MRI Imaging Sequences Research Articles

MRI susceptibility artefacts caused by orthodontic wire.

To evaluate magnetic susceptibility artefacts produced by orthodontic wires on MRI and the influence of wire properties and MRI image sequences on the magnitude of the artefact. Arch form orthodontic wires [four stainless steels (SS), one cobalt chromium (CC) alloy, 13 titanium (Ti) alloys] were embedded in a polyester phantom, and scanned using a 1.5-T superconducting magnet scanner with an eight-channel phased-array coil. All wires were scanned with T1-weighted spin echo (SE) and gradient echo (GRE) sequences according to the American Society for Testing and Materials (ASTM) F2119-07 standard. The phantom also scanned other eight sequences. Artefacts were measured using the ASTM F2119-07 definition and OsiriX software. Artefact volume was analysed according to metal composition, wire length, number of wires, wire thickness, and imaging sequence as factors. With SE/GRE, black/white artefacts volumes from all SS wires were significantly larger than those produced by CC and Ti wires (P < .01). With the GRE, the black artefacts volume was the highest with the SS wires. With the SE, the black artefacts volume was small, whereas white artefacts were noticeable. The cranio-caudal extent of the artefacts was significantly longer with SS wires (P < .01). Although a direct relationship of wire length, number of wires, and wire thickness with artefact volume was noted, these factors did not influence artefact extension in the cranio-caudal direction. Ferromagnetic/paramagnetic orthodontic wires create artefacts due to local alteration of magnetic field homogeneity. The SS-type wires produced the largest artefacts followed by CC and Ti.

Construction of multi-scale feature fusion segmentation model of MRI knee images based on dual attention mechanism weighted aggregation.

Early diagnosis of knee osteoarthritis is an important area of research in the field of clinical medicine. Due to the complexity in the MRI imaging sequences and the diverse structure of cartilage, there are many challenges in the segmentation of knee bone and cartilage. Relevant studies have conducted semantic fusion processing through splicing or summing forms, which results in reduced resolution and the accumulation of redundant information. This study was envisaged to construct an MRI image segmentation model to improve the diagnostic efficiency and accuracy of different grade knee osteoarthritis by adopting the Dual Attention and Multi-scale Feature Fusion Segmentation network (DA-MFFSnet). The feature information of different scales was fused through the Multi-scale Attention Downsample module to extract more accurate feature information, and the Global Attention Upsample module weighted lower-level feature information to reduce the loss of key information. The collected MRI knee images were screened and labeled, and the study results showed that the segmentation effect of DA-MFFSNet model was closer to that of the manually labeled images. The mean intersection over union, the dice similarity coefficient and the volumetric overlap error was 92.74%, 91.08% and 7.44%, respectively, and the accuracy of the differential diagnosis of knee osteoarthritis was 84.42%. The model exhibited better stability and classification effect. Our results indicated that the Dual Attention and Multi-scale Feature Fusion Segmentation model can improve the segmentation effect of MRI knee images in mild and medium knee osteoarthritis, thereby offering an important clinical value and improving the accuracy of the clinical diagnosis.

TTMRI: Multislice texture transformer network for undersampled MRI reconstruction

AbstractMagnetic resonance imaging (MRI) is a non‐interposition imaging technique that provides rich anatomical and physiological information. Yet it is limited by the long imaging time. Recently, deep neural networks have shown potential to significantly accelerate MRI. However, most of these approaches ignore the correlation between adjacent slices in MRI image sequences. In addition, the existing deep learning‐based methods for MRI are mainly based on convolutional neural networks (CNNs). They fail to capture long‐distance dependencies due to the small receptive field. Inspired by the feature similarity in adjacent slices and impressive performance of Transformer for exploiting the long‐distance dependencies, a novel multislice texture transformer network is presented for undersampled MRI reconstruction (TTMRI). Specifically, the proposed TTMRI is consisted of four modules, namely the texture extraction, correlation calculation, texture transfer and texture synthesis. It takes three adjacent slices as inputs, in which the middle one is the target image to be reconstructed, and the other two are auxiliary images. The multiscale features are extracted by the texture extraction module and their inter‐dependencies are calculated by the correlation calculation module, respectively. Then the relevant features are transferred by the texture transfer module and fused by the texture synthesis module. By considering inter‐slice correlations and leveraging the Transformer architecture, the joint feature learning across target and adjacent slices are encouraged. Moreover, TTMRI can be stacked with multiple layers to recover more texture information at different levels. Extensive experiments demonstrate that the proposed TTMRI outperforms other state‐of‐the‐art methods in both quantitative and qualitative evaluationsions.

Deep learning-based 3D cerebrovascular segmentation workflow on bright and black blood sequences magnetic resonance angiography

BackgroundCerebrovascular diseases have emerged as significant threats to human life and health. Effectively segmenting brain blood vessels has become a crucial scientific challenge. We aimed to develop a fully automated deep learning workflow that achieves accurate 3D segmentation of cerebral blood vessels by incorporating classic convolutional neural networks (CNNs) and transformer models.MethodsWe used a public cerebrovascular segmentation dataset (CSD) containing 45 volumes of 1.5 T time-of-flight magnetic resonance angiography images. We collected data from another private middle cerebral artery (MCA) with lenticulostriate artery (LSA) segmentation dataset (MLD), which encompassed 3.0 T three-dimensional T1-weighted sequences of volumetric isotropic turbo spin echo acquisition MRI images of 107 patients aged 62 ± 11 years (42 females). The workflow includes data analysis, preprocessing, augmentation, model training with validation, and postprocessing techniques. Brain vessels were segmented using the U-Net, V-Net, UNETR, and SwinUNETR models. The model performances were evaluated using the dice similarity coefficient (DSC), average surface distance (ASD), precision (PRE), sensitivity (SEN), and specificity (SPE).ResultsDuring 4-fold cross-validation, SwinUNETR obtained the highest DSC in each fold. On the CSD test set, SwinUNETR achieved the best DSC (0.853), PRE (0.848), SEN (0.860), and SPE (0.9996), while V-Net achieved the best ASD (0.99). On the MLD test set, SwinUNETR demonstrated good MCA segmentation performance and had the best DSC, ASD, PRE, and SPE for segmenting the LSA.ConclusionsThe workflow demonstrated excellent performance on different sequences of MRI images for vessels of varying sizes. This method allows doctors to visualize cerebrovascular structures.Critical relevance statementA deep learning-based 3D cerebrovascular segmentation workflow is feasible and promising for visualizing cerebrovascular structures and monitoring cerebral small vessels, such as lenticulostriate arteries.Key points• The proposed deep learning-based workflow performs well in cerebrovascular segmentation tasks.• Among comparison models, SwinUNETR achieved the best DSC, ASD, PRE, and SPE values in lenticulostriate artery segmentation.• The proposed workflow can be used for different MR sequences, such as bright and black blood imaging.Graphical

Altered structural and functional connectivity in Posterior Cortical Atrophy and Dementia with Lewy bodies

Posterior cortical atrophy (PCA) and dementia with Lewy bodies (DLB) show distinct atrophy and overlapping hypometabolism profiles, but it is unknown how disruptions in structural and functional connectivity compare between these disorders and whether breakdowns in connectivity relate to either atrophy or hypometabolism.Thirty amyloid-positive PCA patients, 24 amyloid-negative DLB patients and 30 amyloid-negative cognitively unimpaired (CU) healthy individuals were recruited at Mayo Clinic, Rochester, MN, and underwent a 3T head MRI, including structural MRI, resting state functional MRI (rsfMRI) and diffusion tensor imaging (DTI) sequences, as well as [18F] fluorodeoxyglucose (FDG) PET. We assessed functional connectivity within and between 12 brain networks using rsfMRI and the CONN functional connectivity toolbox and calculated regional DTI metrics using the Johns Hopkins atlas. Multivariate linear-regression models corrected for multiple comparisons and adjusted for age and sex compared DTI metrics and within-network and between-network functional connectivity across groups. Regional gray-matter volumes and FDG-PET standard uptake value ratios (SUVRs) were calculated and analyzed at the voxel-level using SPM12. We used univariate linear-regression models to investigate the relationship between connectivity measures, gray-matter volume, and FDG-PET SUVR.On DTI, PCA showed degeneration in occipito-parietal white matter, posterior thalamic radiations, splenium of the corpus collosum and sagittal stratum compared to DLB and CU, with greater degeneration in the temporal white matter and the fornix compared to CU. We observed no white-matter degeneration in DLB compared to CU. On rsfMRI, reduced within-network connectivity was present in dorsal and ventral default mode networks (DMN) and the dorsal-attention network in PCA compared to DLB and CU, with reduced within-network connectivity in the visual and sensorimotor networks compared to CU. DLB showed reduced connectivity in the cerebellar network compared to CU. Between-network analysis showed increased connectivity in both cerebellar-to-sensorimotor and cerebellar-to-dorsal attention network connectivity in PCA and DLB. PCA showed reduced anterior DMN-to-cerebellar and dorsal attention-to-sensorimotor connectivity, while DLB showed reduced posterior DMN-to-sensorimotor connectivity compared to CU. PCA showed reduced dorsal DMN-to-visual connectivity compared to DLB. The multimodal analysis revealed weak associations between functional connectivity and volume in PCA, and between functional connectivity and metabolism in DLB.These findings suggest that PCA and DLB have unique connectivity alterations, with PCA showing more widespread disruptions in both structural and functional connectivity; yet some overlap was observed with both disorders showing increased connectivity from the cerebellum.

An Automated Prostate-cancer Prediction System (APPS) Based on Advanced DFO-ConGA2L Model using MRI Imaging Technique

The prostate cancer is a deadly form of cancer that assassinates a significant number of men due of its mediocre identification process. Images from people with cancer include important and intricate details that are difficult for conventional diagnostic methods to extract. This work establishes a novel Automated Prostate-cancer Prediction System (APPS) model for the goal of detecting and classifying prostate cancer utilizing MRI imaging sequences. The supplied medical image is normalized using a Coherence Diffusion Filtering (CDFilter) approach for improved quality and contrast. The appropriate properties are also extracted from the normalized image using the morphological and texture feature extraction approach, which helps to increase the classifier's accuracy. In order to train the classifier, the most important properties are also selected utilizing the cutting-edge Dragon Fly Optimized Feature Selection (DFO-FS) algorithm. Using this method greatly improves the classifier's overall disease diagnosis performance in less time and with faster processing. More specifically, the provided MRI input data are used to categorize the prostate cancer-affected and healthy tissues using the new Convoluted Gated Axial Attention Learning Model (ConGA2L) based on the selected features. This study compares and validates the performance of the APPS model by looking at several aspects using publicly available prostate cancer data.

McTFI QSM MRI ABC/2 intracranial hemorrhage to noncontrast head CT volume measurement equivalence

Background/objectivesIntracranial hemorrhage (ICH) volume assessment is an important part of patient management and is routinely obtained by non-contrast head CT (NCHCT) using the validated ABC/2 measurement method. Because conventional MRI imaging sequences demonstrate variability in ICH appearance, volumetric analyses for MRI bleed volume in a standardized manner using ABC/2 is not possible. The recently introduced multiecho-complex total field inversion quantitative susceptibility mapping (mcTFI QSM) MRI technique, which maps brain tissue susceptibility to both depict brain tissue structures and quantify tissue susceptibility, may provide a viable alternative. In this study we evaluated mcTFI QSM ABC/2 ICH volume assessment relative to NCHCT. MethodsPatients with ICH who had undergone NCHCT and MRI brain scans within 48 h were recruited for this retrospective study. The ABC/2 method was applied to estimate the bleed volume for both NCHCT and MRI by a CAQ-certified neuroradiologist with 10 years of experience and a trained laboratory assistant. Results were analyzed via Bland-Altman (B-A) and linear regression. Results54 patients (27 females) who had undergone NCHCT and MRI within 48 h (<24 h., n = 31, 24-48 h, n = 10) were enrolled. mcTFI QSM ICH volume measurement method showed a positive correlation (99.5%) compared to NCHCT. B-A plot comparing ABC/2 ICH volume on NCHCT and mcTFI MRI done for patients within 24 h demonstrates a bias of −0.09%. ConclusionsICH volume calculation using ABC/2 on mcTFI QSM showed a high correlation with NCHCT measurement. These results suggest mcTFI QSM is a promising MRI method for ABC/2 for bleed volume measurement.

Dual-domain accelerated MRI reconstruction using transformers with learning-based undersampling.

Acceleration in MRI has garnered much attention from the deep-learning community in recent years, particularly for imaging large anatomical volumes such as the abdomen or moving targets such as the heart. A variety of deep learning approaches have been investigated, with most existing works using convolutional neural network (CNN)-based architectures as the reconstruction backbone, paired with fixed, rather than learned, k-space undersampling patterns. In both image domain and k-space, CNN-based architectures may not be optimal for reconstruction due to its limited ability to capture long-range dependencies. Furthermore, fixed undersampling patterns, despite ease of implementation, may not lead to optimal reconstruction. Lastly, few deep learning models to date have leveraged temporal correlation across dynamic MRI data to improve reconstruction. To address these gaps, we present a dual-domain (image and k-space), transformer-based reconstruction network, paired with learning-based undersampling that accepts temporally correlated sequences of MRI images for dynamic reconstruction. We call our model DuDReTLU-net. We train the network end-to-end against fully sampled ground truth dataset. Human cardiac CINE images undersampled at different factors (5-100) were tested. Reconstructed images were assessed both visually and quantitatively via the structural similarity index, mean squared error, and peak signal-to-noise. Experimental results show superior performance of DuDReTLU-net over state-of-the-art methods (LOUPE, k-t SLR, BM3D-MRI) in accelerated MRI reconstruction; ablation studies show that transformer-based reconstruction outperformed CNN-based reconstruction in both image domain and k-space; dual-domain reconstruction architectures outperformed single-domain reconstruction architectures regardless of reconstruction backbone (CNN or transformer); and dynamic sequence input leads to more accurate reconstructions than single frame input. We expect our results to encourage further research in the use of dual-domain architectures, transformer-based architectures, and learning-based undersampling, in the setting of accelerated MRI reconstruction. The code for this project is made freely available at https://github.com/william2343/dual-domain-mri-recon-nets (Hong et al., 2022).

Echocardiographic correlates of MRI imaging markers of cerebral small-vessel disease in patients with atrial-fibrillation-related ischemic stroke.

Atrial fibrillation (AF) has been linked to dementia risk, partly explained by cerebral small vessel disease (CSVD). Since AF and cardiovascular comorbidities were associated with cardiac dysfunction, we aimed to determine the association between echocardiographic parameters and neuroimaging markers of CSVD in patients with AF-related ischemic stroke. This cross-sectional study enrolled patients with AF-related ischemic stroke from March 2013 to December 2019 who underwent transthoracic echocardiography and brain 3T MRI, including T1, T2, Flair, and SWI imaging sequences. We assessed the presence of lacunes and cerebellar microbleeds (CMBs), the severity of white matter hyperintensity (WMH) scored by the Fazekas scale (0-6), and the severity of enlarged perivascular spaces (EPVS) in basal ganglia (BG) and centrum semiovale (CSO) classified into three categories (0-10, 10-25, and >25). CSVD burden was rated on a 0-to-4 ordinal scale. Generalized linear regression analysis and post hoc comparisons with Bonferroni correction were performed to assess the association between various echocardiographic parameters and these lesions, adjusted for demographics and potential confounders. 119 patients (68.38 ± 12.692 years; male 45.4 %) were included for analysis, of whom 55 (46.2%) had lacunes, 40 (33.6%) had CMBs, and median severity for WMH, BG-EPVS, CSO-EPVS, and CSVD burden were 2 (IQR: 1-3), 1 (IQR: 1-2), 1 (IQR: 0-1), and 1 (IQR: 1-2) respectively. In multivariable, fully adjusted models, left ventricular posterior wall thickness (LVPW) was associated with a higher risk of lacunes (RR 1.899, 95% CI: 1.342-2.686) and CSVD burden (RR = 2.081, 95%CI: 1.562-2.070). Right atrial diameter (RAD) was associated with greater CSO-EPVS (RR = 2.243, 95%CI: 1.234-4.075). No echocardiographic parameters were revealed to be associated with CMBs and WMH. In patients with AF-related ischemic stroke, LVPW is associated with a higher risk of lacunes and CSVD burden, while RAD was associated with greater CSO-EPVS. Larger studies are required to determine these associations and to elucidate if these associations can help facilitate cognitive evaluation and brain MRI screening.

P03.11.A Potential role of pre-radiotherapy MRI for target delineation in high-grade gliomas: a multicenter retro-prospective cohort study

Abstract Background The optimal timing for target identification in high-grade glioma (HGG) remains unclear due to variability in the hyper-signal T2/FLAIR between MRI performed at diagnosis, post-surgery and at radiotherapy (RT) start. The aim of this study was to retrospectively confirm that RT planned on delayed MRI might allow to spare more normal tissue without decreasing local tumour control, in order to prospectively evaluate the best standard and advanced MRI and metabolic imaging sequences for clinical tumor volume (CTV) adaptation. Material and Methods We analyzed a retrospective cohort of consecutive patients with HGG treated from 2017 to 2020. All patients had a diagnostic MRI and another performed immediately post-surgery or pre-RT. Target volumes were contoured, based on T2/FLAIR, on diagnostic and post-surgery MRI in group A, while in group B on pre-RT MRI. We analyzed GTV and CTV volume, and the percentage increase between them. Moreover, we compared the two groups in terms of clinical-pathological characteristics and progression-free survival (PFS) and overall survival (OS). A prospective study, started on January 2022, has enrolled patients with HGG evaluated by advanced sequences MRI at diagnosis, post-surgery and pre-RT. In addition, some selected patients have undergone diagnostic DOPA-PET and pre-RT DOPA-PET. 2 MRI-guided contours have been performed for each patient: adapted on T2/FLAIR post-surgery and CTV-adapt on pre-RT, to assess study objectives. Results In retrospective cohort we analyzed 54 patients (25 group A, 29 group B). The median age of patients was 61 years (IQR 17,75), 93% had an ECOG PS of 0 or 1, 51 were symptomatic at diagnosis. Patients in group B had more frequently MGMT methylation (59 % vs. 28%, p=0.01) while less frequently frontal lobe involvement (60% vs. 24%, p=0.01). The median percentage increase between GTV and CTV was higher in group A than B: 431% (range 62%-7335%) vs 385% (range 53%-3174%), respectively. No significant difference in the pattern of relapse was observed, since &amp;gt;90% of disease recurrences were in-field in both groups. Median PFS and OS of the overall population were 9.5 months (95% CI 7 - 12) and 18.5 months (95% CI 16 - 24), respectively. Patients in group B had a significant better PFS as compared to those in group A (p=0.03), but similar OS. Nevertheless, imbalance in MTMT methylation status between the two groups was a major driver for PFS. Overall, 37 out of 51 patients had improvement in neurological symptoms (p&amp;lt;0.001), with no difference between the two groups (p=0.54). Conclusion Our data suggest that CTV adaptation to pre-RT T2/FLAIR may allow reducing RT volume, without affecting symptoms relieving and disease control. Results from the prospective study will help identifying the best adaptation of CTV guided by T2/FLAIR, advanced MRI sequences and metabolic imaging, in order to optimize efficacy and safety of treatment planning.

Cerebral microbleeds is associated with dementia in Parkinson's disease.

Emerging evidence suggests that cerebral small vessel disease (CSVD) may worsen cognitive functions in Parkinson's disease (PD). However, the effect of microbleeds on cognitive function in patients with PD remains unknown. This study explored the association between the presence, number and location of microbleeds with dementia in PD patients. This cross-sectional study included 431 patients with PD from Beijing Tiantan Hospital from May 2016 to August 2019. Cognition assessments (MMSE, MoCA) were performed for these patients. MRI imaging sequences were obtained and reviewed independently by two well-trained readers who were blind to all clinical data. Spearman's correlation analysis and logistic regression model analysis were further used for the assessments. An association between cerebral microbleeds with cognitive ability and dementia in PD patients was revealed. A significance was observed between the total number of microbleeds and two widely used scores of cognitive assessments (Spearman R = - 0.120 to MMSE with a p = 0.016, and - 0.117 to MoCA with a p = 0.020). In detail, infratentorial microbleeds were associated with the level of cognition in PD patients (Spearman R = - 0.099 to MMSE with a p = 0.049, and - 0.116 to MoCA with a p = 0.021). Furthermore, logistic regression analysis results also confirmed such correlations between the number of microbleeds and cognitive ability after adjusting for age, cholesterol level, Hamilton Anxiety Scale, Hamilton Depression Scale, and white matter hyperintensity Fazekas score (OR 3.28, p = 0.035, 95% CI 1.090-9.892). The occurrence of microbleeds, especially in the infratentorial locations, may worsen the cognitive function of PD patients and result in dementia. Management of cerebral vascular disease could be beneficial to patients with PD.

Proton and Photon Dosimetry of Carbon Fiber-PEEK Implants for Paraspinal Stereotactic Body Radiation Therapy

<h3>Purpose/Objective(s)</h3> Carbon fiber composite (CFR-PEEK) implants have been shown to improve image quality in volumetric imaging compared with titanium (Ti). We perform a dosimetric comparison of proton and photon SBRT plans with Ti and CFR-PEEK implants for paraspinal SBRT treatments. <h3>Materials/Methods</h3> CT scans of a spine phantom embedded with CFR-PEEK and Ti implants were obtained. PTV and OAR contours for a 24 Gy single fraction plan were generated with d_max spinal cord < 14 Gy. Nine-field IMRT photon plans were optimized on both scans. Three-field pencil beam scanning (PBS) plans using multi-field optimization (MFO) were generated. Robust evaluation for 2 mm setup, ± 3.5% range and ± 10% material density override uncertainties were performed. To evaluate the MR image quality of the carbon fiber implants, an implant set was placed in a water phantom. Spine MRI imaging sequences were acquired at 1.5T and 3T and associated artifacts were evaluated. <h3>Results</h3> Target V95 coverage statistics for all plans are shown in Table 1. Photon plans with CFR-PEEK vs. Ti implants were not significantly different. Target coverage goals and OAR constraints were met. When the CFR-PEEK optimized photon plan was recalculated on Ti CT, PTV V100% was reduced by 12.4%, demonstrating the dose perturbation induced by Ti implants. The MFO proton plans for both implant scenarios had lower coverage than the IMRT plans due to the sharper gradients on the IMRT plans. The PBS plan for CFR-PEEK had slightly better coverage than for Ti. Cord-constraint-free proton plans were generated to investigate how the cord constraint impacted the proton plan. Without the constraint, PTV V95% was 99.0% for CFR-PEEK and 94.9% for Ti plans. The increased coverage for CFR-PEEK plan suggests that the potential advantage of CFR-PEEK is masked by the tight cord constraints. In robustness analysis, no clear benefit was shown for range and setup uncertainties. However, advantage in target coverage robustness was seen under material density perturbation analysis. Specifically, the perturbed PTV V95% ranged from 88.7% to 88.4%. In contrast, for Ti plan, PTV V95% varied from 86.5% to 82.3% in perturbed scenarios. In MRI studies, no metal signal loss image artifacts were observed around the CFR-PEEK screws at either field strengths. Minor signal pile-up artifacts of < 5 mm can be seen on one side of the screw in high field 3 T studies but were negligible in 1.5 T studies. <h3>Conclusion</h3> For paraspinal SBRT, utilizing CFR-PEEK compared to Ti implants do not significantly affect plan quality for both PBS and IMRT. Proton plans are more robust for material composition uncertainties only. The largest advantage of CFR-PEEK vs. Ti will be in MR image quality improvements, which can translate to more accurate target delineations and improved follow-up image quality.

Multiparametric MRI-based radiomics analysis for the prediction of breast tumor regression patterns after neoadjuvant chemotherapy.

ObjectivesBreast cancers show different regression patterns after neoadjuvant chemotherapy. Certain regression patterns are associated with more reliable margins in breast-conserving surgery. Our study aims to establish a nomogram based on radiomic features and clinicopathological factors to predict regression patterns in breast cancer patients.MethodsWe retrospectively reviewed 144 breast cancer patients who received neoadjuvant chemotherapy and underwent definitive surgery in our center from January 2016 to December 2019. Tumor regression patterns were categorized as type 1 (concentric regression + pCR) and type 2 (multifocal residues + SD + PD) based on pathological results. We extracted 1158 multidimensional features from 2 sequences of MRI images. After feature selection, machine learning was applied to construct a radiomic signature. Clinical characteristics were selected by backward stepwise selection. The combined prediction model was built based on both the radiomic signature and clinical factors. The predictive performance of the combined prediction model was evaluated.ResultsTwo radiomic features were selected for constructing the radiomic signature. Combined with two significant clinical characteristics, the combined prediction model showed excellent prediction performance, with an area under the receiver operating characteristic curve of 0.902 (95% confidence interval 0.8343–0.9701) in the primary cohort and 0.826 (95% confidence interval 0.6774–0.9753) in the validation cohort.ConclusionsOur study established a unique model combining a radiomic signature and clinicopathological factors to predict tumor regression patterns prior to the initiation of NAC. The early prediction of type 2 regression offers the opportunity to modify preoperative treatments or aids in determining surgical options.

Ensemble Learning for Early‐Response Prediction of Antidepressant Treatment in Major Depressive Disorder

In order to reduce unsuccessful treatment trials for depression, neuroimaging and genetic information can be considered as biomarkers. Together with machine-learning methods, prediction models have proved to be valuable for baseline prediction. To propose an ensemble learning modeling framework that integrates imaging and genetic information for individualized baseline prediction of early-stage treatment response of antidepressants in major depressive disorder (MDD). Prospective. In all, 98 inpatients with MDD. 3.0T MRI and gradient-echo echo-planar imaging sequence. Participants were divided into responders and nonresponders based on reducing rates of HDRS-6 after early-stage treatment of 2 weeks. Fourteen brain regions of interest were selected according to previous studies. An ensemble learning modeling framework was used to integrate imaging data and genetic data. Support vector machine (SVM) with linear kernel was utilized to integrate multimode information and then to construct the prediction model. Leave-one-out cross-validation (LOOCV) was used to evaluate the performance. The position characteristics obtained through SVM-RFE (recursive feature elimination) algorithm and LOOCV was considered to compare each feature's relative importance for the prediction model. Compared with the single-level prediction model, the ensemble learning prediction model showed improvement in prediction performance (accuracy from 0.61 to 0.86 with imaging data and genetic data). Integrated with 14 priori brain regions, the region of interest (ROI) map ensemble learning prediction model can achieve a performance that is analogous with the model with information from whole-brain regions (both with accuracy of 0.81). The integration of genetic features further improved the sensitivity of prediction (sensitivity from 0.78 to 0.87 under the ensemble learning framework). Our ensemble learning prediction model demonstrated significant advantages in interpretability and information integration. The findings may provide more assistance for clinical treatment selection in MDD at the individual level. 1 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2020;52:161-171.

Chronic inflammation in multiple sclerosis - seeing what was always there.

Activation of innate immune cells and other compartmentalized inflammatory cells in the brains and spinal cords of people with relapsing-remitting multiple sclerosis (MS) and progressive MS has been well described histopathologically. However, conventional clinical MRI is largely insensitive to this inflammatory activity. The past two decades have seen the introduction of quantitative dynamic MRI scanning with contrast agents that are sensitive to the reduction in blood-brain barrier integrity associated with inflammation and to the trafficking of inflammatory myeloid cells. New MRI imaging sequences provide improved contrast for better detection of grey matter lesions. Quantitative lesion volume measures and magnetic resonance susceptibility imaging are sensitive to the activity of macrophages in the rims of white matter lesions. PET and magnetic resonance spectroscopy methods can also be used to detect contributions from innate immune activation in the brain and spinal cord. Some of these advanced research imaging methods for visualization of chronic inflammation are practical for relatively routine clinical applications. Observations made with the use of these techniques suggest ways of stratifying patients with MS to improve their care. The imaging methods also provide new tools to support the development of therapies for chronic inflammation in MS.

Explainable cardiac pathology classification on cine MRI with motion characterization by semi-supervised learning of apparent flow.

We propose a method to classify cardiac pathology based on a novel approach to extract image derived features to characterize the shape and motion of the heart. An original semi-supervised learning procedure, which makes efficient use of a large amount of non-segmented images and a small amount of images segmented manually by experts, is developed to generate pixel-wise apparent flow between two time points of a 2D+t cine MRI image sequence. Combining the apparent flow maps and cardiac segmentation masks, we obtain a local apparent flow corresponding to the 2D motion of myocardium and ventricular cavities. This leads to the generation of time series of the radius and thickness of myocardial segments to represent cardiac motion. These time series of motion features are reliable and explainable characteristics of pathological cardiac motion. Furthermore, they are combined with shape-related features to classify cardiac pathologies. Using only nine feature values as input, we propose an explainable, simple and flexible model for pathology classification. On ACDC training set and testing set, the model achieves 95% and 94% respectively as classification accuracy. Its performance is hence comparable to that of the state-of-the-art. Comparison with various other models is performed to outline some advantages of our model.

Integration of Neural Reward Processing and Appetite-Related Signaling in Obese Females: Evidence From Resting-State fMRI.

The reward-related regions have been considered a crucial component in the regulation of eating behavior. Furthermore, appetite-related regions associated with reward can influence eating behaviors through altered functional activity related to food in brain areas associated with emotion, memory, sensory processing, motor function, and cognitive control. To investigate the key nodes in obese females of reward-related regions and, based on key nodes, to evaluate the directionality of functional connectivity between key nodes and appetite-related regions. Prospective. Twenty-eight obese and 28 normal-weight female controls of similar age. 3.0 T MRI and echo planar imaging (EPI) sequence, 3D BRAVO sequence. The fMRI data preprocessing was based on the Data Processing & Analysis of Brain Imaging and Statistical Parametric Mapping 12. Degree centrality calculation was based on the GRETNA toolkit and granger causality analysis were based on the DynamicBC toolbox. Statistical Tests: Independent two-sample t-tests were used to assess the differences in demographic and clinical data between two groups. Two-sample t-tests were conducted to test the difference in degree centrality and effective connectivity of key nodes between two groups. Compared with normal-weight controls, obese females showed an increased degree centrality in the left ventral striatum/caudate (t = 2.96808, P < 0.05) and decreased degree centrality in right orbitofrontal cortex (OFC) (t = -3.3558, P < 0.05). The obese females showed directional effective connectivity between left ventral striatum/caudate and several regions (left inferior temporal gyrus, fusiform gyrus, postcentral gyrus, and right precentral gyrus) (P < 0.05). Directional effective connectivity was also observed between the right OFC and several regions (left middle temporal gyrus, cuneus, OFC, superior temporal gyrus, middle frontal gyrus, and right inferior parietal lobule) (P < 0.05). The left ventral striatum/caudate and right OFC are key nodes in reward-related regions. The key nodes with reward processing mainly enhance visual processing of information and further participate in cognitive, attention, and sensorimotor processing. 1. Technical Efficacy: Stage 4. J. Magn. Reson. Imaging 2019;50:541-551.

Multimodal Radiomic Features for the Predicting Gleason Score of Prostate Cancer.

Background: Novel radiomic features are enabling the extraction of biological data from routine sequences of MRI images. This study’s purpose was to establish a new model, based on the joint intensity matrix (JIM), to predict the Gleason score (GS) of prostate cancer (PCa) patients. Methods: A retrospective dataset comprised of the diagnostic imaging data of 99 PCa patients was used, extracted from The Cancer Imaging Archive’s (TCIA) T2-Weighted (T2-WI) and apparent diffusion coefficient (ADC) images. Radiomic features derived from JIM and the grey level co-occurrence matrix (GLCM) were extracted from the reported tumor locations. The Kruskal-Wallis test and Spearman’s rank correlation identified features related to the GS. The Random Forest classifier model was implemented to identify the best performing signature of JIM and GLCM radiomic features to predict for GS. Results: Five JIM-derived features: contrast, homogeneity, difference variance, dissimilarity, and inverse difference were independent predictors of GS (p < 0.05). Combined JIM and GLCM analysis provided the best performing area-under-the-curve, with values of 78.40% for GS ≤ 6, 82.35% for GS = 3 + 4, and 64.76% for GS ≥ 4 + 3. Conclusion: This retrospective study produced a novel predictive model for GS by the incorporation of JIM data from standard diagnostic MRI images.

A deep Boltzmann machine-driven level set method for heart motion tracking using cine MRI images.

Heart motion tracking for radiation therapy treatment planning can result in effective motion management strategies to minimize radiation-induced cardiotoxicity. However, automatic heart motion tracking is challenging due to factors that include the complex spatial relationship between the heart and its neighboring structures, dynamic changes in heart shape, and limited image contrast, resolution, and volume coverage. In this study, we developed and evaluated a deep generative shape model-driven level set method to address these challenges. The proposed heart motion tracking method makes use of a heart shape model that characterizes the statistical variations in heart shapes present in a training data set. This heart shape model was established by training a three-layered deep Boltzmann machine (DBM) in order to characterize both local and global heart shape variations. During the tracking phase, a distance regularized level-set evolution (DRLSE) method was applied to delineate the heart contour on each frame of a cine MRI image sequence. The trained shape model was embedded into the DRLSE method as a shape prior term to constrain an evolutional shape to reach the desired heart boundary. Frame-by-frame heart motion tracking was achieved by iteratively mapping the obtained heart contour for each frame to the next frame as a reliable initialization, and performing a level-set evolution. The performance of the proposed motion tracking method was demonstrated using thirty-eight coronal cine MRI image sequences.

Quantitative analysis of brain perfusion parameters in dogs with idiopathic epilepsy by use of magnetic resonance imaging.

OBJECTIVE To quantitatively analyze brain perfusion parameters in dogs with idiopathic epilepsy (IE) by use of MRI and to compare those findings with brain perfusion parameters for healthy dogs. ANIMALS 12 client-owned dogs with IE. PROCEDURES For each dog, standard MRI and perfusion-weighted imaging (before and after injection of gadoteric acid contrast medium) sequences of the brain were obtained during the interictal period by means of the same protocol used in a comparable study of healthy dogs. Time of contrast medium arrival, time to peak contrast enhancement, mean contrast transit time, and cerebral blood flow were calculated for the caudate nucleus, thalamus, piriform lobe, hippocampus, semioval center, and temporal cerebral cortex. Parameters for each structure were compared between dogs with IE and healthy dogs. RESULTS Dogs with IE had a significantly greater mean time of contrast arrival and lower mean cerebral blood flow than healthy dogs. Differences in cerebral blood flow between dogs with IE and healthy dogs were most pronounced in the piriform lobe, thalamus, and temporal cerebral cortex. The mean contrast transit time did not differ between dogs with IE and healthy dogs. CONCLUSIONS AND CLINICAL RELEVANCE Results suggested that, compared with healthy dogs, dogs with IE have decreased blood perfusion of the brain. Findings of this study can be used as a basis for further research into functional changes within the brains of epileptic dogs during the interictal phase.