Application of Texture Analysis to Study Small Vessel Disease and Blood-Brain Barrier Integrity

Machine-learning with region-level radiomic and dosimetric features for predicting radiotherapy-induced rectal toxicities in prostate cancer patients

BACKGROUND AND PURPOSE

This study aims to build machine learning models to predict radiation-induced rectal toxicities for three clinical endpoints and explore whether the inclusion of radiomic features calculated on radiotherapy planning computerised tomography (CT) scans combined with dosimetric features can enhance the prediction performance.

MATERIALS AND METHODS

183 patients recruited to the VoxTox study (UK-CRN-ID-13716) were included. Toxicity scores were prospectively collected after 2 years with grade ≥ 1 proctitis, haemorrhage (CTCAEv4.03); and gastrointestinal (GI) toxicity (RTOG) recorded as the endpoints of interest. The rectal wall on each slice was divided into 4 regions according to the centroid, and all slices were divided into 4 sections to calculate region-level radiomic and dosimetric features. The patients were split into a training set (75%, N = 137) and a test set (25%, N = 46). Highly correlated features were removed using four feature selection methods. Individual radiomic or dosimetric or combined (radiomic + dosimetric) features were subsequently classified using three machine learning classifiers to explore their association with these radiation-induced rectal toxicities.

RESULTS

The test set area under the curve (AUC) values were 0.549, 0.741 and 0.669 for proctitis, haemorrhage and GI toxicity prediction using radiomic combined with dosimetric features. The AUC value reached 0.747 for the ensembled radiomic-dosimetric model for haemorrhage.

CONCLUSIONS

Our preliminary results show that region-level pre-treatment planning CT radiomic features have the potential to predict radiation-induced rectal toxicities for prostate cancer. Moreover, when combined with region-level dosimetric features and using ensemble learning, the model prediction performance slightly improved.

AI-Based Automated Lipomatous Tumor Segmentation in MR Images: Ensemble Solution to Heterogeneous Data

Deep learning (DL) has been proposed to automate image segmentation and provide accuracy, consistency, and efficiency. Accurate segmentation of lipomatous tumors (LTs) is critical for correct tumor radiomics analysis and localization. The major challenge of this task is data heterogeneity, including tumor morphological characteristics and multicenter scanning protocols. To mitigate the issue, we aimed to develop a DL-based Super Learner (SL) ensemble framework with different data correction and normalization methods. Pathologically proven LTs on pre-operative T1-weighted/proton-density MR images of 185 patients were manually segmented. The LTs were categorized by tumor locations as distal upper limb (DUL), distal lower limb (DLL), proximal upper limb (PUL), proximal lower limb (PLL), or Trunk (T) and grouped by 80%/9%/11% for training, validation and testing. Six configurations of correction/normalization were applied to data for fivefold-cross-validation trainings, resulting in 30 base learners (BLs). A SL was obtained from the BLs by optimizing SL weights. The performance was evaluated by dice-similarity-coefficient (DSC), sensitivity, specificity, and Hausdorff distance (HD95). For predictions of the BLs, the average DSC, sensitivity, and specificity from the testing data were 0.72 [Formula: see text] 0.16, 0.73 [Formula: see text] 0.168, and 0.99 [Formula: see text] 0.012, respectively, while for SL predictions were 0.80 [Formula: see text] 0.184, 0.78 [Formula: see text] 0.193, and 1.00 [Formula: see text] 0.010. The average HD95 of the BLs were 11.5 (DUL), 23.2 (DLL), 25.9 (PUL), 32.1 (PLL), and 47.9 (T) mm, whereas of SL were 1.7, 8.4, 15.9, 2.2, and 36.6 mm, respectively. The proposed method could improve the segmentation accuracy and mitigate the performance instability and data heterogeneity aiding the differential diagnosis of LTs in real clinical situations.

Texture analysis of apparent diffusion coefficient maps in predicting the clinical functional outcomes of acute ischemic stroke

Purpose

To investigate texture analysis (TA) based on apparent diffusion coefficient (ADC) map in predicting acute ischemic stroke (AIS) prognosis and discriminating TA features in stroke subtypes.

Methods

This retrospective study included patients with AIS between January 2018 and April 2021. The patients were assigned to the favorable [modified Rankin Scale (mRS) score ≤ 2] and unfavorable (mRS score > 2) outcome groups. All patients underwent stroke subtyping according to the Trial of Org 10,172 in Acute Stroke Treatment (TOAST) classification. The TA features were extracted from infarction lesions on the ADC map. The demographic characteristics, clinical characteristics, and texture features were used to construct prediction models with recurrent neural network (RNN). The receiver operating characteristic (ROC) curves were implemented to evaluate the performance of the predictive models.

Results

A total of 1,003 patients (682 male; mean age 65.90 ± 12.44) with AIS having documented the 90-day mRS score were identified, including 840 with favorable outcomes. In the validation set, the area under the curve (AUC) of the predictive model using only clinical characteristics achieved an AUC of 0.56, texture model 0.77, the model combining both clinical and texture features showed better with an AUC of 0.78. The texture feature profiles differed between large artery atherosclerosis (LAA) and small artery occlusion (SAO) subtypes (all p < 0.05). The AUC of combined prediction models for LAA and SAO subtypes was 0.80 and 0.81.

Conclusion

Texture analysis based on ADC map could be useful as an adjunctive tool for predicting ischemic stroke prognosis.

Intracerebral haemorrhage

Intracerebral haemorrhage (ICH) is a dramatic condition caused by the rupture of a cerebral vessel and the entry of blood into the brain parenchyma. ICH is a major contributor to stroke-related mortality and dependency: only half of patients survive for 1 year after ICH, and patients who survive have sequelae that affect their quality of life. The incidence of ICH has increased in the past few decades with shifts in the underlying vessel disease over time as vascular prevention has improved and use of antithrombotic agents has increased. The pathophysiology of ICH is complex and encompasses mechanical mass effect, haematoma expansion and secondary injury. Identifying the causes of ICH and predicting the vital and functional outcome of patients and their long-term vascular risk have improved in the past decade; however, no specific treatment is available for ICH. ICH remains a medical emergency, with prevention of haematoma expansion as the key therapeutic target. After discharge, secondary prevention and management of vascular risk factors in patients remains challenging and is based on an individual benefit-risk balance evaluation.

Radiomics-based infarct features on CT predict hemorrhagic transformation in patients with acute ischemic stroke

Objective

To develop and validate a model based on the radiomics features of the infarct areas on non-contrast-enhanced CT to predict hemorrhagic transformation (HT) in acute ischemic stroke.

Materials and methods

A total of 118 patients diagnosed with acute ischemic stroke in two centers from January 2019 to February 2022 were included. The radiomics features of infarcted areas on non-contrast-enhanced CT were extracted using 3D-Slicer. A univariate analysis and the least absolute shrinkage and selection operator (LASSO) were used to select features, and the radiomics score (Rad-score) was then constructed. The predictive model of HT was constructed by analyzing the Rad-score and clinical and imaging features in the training cohort, and it was verified in the validation cohort. The model was evaluated with the receiver operating characteristic curve, calibration curve and decision curve, and the prediction performance of the model in different scenarios was further discussed hierarchically.

Results

Of the 118 patients, 52 developed HT, including 21 cases of hemorrhagic infarct (HI) and 31 cases of parenchymal hematoma (PH). The Rad-score was constructed from five radiomics features and was the only independent predictor for HT. The predictive model was constructed from the Rad-score. The area under the curve (AUCs) of the model for predicting HT in the training and validation cohorts were 0.845 and 0.750, respectively. Calibration curve and decision curve analyses showed that the model performed well. Further analysis found that the model predicted HT for different infarct sizes or treatment methods in the training and validation cohorts with 78.3 and 71.4% accuracy, respectively. For all samples, the model predicted an AUC of 0.754 for HT in patients within 4.5 h since stroke onset, and predicted an AUC of 0.648 for PH.

Conclusion

This model, which was based on CT radiomics features, could help to predict HT in the setting of acute ischemic stroke for any infarct size and provide guiding suggestions for clinical treatment and prognosis evaluation.

The role of MRI-based texture analysis to predict the severity of brain injury in neonates with perinatal asphyxia

OBJECTIVE

To evaluate the efficacy of the MRI-based texture analysis (TA) of the basal ganglia and thalami to distinguish moderate-to-severe hypoxic-ischemic encephalopathy (HIE) from mild HIE in neonates.

METHODS

This study included 68 neonates (15 with mild, 20 with moderate-to-severe HIE, and 33 control) were born at 37 gestational weeks or later and underwent MRI in first 10 days after birth. The basal ganglia and thalami were delineated for TA on the apparent diffusion coefficient (ADC) maps, T₁-, and T₂ weighted images. The basal ganglia, thalami, and the posterior limb of the internal capsule (PLIC) were also evaluated visually on diffusion-weighted imaging and T₁ weighted sequence. Receiver operating characteristic curve and logistic regression analyses were used.

RESULTS

Totally, 56 texture features for the basal ganglia and 46 features for the thalami were significantly different between the HIE groups on the ADC maps, T₂-, and T₂ weighted sequences. Using a Histogram_entropy log-10 value as >1.8 from the basal ganglia on the ADC maps (p < 0.001; OR, 266) and the absence of hyperintensity of the PLIC on T₁ weighted images (p = 0.012; OR, 17.11) were found as independent predictors for moderate-to-severe HIE. Using only a Histogram_entropy log-10 value had an equal diagnostic yield when compared to its combination with other texture features and imaging findings.

CONCLUSION

The Histogram_entropy log-10 value can be used as an indicator to differentiate from moderate-to-severe to mild HIE.

ADVANCES IN KNOWLEDGE

MRI-based TA may provide quantitative findings to indicate different stages in neonates with perinatal asphyxia.

Fluorine-18 fluorodeoxyglucose positron emission tomography-based textural features for prediction of event prone carotid atherosclerotic plaques

BACKGROUND

Texture analysis has been increasingly used in the field of positron emission tomography (PET)/computed tomography (CT) imaging with Fluorine-18 fluorodeoxyglucose (18F-FDG), aiming at assessing tumor heterogeneity. The purpose of the present study is to examine the feasibility of performing texture analysis in carotid arteries, investigate the value of textural features as predictors of potential plaque vulnerability using as reference standards histological and immunohistochemical data and compare their performance with conventional uptake measurements.

METHODS

67 different 18F-FDG PET-based textural features were extracted from carotid images of 21 patients with high-grade carotid stenosis undergoing endarterectomy. To identify the more reliable predictors, univariate logistic regression analysis was performed. The accuracy was satisfactory in case of an Area Under the Receiver Operating Characteristic (ROC) curve (AUC) ≥ 0.80.

RESULTS

First measure of information correlation (AUC = 0.87, P < 0.001), large zone low gray level emphasis (AUC = 0.87, P < 0.001), and normalized run length non-uniformity (AUC = 0.84, P < 0.001) were the most optimal textural features for identifying characteristics of plaque vulnerability based on histological analysis. Addition of textural features to target-to-background ratio (TBR) (AUC = 0.74, P = 0.031) resulted in an AUC = 0.92 (P < 0.001), however, this did not reach statistical significance (Pdiff = 0.09). Intensity histogram standard deviation (AUC = 0.87, P < 0.001) and joint variance (AUC = 0.81, P = 0.001) were the most efficient features for signal differential in relation to immunohistochemical findings and provided incremental value compared to TBR (Pdiff = 0.02).

CONCLUSION

Texture analysis can be applied in 18F-FDG PET carotid imaging providing valuable information for plaque characterization.

MRI Radiomic Signature of White Matter Hyperintensities Is Associated With Clinical Phenotypes

OBJECTIVE

Neuroimaging measurements of brain structural integrity are thought to be surrogates for brain health, but precise assessments require dedicated advanced image acquisitions. By means of quantitatively describing conventional images, radiomic analyses hold potential for evaluating brain health. We sought to: (1) evaluate radiomics to assess brain structural integrity by predicting white matter hyperintensities burdens (WMH) and (2) uncover associations between predictive radiomic features and clinical phenotypes.

METHODS

We analyzed a multi-site cohort of 4,163 acute ischemic strokes (AIS) patients with T2-FLAIR MR images with total brain and WMH segmentations. Radiomic features were extracted from normal-appearing brain tissue (brain mask-WMH mask). Radiomics-based prediction of personalized WMH burden was done using ElasticNet linear regression. We built a radiomic signature of WMH with stable selected features predictive of WMH burden and then related this signature to clinical variables using canonical correlation analysis (CCA).

RESULTS

Radiomic features were predictive of WMH burden (R 2 = 0.855 ± 0.011). Seven pairs of canonical variates (CV) significantly correlated the radiomics signature of WMH and clinical traits with respective canonical correlations of 0.81, 0.65, 0.42, 0.24, 0.20, 0.15, and 0.15 (FDR-corrected p-values CV 1 - 6 < 0.001, p-value CV 7 = 0.012). The clinical CV1 was mainly influenced by age, CV2 by sex, CV3 by history of smoking and diabetes, CV4 by hypertension, CV5 by atrial fibrillation (AF) and diabetes, CV6 by coronary artery disease (CAD), and CV7 by CAD and diabetes.

CONCLUSION

Radiomics extracted from T2-FLAIR images of AIS patients capture microstructural damage of the cerebral parenchyma and correlate with clinical phenotypes, suggesting different radiographical textural abnormalities per cardiovascular risk profile. Further research could evaluate radiomics to predict the progression of WMH and for the follow-up of stroke patients' brain health.

CT Radiomics in Colorectal Cancer: Detection of KRAS Mutation Using Texture Analysis and Machine Learning

In this work, by using descriptive techniques, the characteristics of the texture of the CT (computed tomography) image of patients with colorectal cancer were extracted and, subsequently, classified in KRAS+ or KRAS-. This was accomplished by using different classifiers, such as Support Vector Machine (SVM), Grading Boosting Machine (GBM), Neural Networks (NNET), and Random Forest (RF). Texture analysis can provide a quantitative assessment of tumour heterogeneity by analysing both the distribution and relationship between the pixels in the image. The objective of this research is to demonstrate that CT-based Radiomics can predict the presence of mutation in the KRAS gene in colorectal cancer. This is a retrospective study, with 47 patients from the University Hospital, with a confirmatory pathological analysis of KRAS mutation. The highest accuracy and kappa achieved were 83% and 64.7%, respectively, with a sensitivity of 88.9% and a specificity of 75.0%, achieved by the NNET classifier using the texture feature vectors combining wavelet transform and Haralick coefficients. The fact of being able to identify the genetic expression of a tumour without having to perform either a biopsy or a genetic test is a great advantage, because it prevents invasive procedures that involve complications and may present biases in the sample. As well, it leads towards a more personalized and effective treatment.

Texture analysis based on ADC maps and T2-FLAIR images for the assessment of the severity and prognosis of ischaemic stroke

OBJECTIVES

To explore the feasibility of texture analysis based on T2-weighted fluid-attenuated inversion recovery (T2-FLAIR) images and apparent diffusion coefficient (ADC) maps in the assessment of the severity and prognosis of ischaemic stroke using the National Institutes of Health Stroke Scale (NIHSS) and modified Rankin scale (mRS) scores, respectively.

METHODS

Overall, 116 patients diagnosed with subacute ischaemic stroke were included in this retrospective study. Based on T2-FLAIR images and ADC maps, 15 texture features were extracted from the ROIs of each patient using grey-level co-occurrence matrix (GLCM) and local binary pattern histogram Fourier (LBP-HF) methods. The correlations of NIHSS score on admission (NIHSS_baseline), NIHSS score 24 h after stroke onset (NIHSS_24h) and mRS score with the texture features were evaluated using Spearman's partial correlations. The receiver operating characteristic (ROC) curve was used to compare the performance of the selected texture features in the evaluation of stroke severity and prognosis.

RESULTS

Texture features derived from the T2-FLAIR images and ADC maps were correlated with NIHSS score and mRS score. Entropy_ADC and 0.75Quantile_T2-FLAIR showed the best diagnostic performance for assessing stroke severity. The combination of Entropy_ADC and 0.75Quantile_T2-FLAIR achieved a better performance in the evaluation of stroke severity (AUC = 0.7, p = 0.01) than either feature alone. Only 0.05Quantile_T2-FLAIR was found to be correlated with mRS score, and none of the texture features were predictive of mRS score.

CONCLUSION

Texture features derived from T2-FLAIR images and ADC maps might serve as biomarkers to evaluate stroke severity, but were insufficient to predict stroke prognosis.

Structure and function of the perivascular fluid compartment and vertebral venous plexus: Illumining a novel theory on mechanisms underlying the pathogenesis of Alzheimer's, cerebral small vessel, and neurodegenerative diseases

Blood dynamically and richly supplies the cerebral tissue via microvessels invested in pia matter perforating the cerebral substance. Arteries penetrating the cerebral substance derive an investment from one or two successive layers of pia mater, luminally apposed to the pial-glial basal lamina of the microvasculature and abluminally apposed to a series of aquaporin IV-studded astrocytic end feet constituting the soi-disant glia limitans. The full investment of successive layers forms the variably continuous walls of the periarteriolar, pericapillary, and perivenular divisions of the perivascular fluid compartment. The pia matter disappears at the distal periarteriolar division of the perivascular fluid compartment. Plasma from arteriolar blood sequentially transudates into the periarteriolar division of the perivascular fluid compartment and subarachnoid cisterns in precession to trickling into the neural interstitium. Fluid from the neural interstitium successively propagates into the venules through the subarachnoid cisterns and perivenular division of the perivascular fluid compartment. Fluid fluent within the perivascular fluid compartment flows gegen the net direction of arteriovenular flow. Microvessel oscillations at the central tendency of the cerebral vasomotion generate corresponding oscillations of within the surrounding perivascular fluid compartment, interposed betwixt the abluminal surface of the vessels and internal surface of the pia mater. The precise microanatomy of this most fascinating among designable spaces has eluded the efforts of various investigators to interrogate its structure, though most authors non-consensusly concur the investing layers effectively and functionally segregate the perivascular and subarachnoid fluid compartments. Enlargement of the perivascular fluid compartment in a variety of neurological disorders, including senile dementia of the Alzheimer's type and cerebral small vessel disease, may alternately or coordinately constitute a correlative marker of disease severity and a possible cause implicated in the mechanistic pathogenesis of these conditions. Venular pressures modulating oscillatory dynamic flow within the perivascular fluid compartment may similarly contribute to the development of a variety among neurological disorders. An intimate understanding of subtle features typifying microanatomy and microphysiology of the investing structures and spaces of the cerebral microvasculature may powerfully inform mechanistic pathophysiology mediating a variety of neurovascular ischemic, neuroinfectious, neuroautoimmune, and neurodegenerative diseases.

Parkinson's Disease Diagnosis Using Neostriatum Radiomic Features Based on T2-Weighted Magnetic Resonance Imaging

Background: Parkinson's disease (PD) is a neurodegenerative disease in which the neostriatum, including the caudate nucleus (CN) and putamen (PU), has an important role in the pathophysiology. However, conventional magnetic resonance imaging (MRI) lacks sufficient specificity to diagnose PD. Therefore, the study's aim was to investigate the feasibility of using a radiomics approach to distinguish PD patients from healthy controls on T2-weighted images of the neostriatum and provide a basis for the clinical diagnosis of PD.

Methods: T2-weighted images from 69 PD patients and 69 age- and sex-matched healthy controls were obtained on the same 3.0T MRI scanner. Regions of interest (ROIs) were manually placed at the CN and PU on the slices showing the largest respective sizes of the CN and PU. We extracted 274 texture features from each ROI and then used the least absolute shrinkage and selection operator regression to perform feature selection and radiomics signature building to identify the CN and PU radiomics signatures consisting of optimal features. We used a receiver operating characteristic curve analysis to assess the diagnostic performance of two radiomics signatures in a training group and estimate the generalization performance in the test group.

Results: There were no significant differences in the demographic and clinical characteristics between the PD patients and healthy controls. The CN and PU radiomics signatures were built using 12 and 7 optimal features, respectively. The performance of the two radiomics signatures to distinguish PD patients from healthy controls was good. In the training and test groups, the AUCs of the CN radiomics signatures were 0.9410 (95% confidence interval [CI]: 0.8986–0.9833) and 0.7732 (95% CI: 0.6292–0.9173), respectively, and the AUCs of the PU radiomics signature were 0.8767 (95% CI: 0.8066–0.9469) and 0.7143 (95% CI: 0.5540–0.8746), respectively. Vertl_GlevNonU_R appeared simultaneously in both the CN and PU radiomics signatures as an optimal feature. A t-test analysis revealed significantly higher levels of texture values of the CN and PU in the PD patients than healthy controls (P < 0.05).

Conclusion: Neostriatum radiomics signatures achieved good diagnostic performance for PD and potentially could serve as a basis for the clinical diagnosis of PD.

Radiomics from magnetic resonance imaging may be used to predict the progression of white matter hyperintensities and identify associated risk factors

OBJECTIVE

The progression of white matter hyperintensities (WMH) varies considerably in adults. In this study, we aimed to predict the progression and related risk factors of WMH based on the radiomics of whole-brain white matter (WBWM).

METHODS

A retrospective analysis was conducted on 141 patients with WMH who underwent two consecutive brain magnetic resonance (MR) imaging sessions from March 2014 to May 2018. The WBWM was segmented to extract and score the radiomics features at baseline. Follow-up images were evaluated using the modified Fazekas scale, with progression indicated by scores ≥ 1. Patients were divided into progressive (n = 65) and non-progressive (n = 76) groups. The progressive group was subdivided into any WMH (AWMH), periventricular WMH (PWMH), and deep WMH (DWMH). Independent risk factors were identified using logistic regression.

RESULTS

The area under the curve (AUC) values for the radiomics signatures of the training sets were 0.758, 0.749, and 0.775 for AWMH, PWMH, and DWMH, respectively. The AUC values of the validation set were 0.714, 0.697, and 0.717, respectively. Age and hyperlipidemia were independent predictors of progression for AWMH. Age and body mass index (BMI) were independent predictors of progression for DWMH, while hyperlipidemia was an independent predictor of progression for PWMH. After combining clinical factors and radiomics signatures, the AUC values were 0.848, 0.863, and 0.861, respectively, for the training set, and 0.824, 0.818, and 0.833, respectively, for the validation set.

CONCLUSIONS

MRI-based radiomics of WBWM, along with specific risk factors, may allow physicians to predict the progression of WMH.

KEY POINTS

• Radiomics features detected by magnetic resonance imaging may be used to predict the progression of white matter hyperintensities. • Radiomics may be used to identify risk factors associated with the progression of white matter hyperintensities. • Radiomics may serve as non-invasive biomarkers to monitor white matter status.

Analysis of dynamic texture and spatial spectral descriptors of dynamic contrast-enhanced brain magnetic resonance images for studying small vessel disease

Cerebral small vessel disease (SVD) comprises various pathological processes affecting small brain vessels and damaging white and grey matter. In this paper, we propose a framework comprising region of interest sampling, dynamic spectral and texture description, functional principal component analysis, and statistical analysis to study exogenous contrast agent distribution over time in various brain regions in patients with recent mild stroke and SVD features.We compared our results against current semi-quantitative surrogates of dysfunction such as signal enhancement area and slope. Biological sex, stroke lesion type and overall burden of white matter hyperintensities (WMH) were significant predictors of intensity, spectral, and texture features extracted from the ventricular region (p-value < 0.05), explaining between a fifth and a fourth of the data variance (0.20 ≤Adj.R² ≤ 0.25). We observed that spectral feature reflected more the dysfunction compared to other descriptors since the overall WMH burden explained consistently the power spectra variability in blood vessels, cerebrospinal fluid, deep grey matter and white matter. Our preliminary results show the potential of the framework for the analysis of dynamic contrast-enhanced brain magnetic resonance imaging acquisitions in SVD since significant variation in our metrics was related to the burden of SVD features. Therefore, our proposal may increase sensitivity to detect subtle features of small vessel dysfunction. A public version of the code will be released on our research website.

Radiomics nomogram based on MRI for predicting white matter hyperintensity progression in elderly adults

BACKGROUND

White matter hyperintensity (WMH) is widely observed in aging brain and is associated with various diseases. A pragmatic and handy method in the clinic to assess and follow up white matter disease is strongly in need.

PURPOSE

To develop and validate a radiomics nomogram for the prediction of WMH progression.

STUDY TYPE

Retrospective.

POPULATION

Brain images of 193 WMH patients from the Picture Archiving and Communication Systems (PACS) database in the A Medical Center (Zhejiang Provincial People's Hospital). MRI data of 127 WMH patients from the PACS database in the B Medical Center (Zhejiang Lishui People's Hospital) were included for external validation. All of the patients were at least 60 years old.

FIELD STRENGTH/SEQUENCE

T₁ -fluid attenuated inversion recovery images were acquired using a 3T scanner.

ASSESSMENT

WMH was evaluated utilizing the Fazekas scale based on MRI. WMH progression was assessed with a follow-up MRI using a visual rating scale. Three neuroradiologists, who were blinded to the clinical data, assessed the images independently. Moreover, interobserver and intraobserver reproducibility were performed for the regions of interest for segmentation and feature extraction.

STATISTICAL TESTS

A receiver operating characteristic (ROC) curve, the area under the curve (AUC) of the ROC was calculated, along with sensitivity and specificity. Also, a Hosmer-Lemeshow test was performed.

RESULTS

The AUC of radiomics signature in the primary, internal validation cohort, external validation cohort were 0.886, 0.816, and 0.787, respectively; the specificity were 71.79%, 72.22%, and 81%, respectively; the sensitivity were 92.68%, 87.94% and 78.3%, respectively. The radiomics nomogram in the primary cohort (AUC = 0.899) and the internal validation cohort (AUC = 0.84). The Hosmer-Lemeshow test showed no significant difference between the primary cohort and the internal validation cohort (P > 0.05). The AUC of the radiomics nomogram, radiomics signature, and hyperlipidemia in all patients from the primary and internal validation cohort was 0.878, 0.848, and 0.626, respectively.

DATA CONCLUSION

This multicenter study demonstrated the use of a radiomics nomogram in predicting the progression of WMH with elderly adults (an age of at least 60 years) based on conventional MRI.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2020;51:535-546.

Identification of the presence of ischaemic stroke lesions by means of texture analysis on brain magnetic resonance images

•

Radiomics in conventionally segmented tissues can identify MRI scans that had a stroke.

•

Patient’s advanced age can negatively influence classification results.

•

Feature selection and stroke subtype influence but do not determine accuracy.

•

Stroke subtype cannot be ascertained from texture analysis in brain tissues.

Background

The differential quantification of brain atrophy, white matter hyperintensities (WMH) and stroke lesions is important in studies of stroke and dementia. However, the presence of stroke lesions is usually overlooked by automatic neuroimage processing methods and the-state-of-the-art deep learning schemes, which lack sufficient annotated data. We explore the use of radiomics in identifying whether a brain magnetic resonance imaging (MRI) scan belongs to an individual that had a stroke or not.

Materials and methods

We used 1800 3D sets of MRI data from three prospective studies: one of stroke mechanisms and two of cognitive ageing, evaluated 114 textural features in WMH, cerebrospinal fluid, deep grey and normal-appearing white matter, and attempted to classify the scans using a random forest and support vector machine classifiers with and without feature selection. We evaluated the discriminatory power of each feature independently in each population and corrected the result against Type 1 errors. We also evaluated the influence of clinical parameters in the classification results.

Results

Subtypes of ischaemic strokes (i.e. lacunar vs. cortical) cannot be discerned using radiomics, but the presence of a stroke-type lesion can be ascertained with accuracies ranging from 0.7 < AUC < 0.83. Feature selection, tissue type, stroke subtype and MRI sequence did not seem to determine the classification results. From all clinical variables evaluated, age correlated with the proportion of images classified correctly using either different or the same descriptors (Pearson r = 0.31 and 0.39 respectively, p < 0.001).

Conclusions

Texture features in conventionally automatically segmented tissues may help in the identification of the presence of previous stroke lesions on an MRI scan, and should be taken into account in transfer learning strategies of the-state-of-the-art deep learning schemes.

Assessment of cerebrovascular reserve impairment using the breath-holding index in patients with leukoaraiosis

Many studies have demonstrated that leukoaraiosis is associated with impaired cerebrovascular reserve function. However, the definitive hemodynamic changes that occur in leukoaraiosis are not clear, and there are many controversies. This study aimed to investigate hemodynamic changes in symptomatic leukoaraiosis using transcranial Doppler ultrasonography and the breath-holding test in a Chinese Han population, from northern China. A total of 203 patients who were diagnosed with ischemic stroke or clinical chronic progressive ischemic symptoms were enrolled in this study, including 97 males and 106 females, with an age range of 43–93 years. The severity of leukoaraiosis was evaluated according to the Fazekas grading scale, and patients were divided into four groups accordingly. Grade 0 was no leukoaraiosis, and grades I, II, and III were mild, moderate, and severe leukoaraiosis, respectively, with 44, 79, 44, and 36 cases in each group. Transcranial Doppler ultrasonography and the breath-holding test were performed. The mean blood flow velocity of the bilateral middle cerebral artery was measured and the breath-holding index was calculated. The breath holding index was correlated with leukoaraiosis severity and cognitive impairment. Patients with a low breath holding index presented poor performance in the Montreal Cognitive Assessment (MoCA) and executive function tests. That is, the lower the breath holding index, the lower the scores for the MoCA and the higher for the trail-making test Parts A and B. These results indicate that the breath-holding index is a useful parameter for the evaluation of cerebrovascular reserve impairment in patients with leukoaraiosis. In addition, the breath-holding index can reflect cognitive dysfunction, providing a new insight into the pathophysiology of leukoaraiosis. This study was approved by the Ethics Committee of the Fifth People’s Hospital of Shenyang, China (approval No. 20160301) and registered in the Chinese Clinical Trial Registry (registration number: ChiCTR1800014421).

ALTEA: A Software Tool for the Evaluation of New Biomarkers for Alzheimer's Disease by Means of Textures Analysis on Magnetic Resonance Images

The current criteria for diagnosing Alzheimer’s disease (AD) require the presence of relevant cognitive deficits, so the underlying neuropathological damage is important by the time the diagnosis is made. Therefore, the evaluation of new biomarkers to detect AD in its early stages has become one of the main research focuses. The purpose of the present study was to evaluate a set of texture parameters as potential biomarkers of the disease. To this end, the ALTEA (ALzheimer TExture Analyzer) software tool was created to perform 2D and 3D texture analysis on magnetic resonance images. This intuitive tool was used to analyze textures of circular and spherical regions situated in the right and left hippocampi of a cohort of 105 patients: 35 AD patients, 35 patients with early mild cognitive impairment (EMCI) and 35 cognitively normal (CN) subjects. A total of 25 statistical texture parameters derived from the histogram, the Gray-Level Co-occurrence Matrix and the Gray-Level Run-Length Matrix, were extracted from each region and analyzed statistically to study their predictive capacity. Several textural parameters were statistically significant (p < 0.05) when differentiating AD subjects from CN and EMCI patients, which indicates that texture analysis could help to identify the presence of AD.

Texture Analysis of T1-Weighted and Fluid-Attenuated Inversion Recovery Images Detects Abnormalities That Correlate With Cognitive Decline in Small Vessel Disease

Supplemental Digital Content is available in the text.

Background and Purpose—

Magnetic resonance imaging may be useful to assess disease severity in cerebral small vessel disease (SVD), identify those individuals who are most likely to progress to dementia, monitor disease progression, and act as surrogate markers to test new therapies. Texture analysis extracts information on the relationship between signal intensities of neighboring voxels. A potential advantage over techniques, such as diffusion tensor imaging, is that it can be used on clinically obtained magnetic resonance sequences. We determined whether texture parameters (TP) were abnormal in SVD, correlated with cognitive impairment, predicted cognitive decline, or conversion to dementia.

Methods—

In the prospective SCANS study (St George’s Cognition and Neuroimaging in Stroke), we assessed TP in 121 individuals with symptomatic SVD at baseline, 99 of whom attended annual cognitive testing for 5 years. Conversion to dementia was recorded for all subjects during the 5-year period. Texture analysis was performed on fluid-attenuated inversion recovery and T1-weighted images. The TP obtained from the SVD cohort were cross-sectionally compared with 54 age-matched controls scanned on the same magnetic resonance imaging system.

Results—

There were highly significant differences in several TP between SVD cases and controls. Within the SVD population, TP were highly correlated to other magnetic resonance imaging parameters (brain volume, white matter lesion volume, lacune count). TP correlated with executive function and global function at baseline and predicted conversion to dementia, after controlling for age, sex, premorbid intelligence quotient, and magnetic resonance parameters.

Conclusions—

TP, which can be obtained from routine clinical images, are abnormal in SVD, and the degree of abnormality correlates with executive dysfunction and global cognition at baseline and decline during 5 years. TP may be useful to assess disease severity in clinically collected data. This needs testing in data clinically acquired across multiple sites.