Multidimensional Dual Encoding Network For Liver Lesion Classification From Multi-Phase Magnetic Resonance Imaging.

Deep learning-based diagnosis systems are useful to identify abnormalities in medical images with the greatly increased workload of doctors. Specifically, the rate of new cases and deaths from malignancies is rising for liver diseases. Early detection of liver lesions plays an extremely important role in effective treatment and gives a higher chance of survival for patients. Therefore, automatic detection and classification of common liver lesions are essential for doctors. In fact, radiologists mainly rely on Hounsfield Units to locate liver lesions but previous studies often pay little attention to this factor. In this paper, we propose an improved method for the automatic classification of common liver lesions based on deep learning techniques and the variation of Hounsfield Unit densities on CT images with and without contrast. Hounsfield Unit is used to locate liver lesions accurately and support data labeling for classification. We construct a multi-phase classification model developed on the deep neural networks of Faster R-CNN, R-FCN, SSD, and Mask R-CNN with the transfer learning approach. The experiments are conducted on six scenarios with multi-phase CT images of common liver lesions. Experimental results show that the proposed method improves the detection and classification of liver lesions compared with recent methods because its accuracy achieves up to 97.4%. The proposed models are very useful to assist doctors in the automatic segmentation and classification of liver lesions to solve the problem of depending on the clinician's experience in the diagnosis and treatment of liver lesions.

Recurrent neural network-based sequence-to-sequence attentional models have proven effective in abstractive text summarization. In this paper, we model abstractive text summarization using a dual encoding model. Different from the previous works only using a single encoder, the proposed method employs a dual encoder including the primary and the secondary encoders. Specifically, the primary encoder conducts coarse encoding in a regular way, while the secondary encoder models the importance of words and generates more fine encoding based on the input raw text and the previously generated output text summarization. The two level encodings are combined and fed into the decoder to generate more diverse summary that can decrease repetition phenomenon for long sequence generation. The experimental results on two challenging datasets (i.e., CNN/DailyMail and DUC 2004) demonstrate that our dual encoding model performs against existing methods.

The ability to register image data to a common coordinate system is a critical feature of virtually all imaging studies. However, in spite of the abundance of literature on the subject and the existence of several variants of registration algorithms, their practical utility remains problematic, as commonly acknowledged even by developers of these methods. A new registration method is presented that utilizes a Hamiltonian formalism and constructs registration as a sequence of symplectomorphic maps in conjunction with a novel phase space regularization. For validation of the framework a panel of deformations expressed in analytical form is developed that includes deformations based on known physical processes in MRI and reproduces various distortions and artifacts typically present in images collected using these different MRI modalities. The method is demonstrated on the three different magnetic resonance imaging (MRI) modalities by mapping between high resolution anatomical (HRA) volumes, medium resolution diffusion weighted MRI (DW-MRI) and HRA volumes, and low resolution functional MRI (fMRI) and HRA volumes. The method has shown an excellent performance and the panel of deformations was instrumental to quantify its repeatability and reproducibility in comparison to several available alternative approaches.

Background Liver fibrosis and cirrhosis require accurate, noninvasive diagnostic methods. Deep learning-enhanced magnetic resonance imaging (MRI) modalities, including T2-weighted imaging, gadoxetic acid-enhanced MRI, and magnetic resonance elastography (MRE), have shown promise in improving diagnostic accuracy. Objective To systematically review and evaluate the diagnostic accuracy of deep learning-enhanced MRI modalities for liver fibrosis and cirrhosis detection, using liver biopsy as the reference standard, and to perform a meta-analysis of the diagnostic accuracy. Methods A systematic search was conducted across multiple databases (PubMed, Cochrane Library, Embase, Scopus, Google Scholar, and IEEE Xplore). Studies reporting diagnostic accuracy metrics (sensitivity, specificity, area under the receiver operating characteristic (AUROC)) for deep learning-based MRI modalities in adults with chronic liver disease were included. A meta-analysis was performed for the studies providing sufficient data. Results Seven studies with 6,547 participants were included. MRE-based approaches exhibited the highest diagnostic accuracy (AUROC: 0.759–0.93). A meta-analysis of three studies with liver biopsy as the reference standard showed high sensitivity (0.80–0.91) and specificity (0.79–0.90), but model non-convergence prevented pooled estimates. Conclusion Deep learning-enhanced MRI, particularly MRE, shows promise for noninvasive detection of liver fibrosis and cirrhosis. Despite promising individual results, further validation, standardized protocols, and larger studies are needed to confirm its clinical utility.

Magnetic Resonance Imaging (MRI) biomarkers of multiple sclerosis (MS), particularly fluid-attenuated inversion-recovery (FLAIR) sequences, have long been investigated. However, advanced analytical methods, capable of fusing results from different MRI modalities, could be more informative to have enabled joint biomarkers. Here we estimate disease duration (DD) in MS subjects (n=47) based on fusion of information from Myelin Water Imaging (MWI), Diffusion Tensor Imaging (DTI), and resting state functional MRI (rsfMRI) modalities, by adapting a joint Multimodal Statistical Analysis Framework. Using this data driven, multimodal, latent variable (LV) approach, common and unique information in each dataset was acquired and their relationship with DD is analyzed through the Least Absolute Shrinkage and Selection Operator (LASSO) regression. The common components between the three modalities, but not the unique components of each modality, accurately predicted DD. To further investigate the regions importance for estimating DD, we separated the data into two groups: “early” and “late” DD depending upon their relationship to the median duration of illness (120 months). In early disease, DTI information in the Right Tapetum and the Fornix jointly with MWI information in the Left Superior Cerebellar Penducle, Right Cerebral Penducle, and Cingulum were most informative. In contrast, rsfMRI demonstrated altered connectivity throughout disease duration. Our results demonstrate the power of multimodal imaging markers in MS.

Alzheimer's disease (AD) is the most common cause of dementia. Neuropathological changes in AD patients occur up to 10–20 years before the emergence of clinical symptoms. Specific diagnosis and appropriate intervention strategies are crucial during the phase of mild cognitive impairment (MCI) and AD. The detection of biomarkers has emerged as a promising tool for tracking the efficacy of potential therapies, making an early disease diagnosis, and prejudging treatment prognosis. Specifically, multiple neuroimaging modalities, including magnetic resonance imaging (MRI), positron emission tomography, optical imaging, and single photon emission-computed tomography, have provided a few potential biomarkers for clinical application. The MRI modalities described in this review include structural MRI, functional MRI, diffusion tensor imaging, magnetic resonance spectroscopy, and arterial spin labelling. These techniques allow the detection of presymptomatic diagnostic biomarkers in the brains of cognitively normal elderly people and might also be used to monitor AD disease progression after the onset of clinical symptoms. This review highlights potential biomarkers, merits, and demerits of different neuroimaging modalities and their clinical value in MCI and AD patients. Further studies are necessary to explore more biomarkers and overcome the limitations of multiple neuroimaging modalities for inclusion in diagnostic criteria for AD.

To investigate prospectively multidetector computed tomography (CT) (MDCT) and magnetic resonance (MR) imaging (MRI) in the preoperative assessment of focal liver lesions. Multiphasic MDCT and conventional gadolinium-enhanced MRI were performed on 31 consecutive patients prior to hepatic surgery. All images were blindly analyzed as consensus reading. Lesion counts and their relation to vascular structures and possible extrahepatic disease were determined. The data from the MDCT and MRI were compared with the results obtained by intraoperative ultrasound (IOUS) and palpation. Histopathologic verification was available. At surgery, IOUS and palpation revealed 45 solid liver lesions. From these, preoperative MDCT detected 43 (96%) and MRI 35 (78%) deposits. MDCT performed statistically better than MRI in lesion detection (P=0.008). Assessment of lesion vascular proximity was correctly determined by MDCT in 98% of patients and by MRI in 87%. Statistical difference was found (P=0.002). IOUS and palpation changed the preoperative surgical plan as a result of extrahepatic disease in 8/31 (26%) cases. In MDCT as well in MRI extrahepatic involvement was suspected in two cases. MDCT was superior to MRI and nearly equal to IOUS in liver lesion detection and in the determination of lesion vascular proximity. However, both techniques fail to reliably detect extrahepatic disease.

Brain lesion segmentation is crucial for neurological disease research and diagnosis. As different types of lesions exhibit distinct characteristics on different imaging modalities, segmentation methods are typically developed in a task-specific manner, where each segmentation model is tailored to a specific lesion type and modality. However, the use of task-specific models requires predetermination of the lesion type and imaging modality, which complicates their deployment in real-world scenarios. In this work, we propose a universal foundation model for brain lesion segmentation on magnetic resonance imaging (MRI), which can automatically segment different types of brain lesions given input of various MRI modalities. We develop a novel Mixture of Modality Experts (MoME) framework with multiple expert networks attending to different imaging modalities. A hierarchical gating network is proposed to combine the expert predictions and foster expertise collaboration. Moreover, to avoid the degeneration of each expert network, we introduce a curriculum learning strategy during training to preserve the specialisation of each expert. In addition to MoME, to handle the combination of multiple input modalities, we propose MoME+, which uses a soft dispatch network for input modality routing. We evaluated the proposed method on nine brain lesion datasets, encompassing five imaging modalities and eight lesion types. The results show that our model outperforms state-of-the-art universal models for brain lesion segmentation and achieves promising generalisation performance onto unseen datasets.

Heart problems are common in newborn babies, affecting approximately 5-10 in 1000 babies. Some are more serious than others, but most babies born with heart problems do not have other health issues. Of those babies who have a serious heart problem, almost 1 in 4 will have heart surgery in their first year. In the UK, pregnant women are offered a scan at around 20weeks to try and spot any heart problems. In most cases there is not a clear reason for the problem, but sometimes other issues, such as genetic conditions, are discovered. In recent years the care given to these babies after they are born has improved their chances of surviving. However, it is recognised that babies born with heart problems have a risk of delays in their learning and development. This may be due to their medical condition, or as a result of surgery and complications after birth. In babies with heart problems, there is a need for more research on ultrasound and magnetic resonance imaging (MRI) to understand how the brain develops and why these babies are more likely to have delays in learning and development. This paper discusses the way ultrasound and MRI are used in assessing the baby's brain. Ultrasound is often used to spot any problems, looking at how the baby's brain develops in pregnancy. Advances in ultrasound technologies have made this easier. MRI is well-established and safe in pregnancy, and if problems in the brain have been seen on ultrasound, MRI may be used to look at these problems in more detail. While it is not always clear what unusual MRI findings can mean for the baby in the long term, increased understanding may mean parents can be given more information about possible outcomes for the baby and may help to improve the counselling they are offered before their baby's birth.

In rheumatoid arthritis (RA) hands, we applied high-resolution peripheral quantitative computed tomography (HR-pQCT) and 3 Tesla (3 T) magnetic resonance imaging (MRI), which are new methods for erosion detection and bone marrow edema (BME) quantification. We compared the erosion measurements between these techniques with conventional radiographs (CR) in order to examine their significance for evaluating structural abnormalities. In 16 RA patients, HR-pQCT of metacarpophalangeal and wrist joints, 3 T MRI of wrist joints, as well as CR in both hands and feet were performed. Ten patients had 1-year follow-up CR. CRs were graded according to the modified Sharp score (MSS). Bone erosions were evaluated in HR-pQCT and MRI. BME pattern was quantified from MRI for volume, signal change and total burden. The erosion detection sensitivity of MRI was 85.7% and CR was 60.9% when HR-pQCT was considered as a reference method. The smallest dimensions of erosion detected by HR-pQCT, MRI and CR were 0.09, 0.14 and 0.66 cm, respectively. Baseline total MSS was correlated with HR-pQCT erosion measures, MRI erosion measures and MRI BME volume (P < 0.05). The mean difference between baseline and 1-year follow-up MSS (delta MSS) was 1.2. A trend was observed toward a correlation between delta MSS and MRI BME volume and burden. This study demonstrates that HR-pQCT detects more and smaller bone erosions compared to MRI and CR. In addition, 3 T MRI can provide quantitative measurement of BME. Combination of HR-pQCT and MRI modalities may provide powerful tools to evaluate joint inflammation and bone damage in RA.

Introduction:Bacterial meningitis is an acute infectious inflammation of the protective membranes covering the brain. Its early diagnosis is vital because of its high morbidity and mortality. It is mostly diagnosed by a gold standard diagnostic tool i.e. Cerebrospinal Fluid (CSF) analysis. However, it is sometimes difficult and or impossible to do this procedure and an alternative diagnostic tool is needed. Contrast enhanced magnetic resonance imaging can detect the pus or other changes in subarachnoid space. But our optimal aim is to use an imaging method without using contrast to be useable and available in more specific condition.Methods:This study aimed to survey the role of non-contrast Magnetic Resonance Imaging (MRI) in the diagnosis of the bacterial meningitis. MEDLINE/PubMed Central, Web of Science and Scopus were searched without time period and language limitation until March 2017. We found 6410 papers in our initial search. After assessing the content of the papers based on Cochrane library guidelines and inclusion/exclusion criteria, 6 relevant studies were included in the systematic review. All of included studies were observational studies.Results:MRI studies demonstrated that Fluid Attenuation Inversion Recovery (FLAIR) and Diffusion-Weighted Image (DWI) MR imaging among all MRI modalities can detect some abnormalities compatible with bacterial meningitis. FLAIR and DWI-MR imaging are potentially useful to diagnose bacterial meningitis and can be used in emergent condition in which bacterial meningitis is highly suspicious and the other diagnostic tools are not available or feasible.

Introduction: It is not uncommon for clinicians to see focal liver lesions (FLLs), which can be anything from harmless hemangiomas to dangerous hepatocellular carcinoma (HCC) and metastases. In order to properly treat these lesions, an accurate diagnosis is required. The purpose of this research is to compare magnetic resonance imaging (MRI) results with histological ones in order to determine how well MRI can detect and characterize localized liver lesions. Materials and Methods: This prospective study comprised 50 patients with localized liver lesions identified via ultrasonography or computed tomography (CT) and then assessed using MRI. This study was conducted at the Department of Radiology at Gouri Devi Institute of Medical Science and Hospital, Rajbandh, West Bengal, India from June 2021 to May 2022. Patients received MRI images with T1-weighted, T2-weighted, diffusion-weighted imaging (DWI), and contrast-enhanced dynamic imaging with gadolinium-based contrast agents. The imaging findings were analyzed according to lesion morphology, signal intensity, enhancement patterns, and diffusion restriction. All patients received biopsy or surgical resection, with histological findings serving as the reference standard. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MRI in identifying malignant and benign liver lesions were assessed. Results: There were 30 cancerous tumors and 20 benign ones among the 50 individuals. To distinguish between malignant and benign lesions, MRI showed a sensitivity of 94%, specificity of 89%, PPV of 93%, and NPV of 90%. Malignant lesions affecting the liver were most commonly found as hepatocellular carcinoma (HCC) and metastatic liver lesions, respectively. In terms of benign lesions, hemangiomas, hepatic adenomas, and focal nodular hyperplasia (FNH) were detected most often. Even while most benign lesions showed facilitated diffusion on diffusion-weighted imaging (DWI), 85% of malignant lesions showed restricted diffusion. Contrast-enhanced sequences showed that malignant lesions had arterial-phase hyperenhancement with washout, but benign lesions showed distinctive enhancement patterns as homogenous enhancement in FNH and peripheral nodular enhancement in hemangiomas. Conclusion: Magnetic resonance imaging (MRI) provides a very accurate picture of localized liver lesions and correlates very well with histological results. Diagnostic certainty in distinguishing benign from malignant liver lesions is greatly increased when morphologic evaluation, diffusion restriction, and dynamic contrast enhancement patterns are combined. To better characterize liver lesions, MRI should be regarded the imaging modality of choice. This will help reduce the need for biopsies that aren't essential and guide appropriate clinical therapy.

Background: Detection of meniscal ramp lesions concomitant with anterior cruciate ligament (ACL) injury using conventional magnetic resonance imaging (MRI) has low sensitivity, and these lesions are currently difficult to diagnose preoperatively. Purpose/Hypothesis: The purpose of this study was to assess the accuracy of MRI in detecting the presence of meniscal ramp lesions in the 120° flexed-knee position compared with that in the near–extended-knee position. It was hypothesized that the diagnostic performance of MRI in the 120° flexed-knee position would be better than that in the extended-knee position. Study Design: Cohort study (Diagnosis); Level of evidence, 2. Methods: This retrospective study of prospectively collected data between February 2019 and January 2024 included 154 consecutive patients undergoing ACL reconstruction. All patients underwent 3-T MRI examination in the near extended- and 120° flexed-knee positions preoperatively. The presence and Thaunat classification of ramp lesions were separately detected on each MRI scan and confirmed via arthroscopy during ACL reconstruction. Diagnostic sensitivity, specificity, and conditional relative odds ratios for detecting ramp lesions and the classification accuracy were compared between 2 MRI modalities. The accuracies of these MRI scans in acute and chronic cases were also evaluated. Results: This study included 154 patients (79 male and 75 female; mean age, 29.0 ± 14.2 years). A total of 62 ramp lesions (40.3%) were observed on arthroscopy. The sensitivity and specificity of MRI in near extension were 69.4% and 77.2%, respectively, and those in flexion were 91.9% and 94.6%, respectively, with significant superiority in MRI at knee flexion (P = .003 and P < .001, respectively). The conditional relative odds ratio between the MRI examinations at these 2 positions was 10.3 (95% CI, 4.82-21.8). The classification accuracy of MRI in flexion was significantly higher than that of MRI in near extension (accuracy, 49 vs 11; P < .001). The diagnostic accuracy of MRI in the 120° flexed-knee position was significantly higher than that in the near–extended-knee position in the acute cases (P < .05); however, no significant differences were observed in the chronic cases. Conclusion: The diagnostic accuracy of MRI in the 120° flexed-knee position for detecting and classifying meniscal ramp lesions was superior to that of MRI in the near–extended-knee position.

Research utilizing magnetic resonance imaging (MRI) has been crucial to the understanding of the neuropathological mechanisms behind and clinical identification of Alzheimer’s disease (AD) and mild cognitive impairment (MCI). MRI modalities show patterns of brain damage that discriminate AD from other brain illnesses and brain abnormalities that are associated with risk of conversion to AD from MCI and other behavioural outcomes. This review discusses the application of various MRI techniques to and their clinical usefulness in AD and MCI. MRI modalities covered include structural MRI, diffusion tensor imaging (DTI), arterial spin labelling (ASL), magnetic resonance spectroscopy (MRS), and functional MRI (fMRI). There is much evidence supporting the validity of MRI as a biomarker for these disorders; however, only traditional structural imaging is currently recommended for routine use in clinical settings. Future research is needed to warrant the inclusion for more advanced MRI methodology in forthcoming revisions to diagnostic criteria for AD and MCI.

Synthetic MRI in action: A novel framework in data augmentation strategies for robust multi-modal brain tumor segmentation