Texture Analysis for Muscular Dystrophy Classification in MRI with Improved Class Activation Mapping

Convolutional Neural Networks to Study Contrast-Enhanced Magnetic Resonance Imaging-Based Skeletal Calf Muscle Perfusion in Peripheral Artery Disease

Peripheral artery disease (PAD) is associated with impaired blood flow in the lower extremities and histopathologic changes of the skeletal calf muscles, resulting in abnormal microvascular perfusion. We studied the use of convolution neural networks (CNNs) to differentiate patients with PAD from matched controls using perfusion pattern features from contrast-enhanced magnetic resonance imaging (CE-MRI) of the skeletal calf muscles. We acquired CE-MRI based skeletal calf muscle perfusion in 56 patients (36 patients with PAD and 20 matched controls). Microvascular perfusion imaging was performed after reactive hyperemia at the midcalf level, with a temporal resolution of 409 ms. We analyzed perfusion scans up to 2 minutes indexed from the local precontrast arrival time frame. Skeletal calf muscles, including the anterior muscle, lateral muscle, deep posterior muscle group, and the soleus and gastrocnemius muscles, were segmented semiautomatically. Segmented muscles were represented as 3-dimensional Digital Imaging and Communications in Medicine stacks of CE-MRI perfusion scans for deep learning (DL) analysis. We tested several CNN models for the 3-dimensional CE-MRI perfusion stacks to classify patients with PAD from matched controls. A total of 2 of the best performing CNNs (resNet and divNet) were selected to develop the final classification model. A peak accuracy of 75% was obtained for resNet and divNet. Specificity was 80% and 94% for resNet and divNet, respectively. In conclusion, DL using CNNs and CE-MRI skeletal calf muscle perfusion can discriminate patients with PAD from matched controls. DL methods may be of interest for the study of PAD.

A machine learning-based approach to identify peripheral artery disease using texture features from contrast-enhanced magnetic resonance imaging

Diagnosing and assessing the risk of peripheral artery disease (PAD) has long been a focal point for medical practitioners. The impaired blood circulation in PAD patients results in altered microvascular perfusion patterns in the calf muscles which is the primary location of intermittent claudication pain. Consequently, we hypothesized that changes in perfusion and increase in connective tissue could lead to alterations in the appearance or texture patterns of the skeletal calf muscles, as visualized with non-invasive imaging techniques. We designed an automatic pipeline for textural feature extraction from contrast-enhanced magnetic resonance imaging (CE-MRI) scans and used the texture features to train machine learning models to detect the heterogeneity in the muscle pattern among PAD patients and matched controls. CE-MRIs from 36 PAD patients and 20 matched controls were used for preparing training and testing data at a 7:3 ratio with cross-validation (CV) techniques. We employed feature arrangement and selection methods to optimize the number of features. The proposed method achieved a peak accuracy of 94.11% and a mean testing accuracy of 84.85% in a 2-class classification approach (controls vs. PAD). A three-class classification approach was performed to identify a high-risk PAD sub-group which yielded an average test accuracy of 83.23% (matched controls vs. PAD without diabetes vs. PAD with diabetes). Similarly, we obtained 78.60% average accuracy among matched controls, PAD treadmill exercise completers, and PAD exercise treadmill non-completers. Machine learning and imaging-based texture features may be of interest in the study of lower extremity ischemia.

Explainable Deep Learning-Assisted Self-Calibrating Colorimetric Patches for In Situ Sweat Analysis

Sweat has emerged as a compelling analyte for noninvasive biosensing technology because it contains a wealth of important biomarkers in hormones, organic biomacromolecules, and various ionic mixtures. These components offer valuable insights and can reflect an individual's physiological conditions. Here, we introduced an explainable deep learning (DL)-assisted wearable self-calibrating colorimetric biosensing analysis platform to efficiently and precisely detect the biomarker's concentration in sweat. Specifically, we have integrated the advantages of the colorimetric sensing method, adsorbing-swelling hydrogel, and explainable DL algorithms to develop an enzyme/indicator-immobilized colorimetric patch, which has reliable colorimetric sensing ability and excellent adsorbing-swelling function. A total of 5625 colorimetric images were collected as the analysis data set and assessed two DL algorithms and seven machine learning (ML) algorithms. Zn2+, glucose, and Ca2+ in human sweats could be facilely classified and quantified with 100% accuracy via the convolutional neural network (CNN) model, and the testing results of actual sweats via the DL-assisted colorimetric approach are 91.7-97.2% matching with the classical UV-vis spectrum. Class activation mapping (CAM) was utilized to visualize the inner working mechanism of CNN operation, which contributes to verify and explicate the design rationality of the noninvasive biosensing technology. An "end-to-end" model was established to ascertain the black box of the DL algorithm, promoted software design or principium optimization, and contributed facile indicators for health monitoring, disease prevention, and clinical diagnosis.

SEMPAI: a Self-Enhancing Multi-Photon Artificial Intelligence for Prior-Informed Assessment of Muscle Function and Pathology

Deep learning (DL) shows notable success in biomedical studies. However, most DL algorithms work as black boxes, exclude biomedical experts, and need extensive data. This is especially problematic for fundamental research in the laboratory, where often only small and sparse data are available and the objective is knowledge discovery rather than automation. Furthermore, basic research is usually hypothesis-driven and extensive prior knowledge (priors) exists. To address this, the Self-Enhancing Multi-Photon Artificial Intelligence (SEMPAI) that is designed for multiphoton microscopy (MPM)-based laboratory research is presented. It utilizes meta-learning to optimize prior (and hypothesis) integration, data representation, and neural network architecture simultaneously. By this, the method allows hypothesis testing with DL and provides interpretable feedback about the origin of biological information in 3D images. SEMPAI performs multi-task learning of several related tasks to enable prediction for small datasets. SEMPAI is applied on an extensive MPM database of single muscle fibers from a decade of experiments, resulting in the largest joint analysis of pathologies and function for single muscle fibers to date. It outperforms state-of-the-art biomarkers in six of seven prediction tasks, including those with scarce data. SEMPAI's DL models with integrated priors are superior to those without priors and to prior-only approaches.

Muscle "islands": An MRI signature distinguishing neurogenic from myopathic causes of early onset distal weakness

Muscle MRI has an increasing role in diagnosis of inherited neuromuscular diseases, but no features are known which reliably differentiate myopathic and neurogenic conditions. Using patients presenting with early onset distal weakness, we aimed to identify an MRI signature to distinguish myopathic and neurogenic conditions. We identified lower limb MRI scans from patients with either genetically (n = 24) or clinically (n = 13) confirmed diagnoses of childhood onset distal myopathy or distal spinal muscular atrophy. An initial exploratory phase reviewed 11 scans from genetically confirmed patients identifying a single potential discriminatory marker concerning the pattern of fat replacement within muscle, coined "islands". This pattern comprised small areas of muscle tissue with normal signal intensity completely surrounded by areas with similar intensity to subcutaneous fat. In the subsequent validation phase, islands correctly classified scans from all 12 remaining genetically confirmed patients, and 12/13 clinically classified patients. In the genetically confirmed patients MRI classification of neurogenic/myopathic aetiology had 100% accuracy (24/24) compared with 65% accuracy (15/23) for EMG, and 79% accuracy (15/19) for muscle biopsy. Future studies are needed in other clinical contexts, however the presence of islands appears to highly suggestive of a neurogenic aetiology in patients presenting with early onset distal motor weakness.

Deep-Learning-Driven Quantification of Interstitial Fibrosis in Digitized Kidney Biopsies

Interstitial fibrosis and tubular atrophy (IFTA) on a renal biopsy are strong indicators of disease chronicity and prognosis. Techniques that are typically used for IFTA grading remain manual, leading to variability among pathologists. Accurate IFTA estimation using computational techniques can reduce this variability and provide quantitative assessment. Using trichrome-stained whole-slide images (WSIs) processed from human renal biopsies, we developed a deep-learning framework that captured finer pathologic structures at high resolution and overall context at the WSI level to predict IFTA grade. WSIs (n = 67) were obtained from The Ohio State University Wexner Medical Center. Five nephropathologists independently reviewed them and provided fibrosis scores that were converted to IFTA grades: ≤10% (none or minimal), 11% to 25% (mild), 26% to 50% (moderate), and >50% (severe). The model was developed by associating the WSIs with the IFTA grade determined by majority voting (reference estimate). Model performance was evaluated on WSIs (n = 28) obtained from the Kidney Precision Medicine Project. There was good agreement on the IFTA grading between the pathologists and the reference estimate (κ = 0.622 ± 0.071). The accuracy of the deep-learning model was 71.8% ± 5.3% on The Ohio State University Wexner Medical Center and 65.0% ± 4.2% on Kidney Precision Medicine Project data sets. Our approach to analyzing microscopic- and WSI-level changes in renal biopsies attempts to mimic the pathologist and provides a regional and contextual estimation of IFTA. Such methods can assist clinicopathologic diagnosis.

The increasing role of muscle MRI to monitor changes over time in untreated and treated muscle diseases

PURPOSE OF REVIEW

This review aims to discuss the recent results of studies published applying quantitative MRI sequences to large cohorts of patients with neuromuscular diseases.

RECENT FINDINGS

Quantitative MRI sequences are now available to identify and quantify changes in muscle water and fat content. These two components have been associated with acute and chronic injuries, respectively. Studies show that the increase in muscle water is not only reversible if therapies are applied successfully but can also predict fat replacement in neurodegenerative diseases. Muscle fat fraction correlates with muscle function tests and increases gradually over time in parallel with the functional decline of patients with neuromuscular diseases. There are new spectrometry-based sequences to quantify other components, such as glycogen, electrolytes or the pH of the muscle fibre, extending the applicability of MRI to the study of several processes in neuromuscular diseases.

SUMMARY

The latest results obtained from the study of long cohorts of patients with various neuromuscular diseases open the door to the use of this technology in clinical trials, which would make it possible to obtain a new measure for assessing the effectiveness of new treatments. The challenge is currently the popularization of these studies and their application to the monitoring of patients in the daily clinic.

Respiratory muscle imaging by ultrasound and MRI in neuromuscular disorders

Respiratory muscle weakness is common in neuromuscular disorders and leads to significant respiratory difficulties. Therefore, reliable and easy assessment of respiratory muscle structure and function in neuromuscular disorders is crucial. In the last decade, ultrasound and MRI emerged as promising imaging techniques to assess respiratory muscle structure and function. Respiratory muscle imaging directly measures the respiratory muscles and, in contrast to pulmonary function testing, is independent of patient effort. This makes respiratory muscle imaging suitable to use as tool in clinical respiratory management and as outcome parameter in upcoming drug trials for neuromuscular disorders, particularly in children. In this narrative review, we discuss the latest studies and technological developments in imaging of the respiratory muscles by US and MR, and its clinical application and limitations. We aim to increase understanding of respiratory muscle imaging and facilitate its use as outcome measure in daily practice and clinical trials.

A deep learning model for diagnosing dystrophinopathies on thigh muscle MRI images

Background

Dystrophinopathies are the most common type of inherited muscular diseases. Muscle biopsy and genetic tests are effective to diagnose the disease but cost much more than primary hospitals can reach. The more available muscle MRI is promising but its diagnostic results highly depends on doctors’ experiences. This study intends to explore a way of deploying a deep learning model for muscle MRI images to diagnose dystrophinopathies.

Methods

This study collected 2536 T1WI images from 432 cases who had been diagnosed by genetic analysis and/or muscle biopsy, including 148 cases with dystrophinopathies and 284 cases with other diseases. The data was randomly divided into three sets: the data from 233 cases were used to train the CNN model, the data from 97 cases for the validation experiments, and the data from 102 cases for the test experiments. We also validated our models expertise at diagnosing by comparing the model’s results on the 102 cases with those of three skilled radiologists.

Results

The proposed model achieved 91% (95% CI: 0.88, 0.93) accuracy on the test set, higher than the best accuracy of 84% in radiologists. It also performed better than the skilled radiologists in sensitivity : sensitivities of the models and the doctors were 0.89 (95% CI: 0.85 0.93) versus 0.79 (95% CI:0.73, 0.84; p = 0.190).

Conclusions

The deep model achieved excellent accuracy and sensitivity in identifying cases with dystrophinopathies. The comparable performance of the model and skilled radiologists demonstrates the potential application of the model in diagnosing dystrophinopathies through MRI images.

Functional changes of the lateral pterygoid muscle in patients with temporomandibular disorders: a pilot magnetic resonance images texture study

BACKGROUND

Texture features were the intrinsic properties of the human tissues and could efficiently detect the subtle functional changes of involved tissue. The pathologic changes of the lateral pterygoid muscle (LPM) were significantly correlated with the temporomandibular disc displacement. However, the occult functional changes of LPM could not be detected by the naked eye on the medical images. The current study was aimed to evaluate the functional changes of the LPM in the patients with temporomandibular disorders (TMDs) using texture analysis.

METHODS

Twenty-nine patients with TMD were performed with magnetic resonance (MR) imaging on a 3.0T MR scanner, who were consecutively recruited from the TMD clinic of Hainan Hospital of Chinese People's Liberation Army General Hospital from February 2019 to September 2019. The patients were classified into three groups according to the disc displacement: disc without displacement (DWoD), disc displacement with reduction (DDWR) and disc displacement without reduction (DDWoR). The gray-level co-occurrence matrix method was applied with the texture analysis of LPM on the axial T2-weighted imaging. The texture features included angular second moment, contrast, correlation, inverse different moment, and entropy. One-way analysis of variance was used for grouped comparisons and receiver operating characteristics (ROC) curve analysis was applied to evaluate the diagnostic efficacy of the texture parameters.

RESULTS

Texture contrast of LPM presented significantly lower in DDWoR (46.30 [35.03, 94.48]) than that in DWoD (123.85 [105.06, 143.23]; test statistic = 23.05; P < 0.001). Texture entropy of LPM showed significant differences among DWoD (7.62 ± 0.33), DDWR (6.76 ± 0.35), and DDWoR (6.46 ± 0.39) (PDWoD-DDWR < 0.001, PDWoD-DDWoR < 0.001, and PDDWR-DDWoR = 0.014). Area under the ROC curve (AUC) demonstrated that texture entropy had an excellent diagnostic accuracy for DWoD-DDWR (AUC = 0.96) and DWoD-DDWoR (AUC = 0.98).

CONCLUSION

The texture contrast and entropy could identify the altered functional status of LPM in patients with TMD and could be considered as the effective imaging biomarker to evaluate the functional changes of LPM in TMD.

Texture analysis and machine learning to predict water T2 and fat fraction from non-quantitative MRI of thigh muscles in Facioscapulohumeral muscular dystrophy

PURPOSE

Quantitative MRI (qMRI) plays a crucial role for assessing disease progression and treatment response in neuromuscular disorders, but the required MRI sequences are not routinely available in every center. The aim of this study was to predict qMRI values of water T2 (wT2) and fat fraction (FF) from conventional MRI, using texture analysis and machine learning.

METHOD

Fourteen patients affected by Facioscapulohumeral muscular dystrophy were imaged at both thighs using conventional and quantitative MR sequences. Muscle FF and wT2 were calculated for each muscle of the thighs. Forty-seven texture features were extracted for each muscle on the images obtained with conventional MRI. Multiple machine learning regressors were trained to predict qMRI values from the texture analysis dataset.

RESULTS

Eight machine learning methods (linear, ridge and lasso regression, tree, random forest (RF), generalized additive model (GAM), k-nearest-neighbor (kNN) and support vector machine (SVM) provided mean absolute errors ranging from 0.110 to 0.133 for FF and 0.068 to 0.115 for wT2. The most accurate methods were RF, SVM and kNN to predict FF, and tree, RF and kNN to predict wT2.

CONCLUSION

This study demonstrates that it is possible to estimate with good accuracy qMRI parameters starting from texture analysis of conventional MRI.

Super resolution microscopy and deep learning identify Zika virus reorganization of the endoplasmic reticulum

The endoplasmic reticulum (ER) is a complex subcellular organelle composed of diverse structures such as tubules, sheets and tubular matrices. Flaviviruses such as Zika virus (ZIKV) induce reorganization of ER membranes to facilitate viral replication. Here, using 3D super resolution microscopy, ZIKV infection is shown to induce the formation of dense tubular matrices associated with viral replication in the central ER. Viral non-structural proteins NS4B and NS2B associate with replication complexes within the ZIKV-induced tubular matrix and exhibit distinct ER distributions outside this central ER region. Deep neural networks trained to distinguish ZIKV-infected versus mock-infected cells successfully identified ZIKV-induced central ER tubular matrices as a determinant of viral infection. Super resolution microscopy and deep learning are therefore able to identify and localize morphological features of the ER and allow for better understanding of how ER morphology changes due to viral infection.

Prognosis of cervical myelopathy based on diffusion tensor imaging with artificial intelligence methods

Diffusion tensor imaging (DTI) has been proposed for the prognosis of cervical myelopathy (CM), but the manual analysis of DTI features is complicated and time consuming. This study evaluated the potential of artificial intelligence (AI) methods in the analysis of DTI for the prognosis of CM. Seventy-five patients who underwent surgical treatment for CM were recruited for DTI imaging and were divided into two groups based on their one-year follow-up recovery. The DTI features of fractional anisotropy, axial diffusivity, radial diffusivity, and mean diffusivity were extracted from DTI maps of all cervical levels. Conventional AI models using logistic regression (LR), k-nearest neighbors (KNN), and a radial basis function kernel support vector machine (RBF-SVM) were built using these DTI features. In addition, a deep learning model was applied to the DTI maps. Their performances were compared using 50 repeated 10-fold cross-validations. The accuracy of the classifications reached 74.2% ± 1.6% for LR, 85.6% ± 1.4% for KNN, 89.7% ± 1.6% for RBF-SVM, and 59.2% ± 3.8% for the deep leaning model. The RBF-SVM algorithm achieved the best accuracy, with sensitivity and specificity of 85.0% ± 3.4% and 92.4% ± 1.9% respectively. This finding indicates that AI methods are feasible and effective for DTI analysis for the prognosis of CM.

Advancements in magnetic resonance imaging-based biomarkers for muscular dystrophy

Recent years have seen steady progress in the identification of genetic muscle diseases as well as efforts to develop treatment for these diseases. Consequently, sensitive and objective new methods are required to identify and monitor muscle pathology. Magnetic resonance imaging offers multiple potential biomarkers of disease severity in the muscular dystrophies. This Review uses a pathology-based approach to examine the ways in which MRI and spectroscopy have been used to study muscular dystrophies. Methods that have been used to quantitate intramuscular fat, edema, fiber orientation, metabolism, fibrosis, and vascular perfusion are examined, and this Review describes how MRI can help diagnose these conditions and improve upon existing muscle biomarkers by detecting small increments of disease-related change. Important challenges in the implementation of imaging biomarkers, such as standardization of protocols and validating imaging measurements with respect to clinical outcomes, are also described.