scholarly journals Using Deep Learning to Detect Spinal Cord Diseases on Thoracolumbar Magnetic Resonance Images of Dogs

2021 ◽  
Vol 8 ◽  
Author(s):  
Anika Biercher ◽  
Sebastian Meller ◽  
Jakob Wendt ◽  
Norman Caspari ◽  
Johannes Schmidt-Mosig ◽  
...  
Keyword(s):  
Spinal Cord ◽  
Deep Learning ◽  
Magnetic Resonance ◽  
Intervertebral Disc ◽  
Magnetic Resonance Images ◽  
Training Data ◽  
Mr Images ◽  
Disc Protrusion ◽  
Fibrocartilaginous Embolism ◽  
T2 Weighted Images

Deep Learning based Convolutional Neural Networks (CNNs) are the state-of-the-art machine learning technique with medical image data. They have the ability to process large amounts of data and learn image features directly from the raw data. Based on their training, these networks are ultimately able to classify unknown data and make predictions. Magnetic resonance imaging (MRI) is the imaging modality of choice for many spinal cord disorders. Proper interpretation requires time and expertise from radiologists, so there is great interest in using artificial intelligence to more quickly interpret and diagnose medical imaging data. In this study, a CNN was trained and tested using thoracolumbar MR images from 500 dogs. T1- and T2-weighted MR images in sagittal and transverse planes were used. The network was trained with unremarkable images as well as with images showing the following spinal cord pathologies: intervertebral disc extrusion (IVDE), intervertebral disc protrusion (IVDP), fibrocartilaginous embolism (FCE)/acute non-compressive nucleus pulposus extrusion (ANNPE), syringomyelia and neoplasia. 2,693 MR images from 375 dogs were used for network training. The network was tested using 7,695 MR images from 125 dogs. The network performed best in detecting IVDPs on sagittal T1-weighted images, with a sensitivity of 100% and specificity of 95.1%. The network also performed very well in detecting IVDEs, especially on sagittal T2-weighted images, with a sensitivity of 90.8% and specificity of 98.98%. The network detected FCEs and ANNPEs with a sensitivity of 62.22% and a specificity of 97.90% on sagittal T2-weighted images and with a sensitivity of 91% and a specificity of 90% on transverse T2-weighted images. In detecting neoplasms and syringomyelia, the CNN did not perform well because of insufficient training data or because the network had problems differentiating different hyperintensities on T2-weighted images and thus made incorrect predictions. This study has shown that it is possible to train a CNN in terms of recognizing and differentiating various spinal cord pathologies on canine MR images. CNNs therefore have great potential to act as a “second eye” for imagers in the future, providing a faster focus on the altered image area and thus increasing workflow in radiology.

scholarly journals Deep Learning for Accurate Segmentation of Venous Thrombus from Black-Blood Magnetic Resonance Images: A Multicenter Study

2021 ◽  
Vol 2021 ◽  
pp. 1-11
Author(s):  
Chuanqi Sun ◽  
Xiangyu Xiong ◽  
Tianjing Zhang ◽  
Xiuhong Guan ◽  
Huan Mao ◽  
...  
Keyword(s):  
Deep Learning ◽  
Magnetic Resonance ◽  
Ground Truth ◽  
Magnetic Resonance Images ◽  
Black Blood ◽  
Mr Images ◽  
Vessel Segment ◽  
Learning Network ◽  
Venous Thrombus ◽  
Deep Learning Network

Objective. Deep vein thrombosis (DVT) is the third-largest cardiovascular disease, and accurate segmentation of venous thrombus from the black-blood magnetic resonance (MR) images can provide additional information for personalized DVT treatment planning. Therefore, a deep learning network is proposed to automatically segment venous thrombus with high accuracy and reliability. Methods. In order to train, test, and external test the developed network, total images of 110 subjects are obtained from three different centers with two different black-blood MR techniques (i.e., DANTE-SPACE and DANTE-FLASH). Two experienced radiologists manually contoured each venous thrombus, followed by reediting, to create the ground truth. 5-fold cross-validation strategy is applied for training and testing. The segmentation performance is measured on pixel and vessel segment levels. For the pixel level, the dice similarity coefficient (DSC), average Hausdorff distance (AHD), and absolute volume difference (AVD) of segmented thrombus are calculated. For the vessel segment level, the sensitivity (SE), specificity (SP), accuracy (ACC), and positive and negative predictive values (PPV and NPV) are used. Results. The proposed network generates segmentation results in good agreement with the ground truth. Based on the pixel level, the proposed network achieves excellent results on testing and the other two external testing sets, DSC are 0.76, 0.76, and 0.73, AHD (mm) are 4.11, 6.45, and 6.49, and AVD are 0.16, 0.18, and 0.22. On the vessel segment level, SE are 0.95, 0.93, and 0.81, SP are 0.97, 0.92, and 0.97, ACC are 0.96, 0.94, and 0.95, PPV are 0.97, 0.82, and 0.96, and NPV are 0.97, 0.96, and 0.94. Conclusions. The proposed deep learning network is effective and stable for fully automatic segmentation of venous thrombus on black blood MR images.

Prognostic relevance of the postoperative evolution of intramedullary spinal cord changes in signal intensity on magnetic resonance imaging after anterior decompression for cervical spondylotic myelopathy

2007 ◽  
Vol 7 (6) ◽  
pp. 615-622 ◽  
Author(s):  
Luciano Mastronardi ◽  
Ahmed Elsawaf ◽  
Raffaelino Roperto ◽  
Alessandro Bozzao ◽  
Manuela Caroli ◽  
...  
Keyword(s):  
Spinal Cord ◽  
Magnetic Resonance ◽  
Mr Imaging ◽  
Signal Intensity ◽  
Cervical Spondylotic Myelopathy ◽  
Mr Images ◽  
Intramedullary Spinal Cord ◽  
Signal Changes ◽  
T1 And T2 ◽  
T2 Weighted Images

Object Areas of intramedullary signal intensity changes (hypointensity on T1-weighted magnetic resonance [MR] images and hyperintensity on T2-weighted MR images) in patients with cervical spondylotic myelopathy (CSM) have been described by several investigators. The role of postoperative evolution of these alterations is still not well known. Methods A total of 47 patients underwent MR imaging before and at the end of the surgical procedure (intraoperative MR imaging [iMRI]) for cervical spine decompression and fusion using an anterior approach. Imaging was performed with a 1.5-tesla scanner integrated with the operative room (BrainSuite). Patients were followed clinically and evaluated using the Japanese Orthopaedic Association (JOA) and Nurick scales and also underwent MR imaging 3 and 6 months after surgery. Results Preoperative MR imaging showed an alteration (from the normal) of the intramedullary signal in 37 (78.7%) of 47 cases. In 23 cases, signal changes were altered on both T1- and T2-weighted images, and in 14 cases only on T2-weighted images. In 12 (52.2%) of the 23 cases, regression of hyperintensity on T2-weighted imaging was observed postoperatively. In 4 (17.4%) of these 23 cases, regression of hyperintensity was observed during the iMRI at the end of surgery. Residual compression on postoperative iMRI was not detected in any patients. A nonsignificant correlation was observed between postoperative expansion of the transverse diameter of the spinal cord at the level of maximal compression and the postoperative JOA score and Nurick grade. A statistically significant correlation was observed between the surgical result and the length of a patient's clinical history. A significant correlation was also observed according to the preoperative presence of intramedullary signal alteration. The best results were found in patients without spinal cord changes of signal, acceptable results were observed in the presence of changes on T2-weighted imaging only, and the worst results were observed in patients with spinal cord signal changes on both T1- and T2-weighted imaging. Finally, a statistically significant correlation was observed between patients with postoperative spinal cord signal change regression and better outcomes. Conclusions Intramedullary spinal cord changes in signal intensity in patients with CSM can be reversible (hyperintensity on T2-weighted imaging) or nonreversible (hypointensity on T1-weighted imaging). The regression of areas of hyperintensity on T2-weighted imaging is associated with a better prognosis, whereas the T1-weighted hypointensity is an expression of irreversible damage and, therefore, the worst prognosis. The preliminary experience with this patient series appears to exclude a relationship between the time of signal intensity recovery and outcome of CSM.

Association of magnetic resonance imaging signal changes and outcome prediction after surgery for cervical spondylotic myelopathy

2011 ◽  
Vol 15 (6) ◽  
pp. 660-666 ◽  
Author(s):  
Aditya Vedantam ◽  
Ashish Jonathan ◽  
Vedantam Rajshekhar
Keyword(s):  
Magnetic Resonance ◽  
Signal Intensity ◽  
Cervical Spondylotic Myelopathy ◽  
Prognostic Significance ◽  
Magnetic Resonance Images ◽  
Lower Probability ◽  
Mr Images ◽  
T2 Weighted Images

Object Few studies have evaluated the prognostic significance of different types of T2-weighted MR imaging changes in patients with cervical spondylotic myelopathy (CSM). The object of this study was to determine whether the type of increased signal intensity (ISI) was an independent predictor of outcome following central corpectomy in patients with CSM or ossification of the posterior longitudinal ligament (OPLL). Methods Magnetic resonance images obtained in 197 patients who had undergone central corpectomy for CSM or OPLL were assessed for ISI within the cord on sagittal T2-weighted images and hypointensity on T1-weighted images. The T2-weighted changes were categorized as no change (Type 0), fuzzy (Type 1), or sharp (Type 2) based on the ISI characteristics. Outcomes were assessed as a change in Nurick grade of 1 grade or more from preoperatively to postoperatively, and cure as a follow-up Nurick grade of 0 or 1. Multilevel regression analysis was performed to identify predictors of change in Nurick grade ≥ 1 and cure. Results There were 30 patients (15.2%) with Type 0, 104 patients (52.8%) with Type 1, and 63 patients (32%) with Type 2 ISI on MR images. Age, duration of symptoms, and preoperative Nurick grade were similar among the groups. A preoperative Nurick grade of 4 or 5 (OR 0.23, p < 0.001) and presence of Type 2 ISI on T2-weighted images (OR 0.48, p = 0.04) negatively influenced the probability of cure after surgery. Hypointensity on T1-weighted images was only seen in patients who had Type 2 ISI changes. Among the 63 patients with Type 2 ISI, the presence of T1-weighted hypointensity (16 patients) was found to negatively impact cure (OR 0.1, p = 0.04). Conclusions Increased signal intensity on preoperative T2-weighted MR images was seen in more than 80% of the cases. However, only Type 2 ISI on T2-weighted images had a prognostic significance of being associated with a decreased likelihood of cure in patients with CSM or OPLL. Hypointensity on T1-weighted images predicted a lower probability of cure among patients with Type 2 ISI on T2-weighted images.

Automatic segmentation of the temporomandibular joint disc on magnetic resonance images using a deep learning technique

2021 ◽  
pp. 20210185
Author(s):  
Michihito Nozawa ◽  
Hirokazu Ito ◽  
Yoshiko Ariji ◽  
Motoki Fukuda ◽  
Chinami Igarashi ◽  
...  
Keyword(s):  
Deep Learning ◽  
Magnetic Resonance ◽  
Temporomandibular Joint ◽  
Test Data ◽  
Automatic Segmentation ◽  
Learning Model ◽  
Magnetic Resonance Images ◽  
Mr Images ◽  
Validity Test ◽  
Deep Learning Model

Objectives: The aims of the present study were to construct a deep learning model for automatic segmentation of the temporomandibular joint (TMJ) disc on magnetic resonance (MR) images, and to evaluate the performances using the internal and external test data. Methods: In total, 1200 MR images of closed and open mouth positions in patients with temporomandibular disorder (TMD) were collected from two hospitals (Hospitals A and B). The training and validation data comprised 1000 images from Hospital A, which were used to create a segmentation model. The performance was evaluated using 200 images from Hospital A (internal validity test) and 200 images from Hospital B (external validity test). Results: Although the analysis of performance determined with data from Hospital B showed low recall (sensitivity), compared with the performance determined with data from Hospital A, both performances were above 80%. Precision (positive predictive value) was lower when test data from Hospital A were used for the position of anterior disc displacement. According to the intra-articular TMD classification, the proportions of accurately assigned TMJs were higher when using images from Hospital A than when using images from Hospital B. Conclusion: The segmentation deep learning model created in this study may be useful for identifying disc positions on MR images.

Postoperative changes in spinal cord signal intensity in patients with cervical compression myelopathy: comparison between preoperative and postoperative magnetic resonance images

2008 ◽  
Vol 8 (6) ◽  
pp. 524-528 ◽  
Author(s):  
Yasutsugu Yukawa ◽  
Fumihiko Kato ◽  
Keigo Ito ◽  
Yumiko Horie ◽  
Tetsurou Hida ◽  
...  
Keyword(s):  
Spinal Cord ◽  
Magnetic Resonance ◽  
Surgical Outcomes ◽  
Signal Intensity ◽  
Magnetic Resonance Images ◽  
Mr Images ◽  
Japanese Orthopedic Association ◽  
Postoperative Changes ◽  
Cervical Compressive Myelopathy ◽  
Compressive Myelopathy

Object Increased signal intensity of the spinal cord on magnetic resonance (MR) imaging was classified pre- and postoperatively in patients with cervical compressive myelopathy. It was investigated whether postoperative classification and alterations of increased signal intensity could reflect the postoperative severity of symptoms and surgical outcomes. Methods One hundred and four patients with cervical compressive myelopathy were prospectively enrolled. All were treated using cervical expansive laminoplasty. Magnetic resonance imaging was performed in all patients preoperatively and after an average of 39.7 months postoperatively (range 12–90 months). Increased signal intensity of the spinal cord was divided into 3 grades based on sagittal T2-weighted MR images as follows: Grade 0, none; Grade 1, light (obscure); and Grade 2, intense (bright). The severity of myelopathy was evaluated according to the Japanese Orthopedic Association (JOA) score for cervical myelopathy and its recovery rate (100% = full recovery). Results Increased signal intensity was seen in 83% of cases preoperatively and in 70% postoperatively. Preoperatively, there were 18 patients with Grade 0 increased signal intensity, 49 with Grade 1, and 37 with Grade 2; postoperatively, there were 31 with Grade 0, 31 with Grade 1, and 42 with Grade 2. The respective postoperative JOA scores and recovery rates (%) were 13.9/56.7% in patients with postoperative Grade 0, 13.2/50.7% in those with Grade 1, and 12.8/40.1% in those with Grade 2, and these differences were not statistically significant. The postoperative increased signal intensity grade was improved in 16 patients, worsened in 8, and unchanged in 80 (77%). There was no significant correlation between the alterations of increased signal intensity and surgical outcomes. Conclusions The postoperative increased signal intensity classification reflected postoperative symptomatology and surgical outcomes to some extent, without statistically significant differences. The alteration of increased signal intensity was seen postoperatively in 24 patients (23%) and was not correlated with surgical outcome.

scholarly journals Brain Tumor Magnetic Resonance Image Classification: A Deep Learning Approach

2021 ◽  
Author(s):  
Jaya Lakshmi Machiraju ◽  
S. Nagaraja Rao
Keyword(s):  
Deep Learning ◽  
Brain Tumor ◽  
Magnetic Resonance ◽  
Loss Function ◽  
Magnetic Resonance Images ◽  
Training Data ◽  
Learning Approach ◽  
Detection Mechanism ◽  
Validation Data ◽  
Proposed Model

Abstract From the past decade, many researchers are focused on the brain tumor detection mechanism using magnetic resonance images. The traditional approaches follow the feature extraction process from bottom layer in the network. This scenario is not suitable to the medical images. To address this issue, the proposed model employed Inception-v3 convolution neural network model which is a deep learning mechanism. This model extracts the multi-level features and classifies them to find the early detection of brain tumor. The proposed model uses the deep learning approach and hyper parameters. These parameters are optimized using the Adam Optimizer and loss function. The loss function helps the machines to model the algorithm with input data. The softmax classifier is used in the proposed model to classify the images in to multiple classes. It is observed that the accuracy of the Inception-v3 algorithm is recorded as 99.34% in training data and 89% accuracy at validation data.

scholarly journals Application of deep learning in detecting neurological disorders from magnetic resonance images: a survey on the detection of Alzheimer’s disease, Parkinson’s disease and schizophrenia

2020 ◽  
Vol 7 (1) ◽  
Author(s):  
Manan Binth Taj Noor ◽  
Nusrat Zerin Zenia ◽  
M Shamim Kaiser ◽  
Shamim Al Mamun ◽  
Mufti Mahmud
Keyword(s):  
Alzheimer’S Disease ◽  
Parkinson’S Disease ◽  
Alzheimer's Disease ◽  
Parkinson's Disease ◽  
Deep Learning ◽  
Magnetic Resonance ◽  
Neurological Disorders ◽  
Magnetic Resonance Images ◽  
Future Research ◽  
Comparative Performance

Abstract Neuroimaging, in particular magnetic resonance imaging (MRI), has been playing an important role in understanding brain functionalities and its disorders during the last couple of decades. These cutting-edge MRI scans, supported by high-performance computational tools and novel ML techniques, have opened up possibilities to unprecedentedly identify neurological disorders. However, similarities in disease phenotypes make it very difficult to detect such disorders accurately from the acquired neuroimaging data. This article critically examines and compares performances of the existing deep learning (DL)-based methods to detect neurological disorders—focusing on Alzheimer’s disease, Parkinson’s disease and schizophrenia—from MRI data acquired using different modalities including functional and structural MRI. The comparative performance analysis of various DL architectures across different disorders and imaging modalities suggests that the Convolutional Neural Network outperforms other methods in detecting neurological disorders. Towards the end, a number of current research challenges are indicated and some possible future research directions are provided.

scholarly journals Automated Head Tissue Modelling Based on Structural Magnetic Resonance Images for Electroencephalographic Source Reconstruction

2021 ◽  
Author(s):  
Gaia Amaranta Taberna ◽  
Jessica Samogin ◽  
Dante Mantini
Keyword(s):  
Magnetic Resonance ◽  
Brain Structures ◽  
Magnetic Resonance Images ◽  
Head Model ◽  
Source Reconstruction ◽  
Mr Images ◽  
Tissue Segmentation ◽  
Mr Image ◽  
Eeg Recordings ◽  
Tissue Modelling

AbstractIn the last years, technological advancements for the analysis of electroencephalography (EEG) recordings have permitted to investigate neural activity and connectivity in the human brain with unprecedented precision and reliability. A crucial element for accurate EEG source reconstruction is the construction of a realistic head model, incorporating information on electrode positions and head tissue distribution. In this paper, we introduce MR-TIM, a toolbox for head tissue modelling from structural magnetic resonance (MR) images. The toolbox consists of three modules: 1) image pre-processing – the raw MR image is denoised and prepared for further analyses; 2) tissue probability mapping – template tissue probability maps (TPMs) in individual space are generated from the MR image; 3) tissue segmentation – information from all the TPMs is integrated such that each voxel in the MR image is assigned to a specific tissue. MR-TIM generates highly realistic 3D masks, five of which are associated with brain structures (brain and cerebellar grey matter, brain and cerebellar white matter, and brainstem) and the remaining seven with other head tissues (cerebrospinal fluid, spongy and compact bones, eyes, muscle, fat and skin). Our validation, conducted on MR images collected in healthy volunteers and patients as well as an MR template image from an open-source repository, demonstrates that MR-TIM is more accurate than alternative approaches for whole-head tissue segmentation. We hope that MR-TIM, by yielding an increased precision in head modelling, will contribute to a more widespread use of EEG as a brain imaging technique.

scholarly journals Multigrid Nonlocal Gaussian Mixture Model for Segmentation of Brain Tissues in Magnetic Resonance Images

2016 ◽  
Vol 2016 ◽  
pp. 1-10
Author(s):  
Yunjie Chen ◽  
Tianming Zhan ◽  
Ji Zhang ◽  
Hongyuan Wang
Keyword(s):  
Magnetic Resonance ◽  
Gaussian Mixture Model ◽  
Mixture Model ◽  
Gaussian Mixture ◽  
Magnetic Resonance Images ◽  
Intensity Inhomogeneity ◽  
Mr Images ◽  
Proposed Model ◽  
Brain Mr Images ◽  
The Impact

We propose a novel segmentation method based on regional and nonlocal information to overcome the impact of image intensity inhomogeneities and noise in human brain magnetic resonance images. With the consideration of the spatial distribution of different tissues in brain images, our method does not need preestimation or precorrection procedures for intensity inhomogeneities and noise. A nonlocal information based Gaussian mixture model (NGMM) is proposed to reduce the effect of noise. To reduce the effect of intensity inhomogeneity, the multigrid nonlocal Gaussian mixture model (MNGMM) is proposed to segment brain MR images in each nonoverlapping multigrid generated by using a new multigrid generation method. Therefore the proposed model can simultaneously overcome the impact of noise and intensity inhomogeneity and automatically classify 2D and 3D MR data into tissues of white matter, gray matter, and cerebral spinal fluid. To maintain the statistical reliability and spatial continuity of the segmentation, a fusion strategy is adopted to integrate the clustering results from different grid. The experiments on synthetic and clinical brain MR images demonstrate the superior performance of the proposed model comparing with several state-of-the-art algorithms.

Sign in / Sign up

Export Citation Format

Share Document