

Clinical significance of the pallidoreticular pathway in patients with carbon monoxide intoxication.

Whereas globus pallidus lesions resulting from carbon monoxide intoxication have been extensively described in the literature, the clinical significance of pallidoreticular lesions has rarely been mentioned. This study incorporated information from functional and structural imaging to explore the correlations of pallidoreticular lesions with parkinsonian features and neurobehavioural performance. Twenty-five patients (11 males) with globus pallidus lesions after carbon monoxide intoxication and 25 age- and sex-matched controls were enrolled for detailed neurological examinations, cognitive testing, susceptibility weighted imaging, diffusion tensor imaging and 99mTc-TRODAT-1 single photon emission computed tomography. The post-processing analysis of the neuroimaging included voxel-based morphometry to assess the regional atrophy, tract-based spatial statistics related to white matter involvement, tractography to investigate the rostral and caudal projections from the midbrain level and specific uptake ratios of 99mTc-TRODAT-1 for presynaptic dopaminergic transporter activity. In susceptibility weighted imaging, low-intensity pallidoreticular lesions were detected from the minimal-intensity projections, which were visible in only 7.7% of the T(1)-weighted images and 15.4% of the T(2)-weighted images, whereas inhomogeneous intensities were detected in the globus pallidus. The patients were further divided into two subgroups based on the presence (n = 13) or absence (n = 12) of pallidoreticular lesions. The patients with pallidoreticular lesions showed increased parkinsonian features, poorer performances on the neuropsychiatric tests, lower 99mTc-TRODAT-1 availability in both the caudate and the putamen and greater atrophy of the thalamus, posterior corpus callosum, cerebral peduncle and white matter surrounding the globus pallidus compared to those without pallidoreticular lesions. The tractography results obtained with seed regions of interest in the substantia nigra showed rostral projections to the supplementary motor cortex and anterior cingulate cortex via the globus pallidus; the two pathways were distinct but ran in parallel, caudal to the level of the globus pallidus. In conclusion, the presence of pallidoreticular lesions after carbon monoxide intoxication indicates a poorer cognitive state, which is associated with extensive grey and white matter damage in addition to the damage to the nigra-striatal neuronal networks. The presence of parkinsonian features may be related to pallidal and presynaptic dopaminergic dysfunction. The sensitivity for detecting pallidoreticular lesions can be greatly improved by using susceptibility weighted imaging compared with conventional imaging.

Deep Learning-Based Brain Computed Tomography Image Classification with Hyperparameter Optimization through Transfer Learning for Stroke.

Brain computed tomography (CT) is commonly used for evaluating the cerebral condition, but immediately and accurately interpreting emergent brain CT images is tedious, even for skilled neuroradiologists. Deep learning networks are commonly employed for medical image analysis because they enable efficient computer-aided diagnosis. This study proposed the use of convolutional neural network (CNN)-based deep learning models for efficient classification of strokes based on unenhanced brain CT image findings into normal, hemorrhage, infarction, and other categories. The included CNN models were CNN-2, VGG-16, and ResNet-50, all of which were pretrained through transfer learning with various data sizes, mini-batch sizes, and optimizers. Their performance in classifying unenhanced brain CT images was tested thereafter. This performance was then compared with the outcomes in other studies on deep learning-based hemorrhagic or ischemic stroke diagnoses. The results revealed that among our CNN-2, VGG-16, and ResNet-50 analyzed by considering several hyperparameters and environments, the CNN-2 and ResNet-50 outperformed the VGG-16, with an accuracy of 0.9872; however, ResNet-50 required a longer time to present the outcome than did the other networks. Moreover, our models performed much better than those reported previously. In conclusion, after appropriate hyperparameter optimization, our deep learning-based models can be applied to clinical scenarios where neurologist or radiologist may need to verify whether their patients have a hemorrhage stroke, an infarction, and any other symptom.

Deep learning based diagnosis of Parkinson's Disease using diffusion magnetic resonance imaging.

The diagnostic performance of a combined architecture on Parkinson's disease using diffusion tensor imaging was evaluated. A convolutional neural network was trained from multiple parcellated brain regions. A greedy algorithm was proposed to combine the models from individual regions into a complex one. Total 305 Parkinson's disease patients (aged 59.9±9.7 years old) and 227 healthy control subjects (aged 61.0±7.4 years old) were enrolled from 3 retrospective studies. The participants were divided into training with ten-fold cross-validation (N = 432) and an independent blind dataset (N = 100). Diffusion-weighted images were acquired from a 3T scanner. Fractional anisotropy and mean diffusivity were calculated and was subsequently parcellated into 90 cerebral regions of interest based on the Automatic Anatomic Labeling template. A convolutional neural network was implemented which contained three convolutional blocks and a fully connected layer. Each convolutional block consisted of a convolutional layer, activation layer, and pooling layer. This model was trained for each individual region. A greedy algorithm was implemented to combine multiple regions as the final prediction. The greedy algorithm predicted the area under curve of 94.1±3.2% from the combination of fractional anisotropy from 22 regions. The model performance analysis showed that the combination of 9 regions is equivalent. The best area under curve was 74.7±5.4% from the right postcentral gyrus. The current study proposed an architecture of convolutional neural network and a greedy algorithm to combine from multiple regions. With diffusion tensor imaging, the algorithm showed the potential to distinguish patients with Parkinson's disease from normal control with satisfactory performance.