scholarly journals 3D Segmentation of Perivascular Spaces on T1-Weighted 3 Tesla MR Images With a Convolutional Autoencoder and a U-Shaped Neural Network

2021 ◽  
Vol 15 ◽  
Author(s):  
Philippe Boutinaud ◽  
Ami Tsuchida ◽  
Alexandre Laurent ◽  
Filipa Adonias ◽  
Zahra Hanifehlou ◽  
...  
Keyword(s):  
Learning Performance ◽  
Similar Distribution ◽  
Perivascular Spaces ◽  
Visual Rating ◽  
Independent Expert ◽  
Convolutional Autoencoder ◽  
Mri Scans ◽  
Experienced Operator ◽  
Magnetic Resonance Imaging Mri ◽  
The Individual

We implemented a deep learning (DL) algorithm for the 3-dimensional segmentation of perivascular spaces (PVSs) in deep white matter (DWM) and basal ganglia (BG). This algorithm is based on an autoencoder and a U-shaped network (U-net), and was trained and tested using T1-weighted magnetic resonance imaging (MRI) data from a large database of 1,832 healthy young adults. An important feature of this approach is the ability to learn from relatively sparse data, which gives the present algorithm a major advantage over other DL algorithms. Here, we trained the algorithm with 40 T1-weighted MRI datasets in which all “visible” PVSs were manually annotated by an experienced operator. After learning, performance was assessed using another set of 10 MRI scans from the same database in which PVSs were also traced by the same operator and were checked by consensus with another experienced operator. The Sorensen-Dice coefficients for PVS voxel detection in DWM (resp. BG) were 0.51 (resp. 0.66), and 0.64 (resp. 0.71) for PVS cluster detection (volume threshold of 0.5 within a range of 0 to 1). Dice values above 0.90 could be reached for detecting PVSs larger than 10 mm3 and 0.95 for PVSs larger than 15 mm3. We then applied the trained algorithm to the rest of the database (1,782 individuals). The individual PVS load provided by the algorithm showed a high agreement with a semi-quantitative visual rating done by an independent expert rater, both for DWM and for BG. Finally, we applied the trained algorithm to an age-matched sample from another MRI database acquired using a different scanner. We obtained a very similar distribution of PVS load, demonstrating the interoperability of this algorithm.

scholarly journals 3D segmentation of perivascular spaces on T1-weighted 3 Tesla MR images with a convolutional autoencoder and a U-shaped neural network

2020 ◽  
Author(s):  
Philippe Boutinaud ◽  
Ami Tsuchida ◽  
Alexandre Laurent ◽  
Filipa Adonias ◽  
Zahra Hanifehlou ◽  
...  
Keyword(s):  
Learning Performance ◽  
Similar Distribution ◽  
Perivascular Spaces ◽  
Visual Rating ◽  
Independent Expert ◽  
Convolutional Autoencoder ◽  
Mri Scans ◽  
Experienced Operator ◽  
Magnetic Resonance Imaging Mri ◽  
The Individual

AbstractWe implemented a deep learning (DL) algorithm for the 3-dimensional segmentation of perivascular spaces (PVSs) in deep white matter (DWM) and basal ganglia (BG). This algorithm is based on an autoencoder and a U-shaped network (U-net), and was trained and tested using T1-weighted magnetic resonance imaging (MRI) data from a large database of 1,832 healthy young adults. An important feature of this approach is the ability to learn from relatively sparse data, which gives the present algorithm a major advantage over other DL algorithms. Here, we trained the algorithm with 40 T1-weighted MRI datasets in which all “visible” PVSs were manually annotated by an experienced operator. After learning, performance was assessed using another set of 10 MRI scans from the same database in which PVSs were also traced by the same operator and were checked by consensus with another experienced operator. The Sorensen-Dice coefficients for PVS voxel detection in DWM (resp. BG) were 0.51 (resp. 0.66), and 0.64 (resp. 0.71) for PVS cluster detection (volume threshold of 0.5 within a range of 0 to 1). Dice values above 0.90 could be reached for detecting PVSs larger than 10 mm3 and 0.95 for PVSs larger than 15 mm3. We then applied the trained algorithm to the rest of the database (1,782 individuals). The individual PVS load provided by the algorithm showed a high agreement with a semi-quantitative visual rating done by an independent expert rater, both for DWM and for BG. Finally, we applied the trained algorithm to an age-matched sample from another MRI database acquired using a different scanner. We obtained a very similar distribution of PVS load, demonstrating the interoperability of this algorithm.

scholarly journals A detailed analysis of anatomical plausibility of crossed and uncrossed streamline rendition of the dentato-rubro-thalamic tract (DRT(T)) in a commercial stereotactic planning system

2021 ◽  
Author(s):  
Volker A. Coenen ◽  
Bastian E. Sajonz ◽  
Peter C. Reinacher ◽  
Christoph P. Kaller ◽  
Horst Urbach ◽  
...  
Keyword(s):  
Surgical Planning ◽  
Ground Truth ◽  
Planning System ◽  
Single Subject ◽  
Fiber Structure ◽  
Native Space ◽  
Volume Of Interest ◽  
Mri Scans ◽  
Magnetic Resonance Imaging Mri ◽  
The Individual

Abstract Background An increasing number of neurosurgeons use display of the dentato-rubro-thalamic tract (DRT) based on diffusion weighted imaging (dMRI) as basis for their routine planning of stimulation or lesioning approaches in stereotactic tremor surgery. An evaluation of the anatomical validity of the display of the DRT with respect to modern stereotactic planning systems and across different tracking environments has not been performed. Methods Distinct dMRI and anatomical magnetic resonance imaging (MRI) data of high and low quality from 9 subjects were used. Six subjects had repeated MRI scans and therefore entered the analysis twice. Standardized DICOM structure templates for volume of interest definition were applied in native space for all investigations. For tracking BrainLab Elements (BrainLab, Munich, Germany), two tensor deterministic tracking (FT2), MRtrix IFOD2 (https://www.mrtrix.org), and a global tracking (GT) approach were used to compare the display of the uncrossed (DRTu) and crossed (DRTx) fiber structure after transformation into MNI space. The resulting streamlines were investigated for congruence, reproducibility, anatomical validity, and penetration of anatomical way point structures. Results In general, the DRTu can be depicted with good quality (as judged by waypoints). FT2 (surgical) and GT (neuroscientific) show high congruence. While GT shows partly reproducible results for DRTx, the crossed pathway cannot be reliably reconstructed with the other (iFOD2 and FT2) algorithms. Conclusion Since a direct anatomical comparison is difficult in the individual subjects, we chose a comparison with two research tracking environments as the best possible “ground truth.” FT2 is useful especially because of its manual editing possibilities of cutting erroneous fibers on the single subject level. An uncertainty of 2 mm as mean displacement of DRTu is expectable and should be respected when using this approach for surgical planning. Tractographic renditions of the DRTx on the single subject level seem to be still illusive.

scholarly journals Three-Dimensional Convolutional Autoencoder Extracts Features of Structural Brain Images With a “Diagnostic Label-Free” Approach: Application to Schizophrenia Datasets

2021 ◽  
Vol 15 ◽  
Author(s):  
Hiroyuki Yamaguchi ◽  
Yuki Hashimoto ◽  
Genichi Sugihara ◽  
Jun Miyata ◽  
Toshiya Murai ◽  
...  
Keyword(s):  
Psychiatric Disorders ◽  
Brain Imaging ◽  
Spatial Information ◽  
Three Dimensional ◽  
Clinical Information ◽  
Label Free ◽  
Brain Images ◽  
Convolutional Autoencoder ◽  
Mri Scans ◽  
Magnetic Resonance Imaging Mri

There has been increasing interest in performing psychiatric brain imaging studies using deep learning. However, most studies in this field disregard three-dimensional (3D) spatial information and targeted disease discrimination, without considering the genetic and clinical heterogeneity of psychiatric disorders. The purpose of this study was to investigate the efficacy of a 3D convolutional autoencoder (3D-CAE) for extracting features related to psychiatric disorders without diagnostic labels. The network was trained using a Kyoto University dataset including 82 patients with schizophrenia (SZ) and 90 healthy subjects (HS) and was evaluated using Center for Biomedical Research Excellence (COBRE) datasets, including 71 SZ patients and 71 HS. We created 16 3D-CAE models with different channels and convolutions to explore the effective range of hyperparameters for psychiatric brain imaging. The number of blocks containing two convolutional layers and one pooling layer was set, ranging from 1 block to 4 blocks. The number of channels in the extraction layer varied from 1, 4, 16, and 32 channels. The proposed 3D-CAEs were successfully reproduced into 3D structural magnetic resonance imaging (MRI) scans with sufficiently low errors. In addition, the features extracted using 3D-CAE retained the relation to clinical information. We explored the appropriate hyperparameter range of 3D-CAE, and it was suggested that a model with 3 blocks may be related to extracting features for predicting the dose of medication and symptom severity in schizophrenia.

scholarly journals Three-dimensional convolutional autoencoder extracts features of structural brain images with a “diagnostic label-free” approach: Application to schizophrenia datasets

2020 ◽  
Author(s):  
Hiroyuki Yamaguchi ◽  
Yuki Hashimoto ◽  
Genichi Sugihara ◽  
Jun Miyata ◽  
Toshiya Murai ◽  
...  
Keyword(s):  
Psychiatric Disorders ◽  
Spatial Information ◽  
Three Dimensional ◽  
Clinical Information ◽  
Label Free ◽  
Brain Images ◽  
Diagnostic Labels ◽  
Convolutional Autoencoder ◽  
Mri Scans ◽  
Magnetic Resonance Imaging Mri

ABSTRACTThere has been increasing interest in performing psychiatric brain imaging studies using deep learning. However, most studies in this field disregard three-dimensional (3D) spatial information and targeted disease discrimination, without considering the genetic and clinical heterogeneity of psychiatric disorders. The purpose of this study was to investigate the efficacy of a 3D convolutional autoencoder (CAE) for extracting features related to psychiatric disorders without diagnostic labels. The network was trained using a Kyoto University dataset including 82 patients with schizophrenia (SZ) and 90 healthy subjects (HS), and was evaluated using Center for Biomedical Research Excellence (COBRE) datasets including 71 SZ patients and 71 HS. The proposed 3D-CAEs were successfully reconstructed into high-resolution 3D structural magnetic resonance imaging (MRI) scans with sufficiently low errors. In addition, the features extracted using 3D-CAE retained the relevant clinical information. We explored the appropriate hyper parameter range of 3D-CAE, and it was suggested that a model with eight convolution layers might be relevant to extract features for predicting the dose of medication and symptom severity in schizophrenia.

scholarly journals Multimodal Neuroimaging Predictors of Learning Performance of Sensorimotor Rhythm Up-Regulation Neurofeedback

2021 ◽  
Vol 15 ◽  
Author(s):  
Linling Li ◽  
Yinxue Wang ◽  
Yixuan Zeng ◽  
Shaohui Hou ◽  
Gan Huang ◽  
...  
Keyword(s):  
Individual Difference ◽  
Resting State ◽  
Resting State Fmri ◽  
Learning Performance ◽  
Multimodal Neuroimaging ◽  
Multimodal Features ◽  
Neuroimaging Data ◽  
Eeg Neurofeedback ◽  
Magnetic Resonance Imaging Mri ◽  
The Individual

Electroencephalographic (EEG) neurofeedback (NFB) is a popular neuromodulation method to help one selectively enhance or inhibit his/her brain activities by means of real-time visual or auditory feedback of EEG signals. Sensory motor rhythm (SMR) NFB protocol has been applied to improve cognitive performance, but a large proportion of participants failed to self-regulate their brain activities and could not benefit from NFB training. Therefore, it is important to identify the neural predictors of SMR up-regulation NFB training performance for a better understanding the mechanisms of individual difference in SMR NFB. Twenty-seven healthy participants (12 males, age: 23.1 ± 2.36) were enrolled to complete three sessions of SMR up-regulation NFB training and collection of multimodal neuroimaging data [resting-state EEG, structural magnetic resonance imaging (MRI), and resting-state functional MRI (fMRI)]. Correlation analyses were performed between within-session NFB learning index and anatomical and functional brain features extracted from multimodal neuroimaging data, in order to identify the neuroanatomical and neurophysiological predictors for NFB learning performance. Lastly, machine learning models were trained to predict NFB learning performance using features from each modality as well as multimodal features. According to our results, most participants were able to successfully increase the SMR power and the NFB learning performance was significantly correlated with a set of neuroimaging features, including resting-state EEG powers, gray/white matter volumes from MRI, regional and functional connectivity (FC) of resting-state fMRI. Importantly, results of prediction analysis indicate that NFB learning index can be better predicted using multimodal features compared with features of single modality. In conclusion, this study highlights the importance of multimodal neuroimaging technique as a tool to explain the individual difference in within-session NFB learning performance, and could provide a theoretical framework for early identification of individuals who cannot benefit from NFB training.

scholarly journals MRI-Visible Perivascular Space (PVS) Changes with Long-Duration Spaceflight

2021 ◽  
Author(s):  
Kathleen Hupfeld ◽  
Sutton Richmond ◽  
Heather McGregor ◽  
Daniel Schwartz ◽  
Madison Luther ◽  
...  
Keyword(s):  
Morphological Changes ◽  
Space Station ◽  
Ventricular Volume ◽  
Perivascular Spaces ◽  
Median Length ◽  
Long Duration ◽  
Mri Scans ◽  
Magnetic Resonance Imaging Mri ◽  
And Shifts ◽  
Brain Homeostasis

Humans are exposed to extreme environmental stressors during spaceflight and return with alterations in brain structure and shifts in intracranial fluids. To date, no studies have evaluated the effects of spaceflight on perivascular spaces (PVSs) within the brain, which are believed to facilitate fluid drainage and brain homeostasis. Here, we examined how the number and morphology of magnetic resonance imaging (MRI)-visible PVSs are affected by spaceflight, including prior spaceflight experience. Fifteen astronauts underwent six T1-weighted 3T MRI scans, twice prior to launch and four times following their return to Earth after ~6-month missions to the International Space Station. White matter MRI-visible PVS number and morphology were calculated using an established automated segmentation algorithm. We found that novice astronauts showed an increase in total PVS volume from pre- to post-flight, whereas experienced crewmembers did not (adjusted for age, sex, and time between landing and first MRI scan). Moreover, experienced astronauts exhibited a significant correlation between more previous flight days and greater PVS median length at baseline, suggesting that experienced astronauts exhibit holdover effects from prior spaceflight(s). There was also a significant positive correlation between pre- to post-flight increases in PVS median length and increases in right lateral ventricular volume. The presence of spaceflight associated neuro-ocular syndrome (SANS) was not associated with PVS number or morphology. Together, these findings demonstrate that spaceflight is associated with PVS morphological changes, and specifically that spaceflight experience is an important factor in determining PVS characteristics.

Semi-automated methods for 3-D reconstruction of macromolecular complexes

1995 ◽  
Vol 53 ◽  
pp. 42-43
Author(s):  
B. Carragher ◽  
M. Whittaker
Keyword(s):  
Three Dimensional ◽  
Time Saving ◽  
Macromolecular Complexes ◽  
Symmetry Properties ◽  
Dimensional Reconstruction ◽  
Experienced Operator ◽  
Projection Images ◽  
The Individual ◽  
Natural Symmetry

Techniques for three-dimensional reconstruction of macromolecular complexes from electron micrographs have been successfully used for many years. These include methods which take advantage of the natural symmetry properties of the structure (for example helical or icosahedral) as well as those that use single axis or other tilting geometries to reconstruct from a set of projection images. These techniques have traditionally relied on a very experienced operator to manually perform the often numerous and time consuming steps required to obtain the final reconstruction. While the guidance and oversight of an experienced and critical operator will always be an essential component of these techniques, recent advances in computer technology, microprocessor controlled microscopes and the availability of high quality CCD cameras have provided the means to automate many of the individual steps.During the acquisition of data automation provides benefits not only in terms of convenience and time saving but also in circumstances where manual procedures limit the quality of the final reconstruction.

Accuracy of Rotator Cuff Tears and Tendinosis Diagnoses on Shoulder Ultrasound Performed by a Short-experienced Operator

2018 ◽  
Vol 1 ◽  
pp. 9
Author(s):  
Harshad Arvind Vanjare ◽  
Jyoti Panwar
Keyword(s):  
Rotator Cuff ◽  
Negative Predictive Value ◽  
Rotator Cuff Tear ◽  
Predictive Value ◽  
Rotator Cuff Tears ◽  
Mri Findings ◽  
Cuff Tear ◽  
Experienced Operator ◽  
Magnetic Resonance Imaging Mri ◽  
Sensitivity Specificity

Objective The objective of the study was to assess the accuracy of ultrasound examination for the diagnosis of rotator cuff tear and tendinosis performed by a short experienced operator, compared to magnetic resonance imaging (MRI) results. Method A total of 70 subjects suspected to have rotator cuff tear or tendinosis and planned for shoulder MRI were included in the study. Shoulder ultrasound was performed either before or after the MRI scan on the same day. Ultrasound operator had a short experience in performing an ultrasound of the shoulder. Ultrasound findings were correlated to MRI findings. Results Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for the diagnosis of tendinosis were 58%, 84%, 63%, 80%, and 75%, respectively, and it was 68%, 91%, 73%, 88%, and 85%, respectively, for the diagnosis of rotator cuff tear. Conclusions Sensitivity for diagnosing rotator cuff tear or tendinosis was moderate but had a higher negative predictive value. Thus, the ultrasound operator with a short experience in performing shoulder ultrasound had moderate sensitivity in diagnosing tendinosis or tears; however, could exclude them with confidence.

scholarly journals Regional medical practice variation in high-cost healthcare services

2021 ◽  
Author(s):  
Michael Berger ◽  
Thomas Czypionka
Keyword(s):  
Medical Practice ◽  
Regional Variation ◽  
Multivariate Regression ◽  
Healthcare Services ◽  
Practice Variation ◽  
District Level ◽  
Healthcare Data ◽  
Regional Medical ◽  
Mri Scans ◽  
Magnetic Resonance Imaging Mri

AbstractMagnetic resonance imaging (MRI) is a popular yet cost-intensive diagnostic measure whose strengths compared to other medical imaging technologies have led to increased application. But the benefits of aggressive testing are doubtful. The comparatively high MRI usage in Austria in combination with substantial regional variation has hence become a concern for its policy makers. We use a set of routine healthcare data on outpatient MRI service consumption of Austrian patients between Q3-2015 and Q2-2016 on the district level to investigate the extent of medical practice variation in a two-step statistical analysis combining multivariate regression models and Blinder–Oaxaca decomposition. District-level MRI exam rates per 1.000 inhabitants range from 52.38 to 128.69. Controlling for a set of regional characteristics in a multivariate regression model, we identify payer autonomy in regulating access to MRI scans as the biggest contributor to regional variation. Nevertheless, the statistical decomposition highlights that more than 70% of the regional variation remains unexplained by differences between the observable district characteristics. In the absence of epidemiological explanations, the substantial regional medical practice variation calls the efficiency of resource deployment into question.

scholarly journals Brain atrophy and clinical characteristics predicting SDMT performance in multiple sclerosis: A 10-year follow-up study

2021 ◽  
Vol 7 (1) ◽  
pp. 205521732199239
Author(s):  
Cecilie Jacobsen ◽  
Robert Zivadinov ◽  
Kjell-Morten Myhr ◽  
Turi O Dalaker ◽  
Ingvild Dalen ◽  
...  
Keyword(s):  
Clinical Data ◽  
Grey Matter Volume ◽  
Lesion Volume ◽  
Information Processing Speed ◽  
Disability Status ◽  
Mri Scans ◽  
Patients At Risk ◽  
Magnetic Resonance Imaging Mri ◽  
Specific Volumes

Objectives To identify Magnetic Resonance Imaging (MRI), clinical and demographic biomarkers predictive of worsening information processing speed (IPS) as measured by Symbol Digit Modalities Test (SDMT). Methods Demographic, clinical data and 1.5 T MRI scans were collected in 76 patients at time of inclusion, and after 5 and 10 years. Global and tissue-specific volumes were calculated at each time point. For the primary outcome of analysis, SDMT was used. Results Worsening SDMT at 5-year follow-up was predicted by baseline age, Expanded Disability Status Scale (EDSS), SDMT, whole brain volume (WBV) and T2 lesion volume (LV), explaining 30.2% of the variance of SDMT. At 10-year follow-up, age, EDSS, grey matter volume (GMV) and T1 LV explained 39.4% of the variance of SDMT change. Conclusion This longitudinal study shows that baseline MRI-markers, demographic and clinical data can help predict worsening IPS. Identification of patients at risk of IPS decline is of importance as follow-up, treatment and rehabilitation can be optimized.

Sign in / Sign up

Export Citation Format

Share Document