StAR: a simple tool for the statistical comparison of ROC curves

ELW-CNN: An extremely lightweight convolutional neural network for enhancing interoperability in colon and lung cancer identification using explainable AI

Cancer is a condition in which cells in the body grow uncontrollably, often forming tumours and potentially spreading to various areas of the body. Cancer is a hazardous medical case in medical history analysis. Every year, many people die of cancer at an early stage. Therefore, it is necessary to accurately and early identify cancer to effectively treat and save human lives. However, various machine and deep learning models are effective for cancer identification. Therefore, the effectiveness of these efforts is limited by the small dataset size, poor data quality, interclass changes between lung squamous cell carcinoma and adenocarcinoma, difficulties with mobile device deployment, and lack of image and individual-level accuracy tests. To overcome these difficulties, this study proposed an extremely lightweight model using a convolutional neural network that achieved 98.16% accuracy for a large lung and colon dataset and individually achieved 99.02% for lung cancer and 99.40% for colon cancer. The proposed lightweight model used only 70 thousand parameters, which is highly effective for real-time solutions. Explainability methods such as Grad-CAM and symmetric explanation highlight specific regions of input data that affect the decision of the proposed model, helping to identify potential challenges. The proposed models will aid medical professionals in developing an automated and accurate approach for detecting various types of colon and lung cancer.

Discriminating Parkinson's disease patients from healthy controls using nasal respiratory airflow

BACKGROUND

Breathing patterns may inform on health. We note that the sites of earliest brain damage in Parkinson's disease (PD) house the neural pace-makers of respiration. We therefore hypothesized that ongoing long-term temporal dynamics of respiration may be altered in PD.

METHODS

We applied a wearable device that precisely logs nasal airflow over time in 28 PD patients (mostly H&Y stage-II) and 33 matched healthy controls. Each participant wore the device for 24 h of otherwise routine daily living.

RESULTS

We observe significantly altered temporal patterns of nasal airflow in PD, where inhalations are longer and less variable than in matched controls (mean PD = -1.22 ± 1.9 (combined respiratory features score), Control = 1.04 ± 2.16, Wilcoxon rank-sum test, z = -4.1, effect size Cliff's δ = -0.61, 95% confidence interval = -0.79 - (-0.34), P = 4.3 × 10-5). The extent of alteration is such that using only 30 min of recording we detect PD at 87% accuracy (AUC = 0.85, 79% sensitivity (22 of 28), 94% specificity (31 of 33), z = 5.7, p = 3.5 × 10-9), and also predict disease severity (correlation with UPDRS-Total score: r = 0.49; P = 0.008).

CONCLUSIONS

We conclude that breathing patterns are altered by H&Y stage-II in the disease cascade, and our methods may be further refined in the future to provide an indication with diagnostic and prognostic value.

Utility of diffusion-weighted imaging in differentiating benign and malignant breast lesions

Background

Breast cancer presents a significant global health burden. An accurate differentiation between benign and malignant lesions is imperative for timely intervention. While dynamic contrast enhanced MRI (DCE-MRI) is highly sensitive, its specificity is limited. This has led to the exploration of diffusion-weighted imaging (DWI) in distinguishing between benign and malignant breast lesions.

Objectives

The study aimed to explore the diagnostic utility of DWI in distinguishing between benign and malignant breast lesions.

Method

Assessment of 38 breast lesions using DWI with a b value of 800 s/mm2, performed with 3 Tesla MRI. The diagnostic performance of two different region of Interest (ROI) placement approaches was compared to obtain a feasible cut-off value of apparent diffusion coefficient (ADC) to differentiate between malignant and benign lesions. The histopathological reports were used as the gold standard.

Results

ADC values of malignant lesions were significantly lower than those of benign lesions (0.84 × 10-3 mm2/s vs. 1.54 × 10-3 mm2/s). The average ADC measured using a small-sized 2D ROI including the darkest part in the ADC map, performed better than the large 2D ROI covering the entire lesion.

Conclusion

Using a cut-off value of 0.98 × 10-3 mm2/s, ADC obtained high sensitivity (90%) and specificity (88.9%) in distinguishing between benign and malignant breast lesions.

Contribution

Utilising quantitative analysis of DWI with ADC value measurement, reliably distinguished between benign and malignant breast lesions in this cohort, especially when employing a higher b value of 800 s/mm2.

Validation of the NUT CRACKER Diagnostic Algorithm and Prediction for Cashew and Pistachio Co-Allergy

BACKGROUND

Because of the high cross-sensitization among tree nuts, the NUT CRACKER (Nut Co-reactivity-Acquiring Knowledge for Elimination Recommendations) study proposed a diagnostic algorithm to minimize the number of required oral food challenges (OFCs).

OBJECTIVE

To validate the algorithm for cashew and pistachio allergy and determine markers for allergic severity.

METHODS

Patients (n = 125) with a median age of 7.8 (interquartile range, 5.9-11.2) years with suspected tree nut allergy were evaluated prospectively with decision tree points on the basis of skin prick test (SPT), basophil activation test (BAT), and knowledge of the coincidence of allergies. Validation of allergic status was determined by OFC. Markers of clinical severity were evaluated using the combined original and prospective cohort (n = 187) in relationship to SPT, BAT, and Ana o 3-sIgE.

RESULTS

Reactivity to cashew in SPT, BAT, and Ana o 3-sIgE and the incidence of abdominal pain on challenge were significantly higher in dual-allergic cashew/pistachio patients (n = 82) versus single cashew allergic patients (n = 18) (P = .001). All 3 diagnostic tests showed significant inverse correlation with log10 reaction doses for positive cashew OFC. The algorithm reduced overall the total number of OFCs by 72.0%, with a positive predictive value and negative predictive value of 93.0% and 99.0%, respectively. Cashew false-positives were observed primarily in hazelnut-allergic patients (P = .026). In this population, Ana o 3-specific IgE could diagnose cashew allergy with a sensitivity of more than 90% and a specificity of more than 95%.

CONCLUSIONS

The NUT CRACKER diagnostic algorithm was validated and reduced the number of diagnostic OFCs required. Markers for severity phenotypes may guide oral immunotherapy protocols, improving the risk/benefit ratio for patients.

Machine learning-based network intrusion detection for big and imbalanced data using oversampling, stacking feature embedding and feature extraction

Cybersecurity has emerged as a critical global concern. Intrusion Detection Systems (IDS) play a critical role in protecting interconnected networks by detecting malicious actors and activities. Machine Learning (ML)-based behavior analysis within the IDS has considerable potential for detecting dynamic cyber threats, identifying abnormalities, and identifying malicious conduct within the network. However, as the number of data grows, dimension reduction becomes an increasingly difficult task when training ML models. Addressing this, our paper introduces a novel ML-based network intrusion detection model that uses Random Oversampling (RO) to address data imbalance and Stacking Feature Embedding based on clustering results, as well as Principal Component Analysis (PCA) for dimension reduction and is specifically designed for large and imbalanced datasets. This model’s performance is carefully evaluated using three cutting-edge benchmark datasets: UNSW-NB15, CIC-IDS-2017, and CIC-IDS-2018. On the UNSW-NB15 dataset, our trials show that the RF and ET models achieve accuracy rates of 99.59% and 99.95%, respectively. Furthermore, using the CIC-IDS2017 dataset, DT, RF, and ET models reach 99.99% accuracy, while DT and RF models obtain 99.94% accuracy on CIC-IDS2018. These performance results continuously outperform the state-of-art, indicating significant progress in the field of network intrusion detection. This achievement demonstrates the efficacy of the suggested methodology, which can be used practically to accurately monitor and identify network traffic intrusions, thereby blocking possible threats.

NORHA: A NORmal Hippocampal Asymmetry Deviation Index Based on One-Class Novelty Detection and 3D Shape Features

Radiologists routinely analyze hippocampal asymmetries in magnetic resonance (MR) images as a biomarker for neurodegenerative conditions like epilepsy and Alzheimer's Disease. However, current clinical tools rely on either subjective evaluations, basic volume measurements, or disease-specific models that fail to capture more complex differences in normal shape. In this paper, we overcome these limitations by introducing NORHA, a novel NORmal Hippocampal Asymmetry deviation index that uses machine learning novelty detection to objectively quantify it from MR scans. NORHA is based on a One-Class Support Vector Machine model learned from a set of morphological features extracted from automatically segmented hippocampi of healthy subjects. Hence, in test time, the model automatically measures how far a new unseen sample falls with respect to the feature space of normal individuals. This avoids biases produced by standard classification models, which require being trained using diseased cases and therefore learning to characterize changes produced only by the ones. We evaluated our new index in multiple clinical use cases using public and private MRI datasets comprising control individuals and subjects with different levels of dementia or epilepsy. The index reported high values for subjects with unilateral atrophies and remained low for controls or individuals with mild or severe symmetric bilateral changes. It also showed high AUC values for discriminating individuals with hippocampal sclerosis, further emphasizing its ability to characterize unilateral abnormalities. Finally, a positive correlation between NORHA and the functional cognitive test CDR-SB was observed, highlighting its promising application as a biomarker for dementia.

Predictive value of abbreviated olfactory tests in prodromal Parkinson disease

There is disagreement in the literature whether olfaction may show specific impairments in Parkinson Disease (PD) and if olfactory tests comprised of selected odors could be more specific for diagnosis. We sought to validate previously proposed subsets of the University of Pennsylvania Smell Identification Test (UPSIT) odors for predicting conversion to PD in an independent, prodromal cohort. Conversion to PD was assessed in 229 participants in the Parkinson At Risk Study who completed baseline olfactory testing with the UPSIT and up to 12 years of clinical and imaging evaluations. No commercially available or proposed subset performed better than the full 40-item UPSIT. The proposed "PD-specific" subsets also did not perform better than expected by chance. We did not find evidence for selective olfactory impairment in Parkinson disease. Shorter odor identification tests, including commercially available 10-12 item tests, may have utility for ease of use and cost, but not for superior predictive value.

Assessment of Diagnostic Performance of Dermatologists Cooperating With a Convolutional Neural Network in a Prospective Clinical Study: Human With Machine

Importance

Studies suggest that convolutional neural networks (CNNs) perform equally to trained dermatologists in skin lesion classification tasks. Despite the approval of the first neural networks for clinical use, prospective studies demonstrating benefits of human with machine cooperation are lacking.

Objective

To assess whether dermatologists benefit from cooperation with a market-approved CNN in classifying melanocytic lesions.

Design, Setting, and Participants

In this prospective diagnostic 2-center study, dermatologists performed skin cancer screenings using naked-eye examination and dermoscopy. Dermatologists graded suspect melanocytic lesions by the probability of malignancy (range 0-1, threshold for malignancy ≥0.5) and indicated management decisions (no action, follow-up, excision). Next, dermoscopic images of suspect lesions were assessed by a market-approved CNN, Moleanalyzer Pro (FotoFinder Systems). The CNN malignancy scores (range 0-1, threshold for malignancy ≥0.5) were transferred to dermatologists with the request to re-evaluate lesions and revise initial decisions in consideration of CNN results. Reference diagnoses were based on histopathologic examination in 125 (54.8%) lesions or, in the case of nonexcised lesions, on clinical follow-up data and expert consensus. Data were collected from October 2020 to October 2021.

Main Outcomes and Measures

Primary outcome measures were diagnostic sensitivity and specificity of dermatologists alone and dermatologists cooperating with the CNN. Accuracy and receiver operator characteristic area under the curve (ROC AUC) were considered as additional measures.

Results

A total of 22 dermatologists detected 228 suspect melanocytic lesions (190 nevi, 38 melanomas) in 188 patients (mean [range] age, 53.4 [19-91] years; 97 [51.6%] male patients). Diagnostic sensitivity and specificity significantly improved when dermatologists additionally integrated CNN results into decision-making (mean sensitivity from 84.2% [95% CI, 69.6%-92.6%] to 100.0% [95% CI, 90.8%-100.0%]; P = .03; mean specificity from 72.1% [95% CI, 65.3%-78.0%] to 83.7% [95% CI, 77.8%-88.3%]; P < .001; mean accuracy from 74.1% [95% CI, 68.1%-79.4%] to 86.4% [95% CI, 81.3%-90.3%]; P < .001; and mean ROC AUC from 0.895 [95% CI, 0.836-0.954] to 0.968 [95% CI, 0.948-0.988]; P = .005). In addition, the CNN alone achieved a comparable sensitivity, higher specificity, and higher diagnostic accuracy compared with dermatologists alone in classifying melanocytic lesions. Moreover, unnecessary excisions of benign nevi were reduced by 19.2%, from 104 (54.7%) of 190 benign nevi to 84 nevi when dermatologists cooperated with the CNN (P < .001). Most lesions were examined by dermatologists with 2 to 5 years (96, 42.1%) or less than 2 years of experience (78, 34.2%); others (54, 23.7%) were evaluated by dermatologists with more than 5 years of experience. Dermatologists with less dermoscopy experience cooperating with the CNN had the most diagnostic improvement compared with more experienced dermatologists.

Conclusions and Relevance

In this prospective diagnostic study, these findings suggest that dermatologists may improve their performance when they cooperate with the market-approved CNN and that a broader application of this human with machine approach could be beneficial for dermatologists and patients.

Muscle atonia index during multiple sleep latency test: A possible marker to differentiate narcolepsy from other hypersomnias

OBJECTIVE

The complexity and delay of the diagnosis of narcolepsy require several diagnostic tests and invasive procedures such as lumbar puncture. Our study aimed to determine the changes in muscle tone (atonia index, AI) at different levels of vigilance during the entire multiple sleep latency test (MSLT) and each nap in people with narcolepsy type 1 (NT1) and 2 (NT2) compared with other hypersomnias and its potential diagnostic value.

METHODS

Twenty-nine patients with NT1 (11 M 18F, mean age 34.9 years, SD 16.8) and sixteen with NT2 (10 M 6F, mean age 39 years, SD 11.8) and 20 controls with other hypersomnias (10 M, 10F mean age 45.1 years, SD 15.1) were recruited. AI was evaluated at different levels of vigilance (Wake and REM sleep) in each nap and throughout the MSLT of each group. The validity of AI in identifying patients with narcolepsy (NT1 and NT2) was analyzed using receiver operating characteristic (ROC) curves.

RESULTS

AI during wakefulness (WAI) was significantly higher in the narcolepsy groups (NT1 and NT2 p < 0.001) compared to the hypersomniac group. AI during REM sleep (RAI) (p = 0.03) and WAI in nap with sudden onsets of REM sleep periods (SOREMP) (p = 0.001) were lower in NT1 than in NT2. The ROC curves showed high AUC values for WAI (NT1 0.88; Best Cut-off > 0.57, Sensitivity 79.3 % Specificity 90 %; NT2 0.89 Best Cut-off > 0.67 Sensitivity 87.5 % Specificity 95 %; NT1 and NT2 0.88 Best Cut-off > 0.57 Sensitivity 82.2 % Specificity 90 %) in discriminating subjects suffering from other hypersomnias. RAI and WAI in nap with SOREMP showed a poor AUC value (RAI AUC: 0.7 Best cutoff 0.7 Sensitivity 50 % Specificity 87.5 %; WAI in nap before SOREMP AUC: 0.66, Best cut-off < 0.82 sensitivity 61.9 % and specificity 67.35 %) differentiating NT1 and NT2.

CONCLUSIONS

WAI may represent an encouraging electrophysiological marker of narcolepsy and suggests a vulnerable tendency to dissociative wake / sleep dysregulation lacking in other forms of hypersomnia.

SIGNIFICANCE

AI during wakefulness may help distinguish between narcolepsy and other hypersomnias.

PEGALUS: predictivity of elderly age, arterial gas analysis, and lung ultrasound. A new prognostic score for COVID-19 patients in the emergency department-an observational prospective study

BACKGROUND

Periodic surges of COVID-19 patients seeking care in the hospital environment overwhelm systems reduce the availability of resources for treatment of non-COVID-19 cases (Zheng et al. in J Hosp Infect 106:325-329, 2020). Hospital flow and resource management could be greatly enhanced by differentiating patients who are likely at risk of adverse clinical outcomes from those who could safely be discharged after evaluation and managed outside of the hospital setting (Sun et al. in J Infect Dis 223:38-46, 2021). Herein, we propose a prognostic score named PEGALUS (Predictivity of Elderly age, arterial blood Gas Analysis and Lung UltraSound) that could potentially help clinicians properly and rapidly choose the appropriate allocation of COVID-19 patients admitted to the emergency department (ED).

METHODS

This observational prospective study enrolled COVID-19 patients who were admitted to the ED of IRCCS San Raffaele Hospital (HSR).

RESULTS

230 COVID-19 patients were enrolled and 30-day follow-up data was collected. Composite outcome was death or need for oro-tracheal intubation (OTI). 50 patients (21.5%) reached the outcome during the observational period. In multivariate Cox analysis, age, PO₂/FiO₂ ratio, pCO₂, duration of symptoms, and lung ultrasound evaluation were significantly associated with the adverse outcome. We obtained a new scorecard (PEGALUS) according to the hazard ratio of the identified predictors. PEGALUS score performed well in predicting the composite outcome (AUC 0.866, 95% IC 0.812-0.921; p < 0.001). Kaplan-Meier showed that a PEGALUS score < 7 was associated with a good 30-day prognosis (survival rate 97.5%), compared to a PEGALUS score of 7-11 (survival rate 85.9%; p log-rank 0.009) and PEGALUS score > 11 (survival rate 49.3%; p log-rank < 0.001).

CONCLUSIONS

PEGALUS score performed at the admission can predict adverse outcomes in patients with COVID-19. The systematic application of this score might permit a more accurate and rapid treatment allocation in this setting.

Direct Gene Expression Profile Prediction for Uveal Melanoma from Digital Cytopathology Images via Deep Learning and Salient Image Region Identification

Objective

To demonstrate that deep learning (DL) methods can produce robust prediction of gene expression profile (GEP) in uveal melanoma (UM) based on digital cytopathology images.

Design

Evaluation of a diagnostic test or technology.

Subjects Participants and Controls

Deidentified smeared cytology slides stained with hematoxylin and eosin obtained from a fine needle aspirated from UM.

Methods

Digital whole-slide images were generated by fine-needle aspiration biopsies of UM tumors that underwent GEP testing. A multistage DL system was developed with automatic region-of-interest (ROI) extraction from digital cytopathology images, an attention-based neural network, ROI feature aggregation, and slide-level data augmentation.

Main Outcome Measures

The ability of our DL system in predicting GEP on a slide (patient) level. Data were partitioned at the patient level (73% training; 27% testing).

Results

In total, our study included 89 whole-slide images from 82 patients and 121 388 unique ROIs. The testing set included 24 slides from 24 patients (12 class 1 tumors; 12 class 2 tumors; 1 slide per patient). Our DL system for GEP prediction achieved an area under the receiver operating characteristic curve of 0.944, an accuracy of 91.7%, a sensitivity of 91.7%, and a specificity of 91.7% on a slide-level analysis. The incorporation of slide-level feature aggregation and data augmentation produced a more predictive DL model (P = 0.0031).

Conclusions

Our current work established a complete pipeline for GEP prediction in UM tumors: from automatic ROI extraction from digital cytopathology whole-slide images to slide-level predictions. Our DL system demonstrated robust performance and, if validated prospectively, could serve as an image-based alternative to GEP testing.

Changing climate may mitigate the invasiveness risk of non-native salmonids in the Danube and Adriatic basins of the Balkan Peninsula (south-eastern Europe)

Salmonids are an extensively hatchery-reared group of fishes that have been introduced worldwide mainly for their high commercial and recreational value. The Balkan Peninsula (south-eastern Europe) is characterised by an outstanding salmonid diversity that has become threatened by the introduction of non-native salmonids whose potential risk of invasiveness in the region remains unknown and especially so under predicted climate change conditions. In this study, 13 extant and four horizon non-native salmonid species were screened for their risk of invasiveness in the Danube and Adriatic basins of four Balkan countries. Overall, six (35%) of the screened species were ranked as carrying a high risk of invasiveness under current climate conditions, whereas under predicted conditions of global warming, this number decreased to three (17%). Under current climate conditions, the very high risk (‘top invasive’) species were rainbow trout Oncorhynchus mykiss and brown trout Salmo trutta (sensu stricto), whereas under predicted climate change, this was true only of O. mykiss. A high risk was also attributed to horizon vendace Coregonus albula and lake charr Salvelinus namaycush, and to extant Atlantic salmon Salmo salar and brook trout Salvelinus fontinalis, whose risk of invasiveness, except for S. fontinalis, decreased to medium. For the other eleven medium-risk species, the risk score decreased under predicted climate change, but still remained medium. The outcomes of this study reveal that global warming will influence salmonids and that only species with wider temperature tolerance, such as O. mykiss will likely prevail. It is anticipated that the present results may contribute to the implementation of appropriate management plans to prevent the introduction and translocation of non-native salmonids across the Balkan Peninsula. Additionally, adequate measures should be developed for aquaculture facilities to prevent escapees of non-native salmonids with a high risk of invasiveness, especially into recipient areas of high conservation value.

Risk of invasiveness of non-native fishes in the South Caucasus biodiversity and geopolitical hotspot

Aquatic invasions are one of the major threats for freshwater ecosystems. However, in developing countries, knowledge of biological invasions, essential for the implementation of appropriate legislation, is often limited if not entirely lacking. In this regard, the identification of potentially invasive non-native species by risk screening, followed by a full risk assessment of the species ranked as higher risk, enables decision-makers to be informed about the extent of the threats posed to the recipient (risk assessment) area. In this study, 32 non-native extant and horizon fish species were screened for their risk of invasiveness under current and predicted climate conditions for the South Caucasus – a biodiversity and geopolitical hotspot that includes the countries of Armenia, Azerbaijan and Georgia. Overall, the number of very high-risk species increased from four (12.5%) under current climate conditions to 12 (37.5%) under predicted climate conditions. The highest-risk species under both conditions included the already established gibel carp Carassius gibelio and topmouth gudgeon Pseudorasbora parva, the locally translocated pikeperch Sander lucioperca and the horizon North African catfish Clarias gariepinus. Under predicted climate conditions, a very high risk of invasiveness was predicted also for the translocated three-spined stickleback Gasterosteus aculeatus and Eurasian perch Perca fluviatilis, for the already established eastern mosquitofish Gambusia holbrooki, ruffe Gymnocephalus cernua, sharpbelly Hemiculter leucisculus and Nile tilapia Orechromis niloticus, and for the horizon pumpkinseed Lepomis gibbosus and largemouth bass Micropterus salmoides. Future research on the non-native species in the South Caucasus should be conducted both country- and region-wide and should account not only for the high biodiversity, but also for the critical geopolitical situation affecting the study area.

A protocol for screening potentially invasive non-native species using Weed Risk Assessment-type decision-support tools

There is increasing use worldwide of electronic decision-support tools to identify potentially invasive non-native species so as to inform policy and management decisions aimed at preventing or mitigating the environmental and socio-economic impacts of biological invasions. This study reviews the analytical approaches used to calibrate scores generated by the Weed Risk Assessment and subsequent adaptations thereof and provides a protocol for: (i) the identification of the assessor(s) who will carry out the screenings; (ii) the definition of the risk assessment area; (iii) the criteria for selection of the species for screening; and (iv) the a priori categorisation of the species into invasive or non-invasive necessary to compute the thresholds by which to distinguish between high-risk and medium-risk non-native species. This analytical approach represents an evidence-based and statistically robust means with which to inform decision-makers and stakeholders about policy and management of potentially invasive species and is expected to serve as a general reference of forthcoming screening applications of Weed Risk Assessment-type toolkits.

Integrated structure-based protein interface prediction

Background

Identifying protein interfaces can inform how proteins interact with their binding partners, uncover the regulatory mechanisms that control biological functions and guide the development of novel therapeutic agents. A variety of computational approaches have been developed for predicting a protein’s interfacial residues from its known sequence and structure. Methods using the known three-dimensional structures of proteins can be template-based or template-free. Template-based methods have limited success in predicting interfaces when homologues with known complex structures are not available to use as templates. The prediction performance of template-free methods that only rely only upon proteins’ intrinsic properties is limited by the amount of biologically relevant features that can be included in an interface prediction model.

Results

We describe the development of an integrated method for protein interface prediction (ISPIP) to explore the hypothesis that the efficacy of a computational prediction method of protein binding sites can be enhanced by using a combination of methods that rely on orthogonal structure-based properties of a query protein, combining and balancing both template-free and template-based features. ISPIP is a method that integrates these approaches through simple linear or logistic regression models and more complex decision tree models. On a diverse test set of 156 query proteins, ISPIP outperforms each of its individual classifiers in identifying protein binding interfaces.

Conclusions

The integrated method captures the best performance of individual classifiers and delivers an improved interface prediction. The method is robust and performs well even when one of the individual classifiers performs poorly on a particular query protein. This work demonstrates that integrating orthogonal methods that depend on different structural properties of proteins performs better at interface prediction than any individual classifier alone.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04852-2.

Accurate and generalizable quantitative scoring of liver steatosis from ultrasound images via scalable deep learning

BACKGROUND

Hepatic steatosis is a major cause of chronic liver disease. Two-dimensional (2D) ultrasound is the most widely used non-invasive tool for screening and monitoring, but associated diagnoses are highly subjective.

AIM

To develop a scalable deep learning (DL) algorithm for quantitative scoring of liver steatosis from 2D ultrasound images.

METHODS

Using multi-view ultrasound data from 3310 patients, 19513 studies, and 228075 images from a retrospective cohort of patients received elastography, we trained a DL algorithm to diagnose steatosis stages (healthy, mild, moderate, or severe) from clinical ultrasound diagnoses. Performance was validated on two multi-scanner unblinded and blinded (initially to DL developer) histology-proven cohorts (147 and 112 patients) with histopathology fatty cell percentage diagnoses and a subset with FibroScan diagnoses. We also quantified reliability across scanners and viewpoints. Results were evaluated using Bland-Altman and receiver operating characteristic (ROC) analysis.

RESULTS

The DL algorithm demonstrated repeatable measurements with a moderate number of images (three for each viewpoint) and high agreement across three premium ultrasound scanners. High diagnostic performance was observed across all viewpoints: Areas under the curve of the ROC to classify mild, moderate, and severe steatosis grades were 0.85, 0.91, and 0.93, respectively. The DL algorithm outperformed or performed at least comparably to FibroScan control attenuation parameter (CAP) with statistically significant improvements for all levels on the unblinded histology-proven cohort and for “= severe” steatosis on the blinded histology-proven cohort.

CONCLUSION

The DL algorithm provides a reliable quantitative steatosis assessment across view and scanners on two multi-scanner cohorts. Diagnostic performance was high with comparable or better performance than the CAP.

An olfactory self-test effectively screens for COVID-19

Background

Key to curtailing the COVID-19 pandemic are wide-scale screening strategies. An ideal screen is one that would not rely on transporting, distributing, and collecting physical specimens. Given the olfactory impairment associated with COVID-19, we developed a perceptual measure of olfaction that relies on smelling household odorants and rating them online.

Methods

Each participant was instructed to select 5 household items, and rate their perceived odor pleasantness and intensity using an online visual analogue scale. We used this data to assign an olfactory perceptual fingerprint, a value that reflects the perceived difference between odorants. We tested the performance of this real-time tool in a total of 13,484 participants (462 COVID-19 positive) from 134 countries who provided 178,820 perceptual ratings of 60 different household odorants.

Results

We observe that olfactory ratings are indicative of COVID-19 status in a country, significantly correlating with national infection rates over time. More importantly, we observe indicative power at the individual level (79% sensitivity and 87% specificity). Critically, this olfactory screen remains effective in participants with COVID-19 but without symptoms, and in participants with symptoms but without COVID-19.

Conclusions

The current odorant-based olfactory screen adds a component to online symptom-checkers, to potentially provide an added first line of defense that can help fight disease progression at the population level. The data derived from this tool may allow better understanding of the link between COVID-19 and olfaction.

A New Family of Similarity Measures for Scoring Confidence of Protein Interactions Using Gene Ontology

The large-scale protein-protein interaction (PPI) data has the potential to play a significant role in the endeavor of understanding cellular processes. However, the presence of a considerable fraction of false positives is a bottleneck in realizing this potential. There have been continuous efforts to utilize complementary resources for scoring confidence of PPIs in a manner that false positive interactions get a low confidence score. Gene Ontology (GO), a taxonomy of biological terms to represent the properties of gene products and their relations, has been widely used for this purpose. We utilize GO to introduce a new set of specificity measures: Relative Depth Specificity (RDS), Relative Node-based Specificity (RNS), and Relative Edge-based Specificity (RES), leading to a new family of similarity measures. We use these similarity measures to obtain a confidence score for each PPI. We evaluate the new measures using four different benchmarks. We show that all the three measures are quite effective. Notably, RNS and RES more effectively distinguish true PPIs from false positives than the existing alternatives. RES also shows a robust set-discriminating power and can be useful for protein functional clustering as well.

Combining information on nocturnal rapid eye movement sleep latency and atonia to facilitate diagnosis of pediatric narcolepsy type 1

STUDY OBJECTIVES

The diagnosis of narcolepsy type 1 (NT1) at its onset in children and adolescents is often difficult, with substantial diagnostic delay. We aimed to test and validate the effectiveness of rapid eye movement (REM) sleep latency (REML), the REM sleep atonia index (RAI), and their combination for the automatic identification of pediatric patients with NT1 based on the standard scoring of nocturnal polysomnograms.

METHODS

A retrospective cohort of 71 pediatric patients with NT1 and 42 controls was subdivided in test and validation cohorts. A novel index (COM) was developed as a nonlinear function of REML and RAI. The effectiveness of REML, RAI, and COM in identifying patients with NT1 was assessed with receiver operating characteristic (ROC) curves.

RESULTS

REML, RAI, and COM significantly identified patients with NT1 both in the test and validation cohorts. Optimal thresholds that maximized identification accuracy were estimated in the test cohort (REML, 49.5 min; RAI, 0.91; COM, 4.57 AU) and validated in the other cohort. COM performed significantly better in identifying patients with NT1 than either REML or RAI, with ROC area under the curve of 94%-100%, sensitivity 85%-96%, and specificity 92%-100%, and with good night-to-night agreement (Cohen's k = 0.69).

CONCLUSIONS

The analysis of REML, RAI, and particularly their combination in the COM index may help shorten diagnostic delay of NT1 in children and adolescents based on the standard scoring of nocturnal polysomnography.

Which is the optimally defined vestibular cross-section to diagnose unilateral Meniere's disease with delayed post-gadolinium 3D fluid-attenuated inversion recovery MRI?

OBJECTIVES

Delayed post-gadolinium 3D fluid-attenuated inversion recovery (FLAIR) MRI is used to support a diagnosis of Ménière's disease (MD) with the ratio of the endolymphatic space (ES) to the sum of the endolymphatic and perilymphatic spaces (SEPS) on a cross-section through the vestibule being a key diagnostic criterion. It was hypothesised that the exact definition of the vestibular cross-section would influence the ES: SEPS ratio, its ability to diagnose MD, and its reproducibility.

METHODS

Following institutional approval, 22 patients (five male, 17 female; mean age 52.1) with unilateral MD and delayed post-gadolinium 3D FLAIR MRI were retrospectively analysed. Two observers measured the ES and SEPS on predefined axial (superior and inferior) and sagittal vestibular cross-sections. Receiver operating characteristic (ROC) curves, Bland-Altman plots and intraclass correlation (ICC) were analysed for the ES:SEPS ratios.

RESULTS

The area under the curve (AUC) was decreased for the ES:SEPS ratios on the superior axial section through the vestibule (AUC 0.737) compared to the inferior axial (AUC 0.874) and sagittal sections (AUC 0.878). The resulting optimal thresholds (sensitivities/specificities) were 0.21 (0.66/0.75), 0.16 (0.77/0.9) and 0.285 (0.75/0.96). The reproducibility was excellent for all measures with ICCs of 0.97, 0.98 and 0.99.

CONCLUSION

Inferior axial or sagittal vestibular cross-sections are more accurate for the diagnosis of MD ears and have excellent reproducibility.

ADVANCES IN KNOWLEDGE

The choice of vestibular cross-section influences both the ability to distinguish MD from asymptomatic contralateral ears, and the optimum threshold ES:SEPS value.

MHCII3D-Robust Structure Based Prediction of MHC II Binding Peptides

Knowledge of MHC II binding peptides is highly desired in immunological research, particularly in the context of cancer, autoimmune diseases, or allergies. The most successful prediction methods are based on machine learning methods trained on sequences of experimentally characterized binding peptides. Here, we describe a complementary approach called MHCII3D, which is based on structural scaffolds of MHC II-peptide complexes and statistical scoring functions (SSFs). The MHC II alleles reported in the Immuno Polymorphism Database are processed in a dedicated 3D-modeling pipeline providing a set of scaffold complexes for each distinct allotype sequence. Antigen protein sequences are threaded through the scaffolds and evaluated by optimized SSFs. We compared the predictive power of MHCII3D with different sequence-based machine learning methods. The Pearson correlation to experimentally determine IC₅₀ values for MHC II Automated Server Benchmarks data sets from IEDB (Immune Epitope Database) is 0.42, which is in the competitor methods range. We show that MHCII3D is quite robust in leaving one molecule out tests and is therefore not prone to overfitting. Finally, we provide evidence that MHCII3D can complement the current sequence-based methods and help to identify problematic entries in IEDB. Scaffolds and MHCII3D executables can be freely downloaded from our web pages.

Machine learning for filtering out false positive grey matter atrophies in single subject voxel based morphometry: A simulation based study

Single subject VBM (SS-VBM), has been used as an alternative tool to standard VBM for single case studies. However, it has the disadvantage of producing an excessively large number of false positive detections. In this study we propose a machine learning technique widely used for automated data classification, namely Support Vector Machine (SVM), to refine the findings produced by SS-VBM. A controlled set of experiments was conducted to evaluate the proposed approach using three-dimensional T1 MRI scans from control subjects collected from the publicly available IXI dataset. The scans were artificially atrophied at different locations and with different sizes to mimic the behavior of neurological disorders. Results empirically demonstrated that the proposed method is able to significantly reduce the amount of false positive clusters (p < 0.05), with no statistical differences in the true positive findings (p > 0.05). This evidence was observed to be consistent for different atrophied areas and sizes of atrophies. This approach could be potentially be applied to alleviate the intensive manual analysis that radiologists and clinicians have to perform to filter out miss-detections of SS-VBM, increasing its usability for image reading.

Deep hiearchical multi-label classification applied to chest X-ray abnormality taxonomies

Chest X-rays (CXRs) are a crucial and extraordinarily common diagnostic tool, leading to heavy research for computer-aided diagnosis (CAD) solutions. However, both high classification accuracy and meaningful model predictions that respect and incorporate clinical taxonomies are crucial for CAD usability. To this end, we present a deep hierarchical multi-label classification (HMLC) approach for CXR CAD. Different than other hierarchical systems, we show that first training the network to model conditional probability directly and then refining it with unconditional probabilities is key in boosting performance. In addition, we also formulate a numerically stable cross-entropy loss function for unconditional probabilities that provides concrete performance improvements. Finally, we demonstrate that HMLC can be an effective means to manage missing or incomplete labels. To the best of our knowledge, we are the first to apply HMLC to medical imaging CAD. We extensively evaluate our approach on detecting abnormality labels from the CXR arm of the Prostate, Lung, Colorectal and Ovarian (PLCO) dataset, which comprises over 198,000 manually annotated CXRs. When using complete labels, we report a mean area under the curve (AUC) of 0.887, the highest yet reported for this dataset. These results are supported by ancillary experiments on the PadChest dataset, where we also report significant improvements, 1.2% and 4.1% in AUC and average precision, respectively over strong "flat" classifiers. Finally, we demonstrate that our HMLC approach can much better handle incompletely labelled data. These performance improvements, combined with the inherent usefulness of taxonomic predictions, indicate that our approach represents a useful step forward for CXR CAD.

Past and present of computer-assisted dermoscopic diagnosis: performance of a conventional image analyser versus a convolutional neural network in a prospective data set of 1,981 skin lesions

BACKGROUND

Convolutional neural networks (CNNs) have shown a dermatologist-level performance in the classification of skin lesions. We aimed to deliver a head-to-head comparison of a conventional image analyser (CIA), which depends on segmentation and weighting of handcrafted features, to a CNN trained by deep learning.

METHODS

Cross-sectional study using a real-world, prospectively acquired, dermoscopic dataset of 1981 skin lesions to compare the diagnostic performance of a market-approved CNN (Moleanalyzer-Pro™, developed in 2018) to a CIA (Moleanalyzer-3™/Dynamole™; developed in 2004, all FotoFinder Systems Inc, Germany). As a reference standard, we used histopathological diagnoses (n = 785) or, in non-excised benign lesions (n = 1196), expert consensus plus an uneventful follow-up by sequential digital dermoscopy for at least 2 years.

RESULTS

A total of 281 malignant lesions and 1700 benign lesions from 435 patients (62.2% male, mean age: 52 years) were prospectively imaged. The CNN showed a sensitivity of 77.6% (95% confidence interval [CI]: [72.4%-82.1%]), specificity of 95.3% (95% CI: [94.2%-96.2%]), and receiver operating characteristic (ROC)-area under the curve (AUC) of 0.945 (95% CI: [0.930-0.961]). In contrast, the CIA achieved a sensitivity of 53.4% (95% CI: [47.5%-59.1%]), specificity of 86.6% (95% CI: [84.9%-88.1%]) and ROC-AUC of 0.738 (95% CI: [0.701-0.774]). The data set included melanomas originally diagnosed by dynamic changes during sequential digital dermoscopy (52 of 201, 20.6%), which reduced the sensitivities of both classifiers. Pairwise comparisons of sensitivities, specificities, and ROC-AUCs indicated a clear outperformance by the CNN (all p < 0.001).

CONCLUSIONS

The superior diagnostic performance of the CNN argues against a continued application of former CIAs as an aide to physicians' clinical management decisions.

Burden of cerebral hypoperfusion in patients with delayed cerebral ischemia after subarachnoid hemorrhage

OBJECTIVE

Delayed cerebral ischemia (DCI) after subarachnoid hemorrhage (SAH) may result in focal neurological deficits and cerebral infarction, believed to result from critical regional rather than global impairments in cerebral blood flow (CBF). However, the burden of such regional hypoperfusion has not been evaluated by gold-standard voxel-by-voxel CBF measurements. Specifically, the authors sought to determine whether the proportion of brain affected by hypoperfusion was greater in patients with DCI than in SAH controls without DCI and whether the symptomatic hemisphere (in those with lateralizing deficits) exhibited a greater cerebral hypoperfusion burden.

METHODS

Sixty-one patients with aneurysmal SAH underwent 15O PET to measure regional CBF during the period of risk for DCI (median 8 days after SAH, IQR 7-10 days). Regions of visibly abnormal brain on head CT studies, including areas of hemorrhage and infarction, were excluded. Burden of hypoperfusion was defined as the proportion of PET voxels in normal-appearing brain with CBF < 25 ml/100 g/min. Global CBF and hypoperfusion burden were compared between patients with and those without DCI at the time of PET. For patients with focal impairments from DCI, the authors also compared average CBF and hypoperfusion burden in symptomatic versus asymptomatic hemispheres.

RESULTS

Twenty-three patients (38%) had clinical DCI at the time of PET. Those with DCI had higher mean arterial pressure (MAP; 126 ± 14 vs 106 ± 12 mm Hg, p < 0.001) and 18 (78%) were on vasopressor therapy at the time of PET study. While global CBF was not significantly lower in patients with DCI (mean 39.4 ± 11.2 vs 43.0 ± 8.3 ml/100 g/min, p = 0.16), the burden of hypoperfusion was greater (20%, IQR 12%-23%, vs 12%, 9%-16%, p = 0.006). Burden of hypoperfusion performed better than global CBF as a predictor of DCI (area under the curve 0.71 vs 0.65, p = 0.044). Neither global CBF nor hypoperfusion burden differed in patients who responded to therapy compared to those who had not improved by the time of PET. Although hemispheric CBF was not lower in the symptomatic versus contralateral hemisphere in the 13 patients with focal deficits, there was a trend toward greater burden of hypoperfusion in the symptomatic hemisphere (21% vs 18%, p = 0.049).

CONCLUSIONS

The burden of hypoperfusion was greater in patients with DCI, despite hemodynamic therapies, higher MAP, and equivalent global CBF. Similarly, hypoperfusion burden was greater in the symptomatic hemisphere of DCI patients with focal deficits even though the average CBF was similar to that in the contralateral hemisphere. Evaluating the proportion of the brain with critical hypoperfusion after SAH may better capture the extent of DCI than averaging CBF across heterogenous brain regions.

Diagnostic utility of the Addenbrooke's Cognitive Examination - III (ACE-III), Mini-ACE, Mini-Mental State Examination, Montreal Cognitive Assessment, and Hasegawa Dementia Scale-Revised for detecting mild cognitive impairment and dementia

BACKGROUND

Early detection of mild cognitive impairment (MCI) and dementia is important to promptly start appropriate intervention. However, it is difficult to examine a patient using long and thorough cognitive tests in a general clinical setting. In this study, we aimed to investigate the diagnostic validity of the Addenbrooke's Cognitive Examination - III (ACE-III), Mini-ACE (M-ACE), Montreal Cognitive Assessment (MoCA), Hasegawa Dementia Scale-Revised (HDS-R), and Mini-Mental State Examination (MMSE) to identify MCI and dementia.

METHODS

A total of 249 subjects (controls = 50, MCI = 94, dementia = 105) at a memory clinic participated in this study, and took the ACE-III, M-ACE, MoCA, HDS-R, and MMSE. After all examinations had been carried out, a conference was held, and the clinical diagnoses were established.

RESULTS

The areas under the curve (AUC) of the ACE-III, M-ACE, MoCA, HDS-R, and MMSE for diagnosing MCI were 0.891, 0.856, 0.831, 0.808, and 0.782. The AUC of the ACE-III was significantly larger than those of the MoCA, HDS-R, and MMSE. The AUCs of the ACE-III, M-ACE, MoCA, HDS-R, and MMSE for diagnosing dementia were 0.930, 0.917, 0.854, 0.871, and 0.856. Thus, the AUCs of the ACE-III and M-ACE were significantly larger than those of the MoCA, HDS-R, and MMSE.

CONCLUSION

The ACE-III is a useful cognitive instrument to detect MCI. For distinguishing dementia patients from non-dementia patients, the ACE-III and M-ACE are superior to the MoCA, HDS-R, and MMSE.

A novel variant of the 13C-methacetin liver function breath test that eliminates the confounding effect of individual differences in systemic CO2 kinetics

The principle of dynamic liver function breath tests is founded on the administration of a ¹³C-labeled drug and subsequent monitoring of ¹³CO₂ in the breath, quantified as time series delta over natural baseline ¹³CO₂ (DOB) liberated from the drug during hepatic CYP-dependent detoxification. One confounding factor limiting the diagnostic value of such tests is that only a fraction of the liberated ¹³CO₂ is immediately exhaled, while another fraction is taken up by body compartments from which it returns with delay to the plasma. The aims of this study were to establish a novel variant of the methacetin-based breath test LiMAx that allows to estimate and to eliminate the confounding effect of systemic ¹³CO₂ distribution on the DOB curve and thus enables a more reliable assessment of the hepatic detoxification capacity compared with the conventional LiMAx test. We designed a new test variant (named "2DOB") consisting of two consecutive phases. Phase 1 is initiated by the intravenous administration of ¹³C-bicarbonate. Phase 2 starts about 30 min later with the intravenous administration of the ¹³C-labelled test drug. Using compartment modelling, the resulting 2-phasic DOB curve yields the rate constants for the irreversible elimination and the reversible exchange of plasma ¹³CO₂ with body compartments (phase 1) and for the detoxification and exchange of the drug with body compartments (phase 2). We carried out the 2DOB test with the test drug ¹³C-methacetin in 16 subjects with chronic liver pathologies and 22 normal subjects, who also underwent the conventional LiMAx test. Individual differences in the systemic CO₂ kinetics can lead to deviations up to a factor of 2 in the maximum of DOB curves (coefficient of variation CV ≈ 0.2) which, in particular, may hamper the discrimination between subjects with normal or mildly impaired detoxification capacities. The novel test revealed that a significant portion of the drug is not immediately metabolized, but transiently taken up into a storage compartment. Intriguingly, not only the hepatic detoxification rate but also the storage capacity of the drug, turned out to be indicative for a normal liver function. We thus used both parameters to define a scoring function which yielded an excellent disease classification (AUC = 0.95) and a high correlation with the MELD score (R_Spearman = 0.92). The novel test variant 2DOB promises a significant improvement in the assessment of impaired hepatic detoxification capacity. The suitability of the test for the reliable characterization of the natural history of chronic liver diseases (fatty liver-fibrosis-cirrhosis) has to be assessed in further studies.

Use of the Bispectral Index to Predict Recovery of Consciousness in Patients with Spontaneous Intracerebral Hemorrhage After Surgical Hematoma Evacuation: A Prospective Cohort Study

Background

Our study aimed to test the predictive value of the bispectral index (BIS) for the post-operational consciousness recovery in patients undergone hematoma evacuation due to spontaneous intracerebral hemorrhage (ICH).

Material/Methods

In this prospective cohort study, we enrolled adult spontaneous ICH patients after surgical hematoma evacuation who did not recover consciousness on the first postoperative day. After patient enrollment, the BIS was continuously monitored for 12 hours, and the motor response on the Glasgow Coma Scale (GCS-M) was evaluated. The patients were followed up for 30 days and divided into a consciousness recovery group and a nonrecovery group. Receiver operating characteristic curve analysis was performed to investigate the predictive values of the BIS, GCS-M and ICH score on the consciousness recovery. The area under the curve (AUC) and 95% confidence interval (95%CI) were calculated. During the 12-hour monitoring period, the peak BIS value after GCS-M stimulation was used for ROC analysis.

Results

Of the 55 enrolled patients, 19 patients recovered consciousness, and 36 patients did not. The BIS value of the consciousness recovery group was significantly higher than that of the nonrecovery group (P<0.001). For consciousness recovery prediction, the AUC (95%CI) of the BIS values after external stimulation was 0.97 (0.91–1.00), which was superior to the GCS-M (0.75 [0.59–0.91]) and ICH score (0.57 [0.41–0.73]).

Conclusions

Our study demonstrates that BIS might be a potential tool for predicting the consciousness recovery in ICH patients undergone surgical hematoma evacuation.

Validation of Addenbrooke's cognitive examination III for detecting mild cognitive impairment and dementia in Japan

Background

Early detection of mild cognitive impairment (MCI) and dementia is very important to begin appropriate treatment promptly and to prevent disease exacerbation. We investigated the screening accuracy of the Japanese version of Addenbrooke’s Cognitive Examination III (ACE-III) to diagnose MCI and dementia.

Methods

The original ACE-III was translated and adapted to Japanese. It was then administered to a Japanese population. The Hasegawa Dementia Scale-revised (HDS-R) and Mini-mental State Examination (MMSE) were also applied to evaluate cognitive dysfunction. In total, 389 subjects (dementia = 178, MCI = 137, controls = 73) took part in our study.

Results

The optimal ACE-III cut-off scores to detect MCI and dementia were 88/89 (sensitivity 0.77, specificity 0.92) and 75/76 (sensitivity 0.82, specificity 0.90), respectively. ACE-III was superior to HDS-R and MMSE in the detection of MCI or dementia. The internal consistency, test-retest reliability, and inter-rater reliability of ACE-III were excellent.

Conclusions

ACE-III is a useful cognitive test to detect MCI and dementia. ACE-III may be widely useful in clinical practice.

Interoperator Reproducibility of Carotid Elastography for Identification of Vulnerable Atherosclerotic Plaques

Ultrasound-based carotid elastography has been developed to evaluate the vulnerability of carotid atherosclerotic plaques. The aim of this study was to investigate the in vivo interoperator reproducibility of carotid elastography for the identification of vulnerable plaques, with high-resolution magnetic resonance imaging (MRI) as reference. Ultrasound radio-frequency data of 45 carotid arteries (including 53 plaques) from 32 volunteers were acquired separately by two experienced operators in the longitudinal view and then were used to estimate the interframe axial strain rate (ASR) with a two-step optical flow method. The maximum 99th percentile of absolute ASR of all plaques in a carotid artery was used as the elastographic index. MRI scanning was also performed on each volunteer to identify the vulnerable plaque. The results showed no systematic bias in the Bland-Altman plot and an intraclass correlation coefficient of 0.66 between the two operators. In addition, no statistical significance was found between the receiver operating characteristic (ROC) curves from the two operators ( ), and their areas under the ROC curves were 0.83 and 0.77, respectively. Using the mean measurements of the two operators as the classification criterion, a sensitivity of 71.4%, a specificity of 87.1%, and an accuracy of 82.2% were obtained with a cutoff value of 1.37 [Formula: see text]. This study validates the interoperator reproducibility of ultrasound-based carotid elastography for identifying vulnerable carotid plaques.

A baseline study for detection of Parkinson's disease with 3D-transcranial sonography and uni-lateral reconstruction

INTRODUCTION

TCS is a well-established technique for diagnosis of Parkinson's disease (PD). Volumetric 3D-TCS is a promising complementary approach for objective acquisition and analysis, in particular for less experienced sonographers. This study provides baselines for Parkinson detection (sensitivity and specificity), cutoff values and inter-rater agreement in 3D-TCS.

METHODS

We performed 3D-TCS in 52 subjects (healthy controls and PD) bilaterally, and reconstructed in 3D space uni-laterally. Ipsi-lateral hyperechogenicities in the substantia nigra are manually segmented slice-by-slice in the 3D volume by two raters at different experience levels. ROC threshold analysis is performed and compared on features representing 3D volume and axial cross-sections (2.5D) of hyperechogenicities. Pearson correlation and intra-class correlation coefficients were evaluated for assessment of inter-rater agreement.

RESULTS

50 subjects were included. Both raters achieved high classification accuracy with 2.5D/3D features extracted from 3D-TCS volumes (best results sensitivity/specificity/cut-off per rater: 84.6%/88.9%/25.0mm²; 77.8%/88.9%/95.9mm³). The inter-rater agreement in 3D was high (ICC(A,1) = 0.777, p < 10^-3), the classification performance of both sonographers was statistically not significantly different.

CONCLUSION

The study presents first baseline values for uni-lateral 3D-TCS examination, and finds no disadvantage of uni-lateral reconstructions compared to previous bi-lateral fusion. Volumetric 3D-TCS has potential for a high inter-rater agreement and accuracy in detection of PD, in particular for sonographers with less experience.

Comparative Analysis of Markers of Mass Effect after Ischemic Stroke

BACKGROUND AND PURPOSE

Midline shift determined on magnetic resonance imaging (MRI) or computed tomography (CT) images is a well-validated marker of mass effect after large hemispheric infarction and associated with mortality. In this study, we targeted a population with moderately sized strokes. We compared midline shift to other imaging markers and determined their ability to predict long-term outcome.

METHODS

MRI scans were studied from the Echoplanar Imaging Thrombolysis Evaluation Trial (EPITHET) cohort. Midline shift, acute stroke lesion volume, lesional swelling volume, change in ipsilateral hemisphere volume, the ratio of ipsilateral to contralateral hemisphere volume, and the reduction in lateral ventricle volume were measured. The relationships of these markers with poor outcome (modified Rankin scale score 3-6 at day 90) were assessed. Receiver-operating characteristic (ROC) curves were generated to compare the performance of each metric.

RESULTS

Of the 71 included patients, 59.2% had a poor outcome that was associated with significantly larger values for midline shift, lesional swelling volume, and ratio of hemisphere volumes. Lesional swelling volume, change in hemisphere volume, ratio of hemisphere volumes, and lateral ventricle displacement were each correlated with midline shift (Spearman r = .60, .49, .61, and -.56, respectively; all P < .0001). ROC curve analysis showed that lesional swelling volume (area under the curve [AUC] = .791) predicted poor outcome better than midline shift (AUC = .682). For predicting mortality, ROC curve analysis showed that these three markers were equivalent.

CONCLUSION

The ratio of ipsilateral to contralateral hemisphere volume, baseline lesion volume and lesional swelling volume best predicted poor outcome across a spectrum of stroke sizes.

A Potential VEP Biomarker for Mild Cognitive Impairment: Evidence from Selective Visual Deficit of Higher-Level Dorsal Pathway

Visual dysfunctions are common in Alzheimer's disease (AD). Our aim was to establish a neurophysiological biomarker for amnestic mild cognitive impairment (aMCI). Visual evoked potentials (VEPs) were recorded in aMCI patients who later developed AD (n = 15) and in healthy older (n = 15) and younger controls (n = 15). Visual stimuli were optimized to separately activate lower and higher levels of the ventral and dorsal streams. We compared VEP parameters across the three groups of participants and conducted a linear correlation analysis between VEPs and data from neuropsychological tests. We then used a receiver operating characteristic (ROC) analysis to discriminate those with aMCI from those who were healthy older adults. The latency and phase of VEPs to lower-level stimuli (chromatic and achromatic gratings) were significantly affected by age but not by cognitive decline. Conversely, VEP latencies for higher-ventral (faces and kanji-words) and dorsal (kana-words and optic flow motion) stimuli were not affected by age, but they were significantly prolonged in aMCI patients. Interestingly, VEPs for higher-dorsal stimuli were related to outcomes of neuropsychological tests. Furthermore, the ROC analysis showed that the highest areas under the curve were obtained for VEP latencies in response to higher-dorsal stimuli. These results suggest aMCI-related functional impairment specific to higher-level visual processing. Further, dysfunction in the higher-level of the dorsal stream could be an early indicator of cognitive decline. Therefore, we conclude that VEPs associated with higher-level dorsal stream activity can be a sensitive biomarker for early detection of aMCI.

Machine learning algorithms based on signals from a single wearable inertial sensor can detect surface- and age-related differences in walking

The aim of this study was to investigate if a machine learning algorithm utilizing triaxial accelerometer, gyroscope, and magnetometer data from an inertial motion unit (IMU) could detect surface- and age-related differences in walking. Seventeen older (71.5 ± 4.2 years) and eighteen young (27.0 ± 4.7 years) healthy adults walked over flat and uneven brick surfaces wearing an inertial measurement unit (IMU) over the L5 vertebra. IMU data were binned into smaller data segments using 4-s sliding windows with 1-s step lengths. Ninety percent of the data were used as training inputs and the remaining ten percent were saved for testing. A deep learning network with long short-term memory units was used for training (fully supervised), prediction, and implementation. Four models were trained using the following inputs: all nine channels from every sensor in the IMU (fully trained model), accelerometer signals alone, gyroscope signals alone, and magnetometer signals alone. The fully trained models for surface and age outperformed all other models (area under the receiver operator curve, AUC = 0.97 and 0.96, respectively; p ≤ .045). The fully trained models for surface and age had high accuracy (96.3, 94.7%), precision (96.4, 95.2%), recall (96.3, 94.7%), and f1-score (96.3, 94.6%). These results demonstrate that processing the signals of a single IMU device with machine-learning algorithms enables the detection of surface conditions and age-group status from an individual's walking behavior which, with further learning, may be utilized to facilitate identifying and intervening on fall risk.

Characterisation of preproendothelin-1 derived peptides identifies Endothelin-Like Domain Peptide as a modulator of Endothelin-1

Endothelin-1 (ET-1) is involved in the pathogenesis of cardiac and renal diseases, and in the progression of tumour growth in cancer, but current diagnosis and treatment remain inadequate. Peptides derived from the 212 amino acid precursor preproendothelin-1 (ppET-1) may have utility as biomarkers, or cause biological effects that are unaffected by endothelin receptor antagonists. Here, we used specific immunoassays and LC-MS/MS to identify NT-proET-1 (ppET-1_[18–50]), Endothelin-Like Domain Peptide (ELDP, ppET-1_[93–166]) and CT-proET-1 (ppET-1_[169–212]) in conditioned media from cultured endothelial cells. Synthesis of these peptides correlated with ET-1, and plasma ELDP and CT-proET-1 were elevated in patients with chronic heart failure. Clearance rates of NT-proET-1, ELDP and CT-proET-1 were determined after i.v. injection in anaesthetised rats. CT-proET-1 had the slowest systemic clearance, hence providing a biological basis for it being a better biomarker of ET-1 synthesis. ELDP contains the evolutionary conserved endothelin-like domain sequence, which potentially confers biological activity. On isolated arteries ELDP lacked direct vasoconstrictor effects. However, it enhanced ET-1 vasoconstriction and prolonged the increase in blood pressure in anaesthetised rats. ELDP may therefore contribute to disease pathogenesis by augmenting ET-1 responses.

Predicting primary progressive aphasias with support vector machine approaches in structural MRI data

Primary progressive aphasia (PPA) encompasses the three subtypes nonfluent/agrammatic variant PPA, semantic variant PPA, and the logopenic variant PPA, which are characterized by distinct patterns of language difficulties and regional brain atrophy. To validate the potential of structural magnetic resonance imaging data for early individual diagnosis, we used support vector machine classification on grey matter density maps obtained by voxel-based morphometry analysis to discriminate PPA subtypes (44 patients: 16 nonfluent/agrammatic variant PPA, 17 semantic variant PPA, 11 logopenic variant PPA) from 20 healthy controls (matched for sample size, age, and gender) in the cohort of the multi-center study of the German consortium for frontotemporal lobar degeneration. Here, we compared a whole-brain with a meta-analysis-based disease-specific regions-of-interest approach for support vector machine classification. We also used support vector machine classification to discriminate the three PPA subtypes from each other. Whole brain support vector machine classification enabled a very high accuracy between 91 and 97% for identifying specific PPA subtypes vs. healthy controls, and 78/95% for the discrimination between semantic variant vs. nonfluent/agrammatic or logopenic PPA variants. Only for the discrimination between nonfluent/agrammatic and logopenic PPA variants accuracy was low with 55%. Interestingly, the regions that contributed the most to the support vector machine classification of patients corresponded largely to the regions that were atrophic in these patients as revealed by group comparisons. Although the whole brain approach took also into account regions that were not covered in the regions-of-interest approach, both approaches showed similar accuracies due to the disease-specificity of the selected networks. Conclusion, support vector machine classification of multi-center structural magnetic resonance imaging data enables prediction of PPA subtypes with a very high accuracy paving the road for its application in clinical settings.

•

Aim was to evaluate the potential of multi-center MRI data for individual PPA diagnosis.

•

We used support vector machine classification in PPA variants and healthy controls.

•

We compared a whole brain approach with a ROI (taken from meta-analyses) approach.

•

Accuracies were overall quite high, for both, the whole brain and the ROI approach.

Rey's Auditory Verbal Learning Test scores can be predicted from whole brain MRI in Alzheimer's disease

Rey's Auditory Verbal Learning Test (RAVLT) is a powerful neuropsychological tool for testing episodic memory, which is widely used for the cognitive assessment in dementia and pre-dementia conditions. Several studies have shown that an impairment in RAVLT scores reflect well the underlying pathology caused by Alzheimer's disease (AD), thus making RAVLT an effective early marker to detect AD in persons with memory complaints. We investigated the association between RAVLT scores (RAVLT Immediate and RAVLT Percent Forgetting) and the structural brain atrophy caused by AD. The aim was to comprehensively study to what extent the RAVLT scores are predictable based on structural magnetic resonance imaging (MRI) data using machine learning approaches as well as to find the most important brain regions for the estimation of RAVLT scores. For this, we built a predictive model to estimate RAVLT scores from gray matter density via elastic net penalized linear regression model. The proposed approach provided highly significant cross-validated correlation between the estimated and observed RAVLT Immediate (R = 0.50) and RAVLT Percent Forgetting (R = 0.43) in a dataset consisting of 806 AD, mild cognitive impairment (MCI) or healthy subjects. In addition, the selected machine learning method provided more accurate estimates of RAVLT scores than the relevance vector regression used earlier for the estimation of RAVLT based on MRI data. The top predictors were medial temporal lobe structures and amygdala for the estimation of RAVLT Immediate and angular gyrus, hippocampus and amygdala for the estimation of RAVLT Percent Forgetting. Further, the conversion of MCI subjects to AD in 3-years could be predicted based on either observed or estimated RAVLT scores with an accuracy comparable to MRI-based biomarkers.

Does Epoetin Beta Still Have a Place in Peginterferon Alpha-2a Plus Ribavirin Treatment Strategies for Chronic Hepatitis C?

To investigate the impact of epoetin beta (EPO) on sustained virological response (SVR) in hepatitis C virus (HCV)-infected patients treated with peginterferon-ribavirin (RBV). Controlled, randomized, pragmatic multicenter study to assess 2 strategies, ie, the use (EPO group) or nonuse (control group) of EPO in terms of achieving SVR in treatment-naive, genotype non-2/non-3 HCV-infected patients receiving a 48-week treatment regimen of pegylated interferon α-2a (peg-IFN) plus RBV (randomization 2:1). The single-nucleotide polymorphisms of interferon lambda 3 (IFNL3) (rs12979860 and rs8099917), interferon lambda 4 (IFNL4) (ss469415590), and inosine triphosphatase (ITPA) (rs1127354 and rs7270101) were determined retrospectively. Two hundred twenty-seven patients were included in the study. In the global population (n = 227), the overall SVR rate was 52% (118/227). Nonresponse and relapse occurred in respectively 46/227 (20.3%) and 42/227 (18.5%) patients. In the intention-to-treat analysis, 55.5% of patients with anemia (n = 164) had a SVR, specifically 57.4% in the EPO group versus 52.4% in the control group, but the difference was not statistically significant. In the anemic population, independent factors associated with SVR were IFNL3 and IFNL4 polymorphisms, pretreatment HCV RNA level, iron level, and aspartate aminotransferase/alanine aminotransferase (AST/ALT) ratio. EPO has little impact on SVR in patients treated with peg-IFN+RBV and should be recommended only for patients with severe anemia.