Using Support Vector Machine to identify imaging biomarkers of neurological and psychiatric disease: a critical review

Identification of focal epilepsy by diffusion tensor imaging using machine learning

OBJECTIVE

The aim of this study was to evaluate the feasibility of machine learning based on diffusion tensor imaging (DTI) measures to distinguish patients with focal epilepsy versus healthy controls and antiseizure medication (ASM) responsiveness.

METHODS

This was a retrospective study performed at a tertiary hospital. We enrolled 456 patients with focal epilepsy, who underwent DTI and were taking ASMs. We enrolled 100 healthy subjects as a control. We obtained the conventional DTI measures and structural connectomic profiles from the DTI.

RESULTS

The support vector machine (SVM) classifier based on the conventional DTI measures revealed an accuracy of 76.5% and an area under curve (AUC) of 0.604 (95% Confidence interval (CI), 0.506-0.695). Another SVM classifier combined with structural connectomic profiles demonstrated an accuracy of 82.8% and an AUC of 0.701 (95% CI, 0.606-0.784). Of the 456 patients with epilepsy, 242 patients were ASM good responders, whereas 214 patients were ASM poor responders. In the classification of the ASM responders, an SVM classifier based on the conventional DTI measures revealed an accuracy of 54.9% and an AUC of 0.551 (95% CI, 0.443-0.655). Another SVM classifier combined with structural connectomic profiles demonstrated an accuracy of 59.3% and an AUC of 0.594 (95% CI, 0.485-0.695).

CONCLUSION

DTI using a machine learning is useful for differentiating patients with focal epilepsy from healthy controls, but it cannot classify ASM responsiveness. Combining structural connectomic profiles results in a better classification performance than the use of conventional DTI measures alone for identifying focal epilepsy and ASM responsiveness.

Boost in Test-Retest Reliability in Resting State fMRI with Predictive Modeling

Recent studies found low test-retest reliability in functional magnetic resonance imaging (fMRI), raising serious concerns among researchers, but these studies mostly focused on the reliability of individual fMRI features (e.g., individual connections in resting state connectivity maps). Meanwhile, neuroimaging researchers increasingly employ multivariate predictive models that aggregate information across a large number of features to predict outcomes of interest, but the test-retest reliability of predicted outcomes of these models has not previously been systematically studied. Here we apply 10 predictive modeling methods to resting state connectivity maps from the Human Connectome Project dataset to predict 61 outcome variables. Compared with mean reliability of individual resting state connections, we find mean reliability of the predicted outcomes of predictive models is substantially higher for all 10 modeling methods assessed. Moreover, improvement was consistently observed across all scanning and processing choices (i.e., scan lengths, censoring thresholds, volume- vs. surface-based processing). For the most reliable methods, the reliability of predicted outcomes was mostly, though not exclusively, in the "good" range (above 0.60). Finally, we identified three mechanisms that help to explain why predicted outcomes of predictive models have higher reliability than individual imaging features. We conclude that researchers can potentially achieve higher test-retest reliability by making greater use of predictive models.

Sensitivity of unilateral- versus bilateral-onset spike-wave discharges to ethosuximide and carbamazepine in the fluid percussion injury rat model of traumatic brain injury

Unilateral-onset spike-wave discharges (SWDs) following fluid percussion injury (FPI) in rats have been used for nearly two decades as a model for complex partial seizures in human posttraumatic epilepsy (PTE). This study determined if SWDs with a unilateral versus bilateral cortical onset differed. In this experiment, 2-mo-old rats received severe FPI (3 atm) or sham surgery and were instrumented for chronic video-electrocorticography (ECoG) recording (up to 9 mo). The antiseizure drug, carbamazepine (CBZ), and the antiabsence drug, ethosuximide (ETX), were administered separately to determine if they selectively suppressed unilateral- versus bilateral-onset SWDs, respectively. SWDs did not significantly differ between FPI and sham rats on any measured parameter (wave-shape, frequency spectrum, duration, or age-related progression), including unilateral (∼17%) versus bilateral (∼83%) onsets. SWDs with a unilateral onset preferentially originated ipsilateral to the craniotomy in both FPI and sham rats, suggesting that the unilateral-onset SWDs were related to surgical injury and not specifically to FPI. ETX profoundly suppressed SWDs with either unilateral or bilateral onsets, and CBZ had no effect on either type of SWD. These results suggest that SWDs with either a unilateral or bilateral onset have a pharmacosensitivity similar to absence seizures and are very different from the complex partial seizures of PTE. Therefore, SWDs with a unilateral onset after FPI are not a model of the complex partial seizures that occur in PTE, and their use for finding new treatments for PTE could be counterproductive, particularly if their close similarity to normal brain oscillations is not acknowledged.NEW & NOTEWORTHY Unilateral-onset spike-wave discharges (SWDs) in rats have been used to model complex partial seizures in human posttraumatic epilepsy (PTE), compared to bilateral-onset SWDs thought to reflect human absence seizures. Here, we show that both unilateral- and bilateral-onset SWDs following traumatic brain injury are suppressed by the antiabsence drug ethosuximide and are unaffected by the antiseizure drug carbamazepine. We propose that unilateral-onset SWDs are not useful for studying mechanisms of, or treatments for, PTE.

Predicting intraventricular hemorrhage growth with a machine learning-based, radiomics-clinical model

We constructed a radiomics-clinical model to predict intraventricular hemorrhage (IVH) growth after spontaneous intracerebral hematoma. The model was developed using a training cohort (N=626) and validated with an independent testing cohort (N=270). Radiomics features and clinical predictors were selected using the least absolute shrinkage and selection operator (LASSO) method and multivariate analysis. The radiomics score (Rad-score) was calculated through linear combination of selected features multiplied by their respective LASSO coefficients. The support vector machine (SVM) method was used to construct the model. IVH growth was experienced by 13.4% and 13.7% of patients in the training and testing cohorts, respectively. The Rad-score was associated with severe IVH and poor outcome. Independent predictors of IVH growth included hypercholesterolemia (odds ratio [OR], 0.12 [95%CI, 0.02-0.90]; p=0.039), baseline Graeb score (OR, 1.26 [95%CI, 1.16-1.36]; p<0.001), time to initial CT (OR, 0.70 [95%CI, 0.58-0.86]; p<0.001), international normalized ratio (OR, 4.27 [95%CI, 1.40, 13.0]; p=0.011), and Rad-score (OR, 2.3 [95%CI, 1.6-3.3]; p<0.001). In the training cohort, the model achieved an AUC of 0.78, sensitivity of 0.83, and specificity of 0.66. In the testing cohort, AUC, sensitivity, and specificity were 0.71, 0.81, and 0.64, respectively. This radiomics-clinical model thus has the potential to predict IVH growth.

Artificial intelligence in neurodegenerative diseases: A review of available tools with a focus on machine learning techniques

Neurodegenerative diseases have shown an increasing incidence in the older population in recent years. A significant amount of research has been conducted to characterize these diseases. Computational methods, and particularly machine learning techniques, are now very useful tools in helping and improving the diagnosis as well as the disease monitoring process. In this paper, we provide an in-depth review on existing computational approaches used in the whole neurodegenerative spectrum, namely for Alzheimer's, Parkinson's, and Huntington's Diseases, Amyotrophic Lateral Sclerosis, and Multiple System Atrophy. We propose a taxonomy of the specific clinical features, and of the existing computational methods. We provide a detailed analysis of the various modalities and decision systems employed for each disease. We identify and present the sleep disorders which are present in various diseases and which represent an important asset for onset detection. We overview the existing data set resources and evaluation metrics. Finally, we identify current remaining open challenges and discuss future perspectives.

Abnormal Default-Mode Network Homogeneity in Melancholic and Nonmelancholic Major Depressive Disorder at Rest

Background

Melancholic depression has been assumed as a severe type of major depressive disorder (MDD). We aimed to explore if there were some distinctive alterations in melancholic MDD and whether the alterations could be used to discriminate the melancholic MDD and nonmelancholic MDD.

Methods

Thirty-one outpatients with melancholic MDD, thirty-three outpatients with nonmelancholic MDD, and thirty-two age- and gender-matched healthy controls were recruited. All participants were scanned by resting-state functional magnetic resonance imaging (fMRI). Imaging data were analyzed with the network homogeneity (NH) and support vector machine (SVM) methods.

Results

Both patient groups exhibited increased NH in the right PCC/precuneus and right angular gyrus and decreased NH in the right middle temporal gyrus compared with healthy controls. Compared with nonmelancholic patients and healthy controls, melancholic patients exhibited significantly increased NH in the bilateral superior medial frontal gyrus and decreased NH in the left inferior temporal gyrus. But merely for melancholic patients, the NH of the right middle temporal gyrus was negatively correlated with TEPS total and contextual anticipatory scores. SVM analysis showed that a combination of NH values in the left superior medial frontal gyrus and left inferior temporal gyrus could distinguish melancholic patients from nonmelancholic patients with accuracy, sensitivity, and specificity of 79.66% (47/59), 70.97% (22/31), and 89.29%(25/28), respectively.

Conclusion

Our findings showed distinctive network homogeneity alterations in melancholic MDD which may be potential imaging markers to distinguish melancholic MDD and nonmelancholic MDD.

Can perihaematomal radiomics features predict haematoma expansion?

AIM

To evaluate the association between perihaematomal radiomics features and haematoma expansion (HE).

MATERIALS AND METHODS

Clinical and radiological data were collected retrospectively. The 1:1 propensity score matching (PSM) method was used to balance the difference of baseline characteristics between patients with and without HE. Radiomics features were extracted from the intra- and perihaematomal regions. Top HE-associated features were selected using the minimum redundancy, maximum relevancy algorithm. Support vector machine models were used to predict HE. Predictive performance of radiomics features from different regions was evaluated by receiver operating characteristic curve and confusion matrix-derived metrics.

RESULTS

A total of 1,062 patients were enrolled. After PSM analysis, the propensity score-matched cohort (PSM cohort) included 314 patients (HE: n=157; non-HE: n=157). The PSM cohort was distributed into the training (n=218) and the validation cohorts (n=96). The predictive performance of intra- and perihaematomal features were comparable in the training (area under the receiver operating characteristic curve [AUC], 0.751 versus 0.757; p=0.867) and the validation cohorts (AUC, 0.724 versus 0.671; p=0.454). By incorporating intra- and perihaematomal features, the combined model outperformed the single intrahaematomal model in the training cohort (AUC, 0.872 versus 0.751; p<0.001). Decision curve analysis (DCA) further confirmed the clinical usefulness of the combined model.

CONCLUSION

Perihaematomal radiomics features can predict HE. The integration of intra- and perihaematomal signatures may provide additional benefit to the prediction of HE.

Structural MRI and functional connectivity features predict current clinical status and persistence behavior in prescription opioid users

Prescription opioid use disorder (POUD) has reached epidemic proportions in the United States, raising an urgent need for diagnostic biological tools that can improve predictions of disease characteristics. The use of neuroimaging methods to develop such biomarkers have yielded promising results when applied to neurodegenerative and psychiatric disorders, yet have not been extended to prescription opioid addiction. With this long-term goal in mind, we conducted a preliminary study in this understudied clinical group. Univariate and multivariate approaches to distinguishing between POUD (n = 26) and healthy controls (n = 21) were investigated, on the basis of structural MRI (sMRI) and resting-state functional connectivity (restFC) features. Univariate approaches revealed reduced structural integrity in the subcortical extent of a previously reported addiction-related network in POUD subjects. No reliable univariate between-group differences in cortical structure or edgewise restFC were observed. Contrasting these mixed univariate results, multivariate machine learning classification approaches recovered more statistically reliable group differences, especially when sMRI and restFC features were combined in a multi-modal model (classification accuracy = 66.7%, p < .001). The same multivariate multi-modal approach also yielded reliable prediction of individual differences in a clinically relevant behavioral measure (persistence behavior; predicted-to-actual overlap r = 0.42, p = .009). Our findings suggest that sMRI and restFC measures can be used to reliably distinguish the neural effects of long-term opioid use, and that this endeavor numerically benefits from multivariate predictive approaches and multi-modal feature sets. This can serve as theoretical proof-of-concept for future longitudinal modeling of prognostic POUD characteristics from neuroimaging features, which would have clearer clinical utility.

Recent Advances in Neuroimaging of Epilepsy

Human neuroimaging has had a major impact on the biological understanding of epilepsy and the relationship between pathophysiology, seizure management, and outcomes. This review highlights notable recent advancements in hardware, sequences, methods, analyses, and applications of human neuroimaging techniques utilized to assess epilepsy. These structural, functional, and metabolic assessments include magnetic resonance imaging (MRI), positron emission tomography (PET), and magnetoencephalography (MEG). Advancements that highlight non-invasive neuroimaging techniques used to study the whole brain are emphasized due to the advantages these provide in clinical and research applications. Thus, topics range across presurgical evaluations, understanding of epilepsy as a network disorder, and the interactions between epilepsy and comorbidities. New techniques and approaches are discussed which are expected to emerge into the mainstream within the next decade and impact our understanding of epilepsies. Further, an increasing breadth of investigations includes the interplay between epilepsy, mental health comorbidities, and aberrant brain networks. In the final section of this review, we focus on neuroimaging studies that assess bidirectional relationships between mental health comorbidities and epilepsy as a model for better understanding of the commonalities between both conditions.

Neuroimaging, Urinary, and Plasma Biomarkers of Treatment Response in Huntington's Disease: Preclinical Evidence with the p75NTR Ligand LM11A-31

Huntington's disease (HD) is caused by an expansion of the CAG repeat in the huntingtin gene leading to preferential neurodegeneration of the striatum. Disease-modifying treatments are not yet available to HD patients and their development would be facilitated by translatable pharmacodynamic biomarkers. Multi-modal magnetic resonance imaging (MRI) and plasma cytokines have been suggested as disease onset/progression biomarkers, but their ability to detect treatment efficacy is understudied. This study used the R6/2 mouse model of HD to assess if structural neuroimaging and biofluid assays can detect treatment response using as a prototype the small molecule p75NTR ligand LM11A-31, shown previously to reduce HD phenotypes in these mice. LM11A-31 alleviated volume reductions in multiple brain regions, including striatum, of vehicle-treated R6/2 mice relative to wild-types (WTs), as assessed with in vivo MRI. LM11A-31 also normalized changes in diffusion tensor imaging (DTI) metrics and diminished increases in certain plasma cytokine levels, including tumor necrosis factor-alpha and interleukin-6, in R6/2 mice. Finally, R6/2-vehicle mice had increased urinary levels of the p75NTR extracellular domain (ecd), a cleavage product released with pro-apoptotic ligand binding that detects the progression of other neurodegenerative diseases; LM11A-31 reduced this increase. These results are the first to show that urinary p75NTR-ecd levels are elevated in an HD mouse model and can be used to detect therapeutic effects. These data also indicate that multi-modal MRI and plasma cytokine levels may be effective pharmacodynamic biomarkers and that using combinations of these markers would be a viable and powerful option for clinical trials.

Resting-State Functional Connectivity in Mathematical Expertise

To what extent are different levels of expertise reflected in the functional connectivity of the brain? We addressed this question by using resting-state functional magnetic resonance imaging (fMRI) in mathematicians versus non-mathematicians. To this end, we investigated how the two groups of participants differ in the correlation of their spontaneous blood oxygen level-dependent fluctuations across the whole brain regions during resting state. Moreover, by using the classification algorithm in machine learning, we investigated whether the resting-state fMRI networks between mathematicians and non-mathematicians were distinguished depending on features of functional connectivity. We showed diverging involvement of the frontal-thalamic-temporal connections for mathematicians and the medial-frontal areas to precuneus and the lateral orbital gyrus to thalamus connections for non-mathematicians. Moreover, mathematicians who had higher scores in mathematical knowledge showed a weaker connection strength between the left and right caudate nucleus, demonstrating the connections' characteristics related to mathematical expertise. Separate functional networks between the two groups were validated with a maximum classification accuracy of 91.19% using the distinct resting-state fMRI-based functional connectivity features. We suggest the advantageous role of preconfigured resting-state functional connectivity, as well as the neural efficiency for experts' successful performance.

Machine Learning-Based Radiomics Nomogram for Detecting Extramural Venous Invasion in Rectal Cancer

Objective

To establish and validate a radiomics nomogram based on the features of the primary tumor for predicting preoperative pathological extramural venous invasion (EMVI) in rectal cancer using machine learning.

Methods

The clinical and imaging data of 281 patients with primary rectal cancer from April 2012 to May 2018 were retrospectively analyzed. All the patients were divided into a training set (n = 198) and a test set (n = 83) respectively. The radiomics features of the primary tumor were extracted from the enhanced computed tomography (CT), the T2-weighted imaging (T2WI) and the gadolinium contrast-enhanced T1-weighted imaging (CE-TIWI) of each patient. One optimal radiomics signature extracted from each modal image was generated by receiver operating characteristic (ROC) curve analysis after dimensionality reduction. Three kinds of models were constructed based on training set, including the clinical model (the optimal radiomics signature combining with the clinical features), the magnetic resonance imaging model (the optimal radiomics signature combining with the mrEMVI status) and the integrated model (the optimal radiomics signature combining with both the clinical features and the mrEMVI status). Finally, the optimal model was selected to create a radiomics nomogram. The performance of the nomogram to evaluate clinical efficacy was verified by ROC curves and decision curve analysis curves.

Results

The radiomics signature constructed based on T2WI showed the best performance, with an AUC value of 0.717, a sensitivity of 0.742 and a specificity of 0.621. The radiomics nomogram had the highest prediction efficiency, of which the AUC was 0.863, the sensitivity was 0.774 and the specificity was 0.801.

Conclusion

The radiomics nomogram had the highest efficiency in predicting EMVI. This may help patients choose the best treatment strategy and may strengthen personalized treatment methods to further optimize the treatment effect.

Prostac: A New Composite Score With Potential Predictive Value in Prostate Cancer

Prostate cancer (PCa) is the most commonly diagnosed solid organ cancer in men worldwide. Current diagnosis of PCa includes use of initial prostate specific antigen assay which has a high false positive rate, low specificity, and low sensitivity. The side effects of unnecessary prostate biopsies that healthy men are subjected to, often result in unintended health complications. New PCa biomarkers are being discovered to address this unmet need. Here, we report on the creation of a composite score (Prostac) based on three recently discovered PCa biomarkers, Plasmacytoma Variant Translocation 1 (PVT1) exons 4A, 4B, and 9. Statistical analysis of copy numbers derived from a real-time quantitative polymerase chain (qPCR) reaction - based assay, showed these PCa biomarkers to be linearly separable and significantly over expressed in PCa epithelial cells. We train a supervised learning algorithm using support vector machines to generate a classification hyperplane from which a user-friendly composite score is developed. Cross validation of Prostac using data from prostate epithelial cells (RWPE1) and PCa cells (MDA PCa 2b) accurately classified 100% of PCa cells. Creation of the Prostac score lays the groundwork for clinical trial of its use in PCa diagnosis.

Use of machine learning approach to predict depression in the elderly in China: A longitudinal study

BACKGROUND

Early detection of potential depression among elderly people is conducive for timely preventive intervention and clinical care to improve quality of life. Therefore, depression prediction considering sequential progression patterns in elderly needs to be further explored.

METHODS

We selected 1,538 elderly people from Chinese Longitudinal Healthy Longevity Study (CLHLS) wave 3-7 survey. Long short-term memory (LSTM) and six machine learning (ML) models were used to predict different depression risk factors and the depression risks in the elderly population in the next two years. Receiver operating curve (ROC) and decision curve analysis (DCA) were used to evaluate the prediction accuracy of the reference model and ML models.

RESULTS

The area under the ROC curve (AUC) values of logistic regression with lasso regularisation (AUC=0.629, p-value=0.020) was the highest among ML models. DCA results showed that the net benefit of six ML models was similar (threshold: 0.00-0.10), the net benefit of lasso regression was the largest (threshold: 0.10-0.17 and 0.22-0.25), and the net benefit of DNN was the largest (threshold: 0.17-0.22 and 0.25-0.40). In two ML models, activities of daily living (ADL)/ instrumental ADL (IADL), self-rated health, marital status, arthritis, and number of cohabiting were the most important predictors for elderly with depression.

LIMITATIONS

The retrospective waves used in the LSTM model need to be further increased.

CONCLUSION

The decision support system based on the proposed LSTM+ML model may be very valuable for doctors, nurses and community medical providers for early diagnosis and intervention.

High density optical neuroimaging predicts surgeons's subjective experience and skill levels

Measuring cognitive load is important for surgical education and patient safety. Traditional approaches of measuring cognitive load of surgeons utilise behavioural metrics to measure performance and surveys and questionnaires to collect reports of subjective experience. These have disadvantages such as sporadic data, occasionally intrusive methodologies, subjective or misleading self-reporting. In addition, traditional approaches use subjective metrics that cannot distinguish between skill levels. Functional neuroimaging data was collected using a high density, wireless NIRS device from sixteen surgeons (11 attending surgeons and 5 surgery resident) and 17 students while they performed two laparoscopic tasks (Peg transfer and String pass). Participant’s subjective mental load was assessed using the NASA-TLX survey. Machine learning approaches were used for predicting the subjective experience and skill levels. The Prefrontal cortex (PFC) activations were greater in students who reported higher-than-median task load, as measured by the NASA-TLX survey. However in the case of attending surgeons the opposite tendency was observed, namely higher activations in the lower v higher task loaded subjects. We found that response was greater in the left PFC of students particularly near the dorso- and ventrolateral areas. We quantified the ability of PFC activation to predict the differences in skill and task load using machine learning while focussing on the effects of NIRS channel separation distance on the results. Our results showed that the classification of skill level and subjective task load could be predicted based on PFC activation with an accuracy of nearly 90%. Our finding shows that there is sufficient information available in the optical signals to make accurate predictions about the surgeons’ subjective experiences and skill levels. The high accuracy of results is encouraging and suggest the integration of the strategy developed in this study as a promising approach to design automated, more accurate and objective evaluation methods.

Deep learning applications for the classification of psychiatric disorders using neuroimaging data: Systematic review and meta-analysis

Deep learning (DL) methods have been increasingly applied to neuroimaging data to identify patients with psychiatric and neurological disorders. This review provides an overview of the different DL applications within psychiatry and compares DL model accuracy to standard machine learning (SML). Fifty-three articles were included for qualitative analysis, primarily investigating autism spectrum disorder (ASD; n = 22), schizophrenia (SZ; n = 22) and attention-deficit/hyperactivity disorder (ADHD; n = 9). Thirty-two of the thirty-five studies that directly compared DL to SML reported a higher accuracy for DL. Only sixteen studies could be included in a meta-regression to quantitatively compare DL and SML performance. This showed a higher odds ratio for DL models, though the comparison attained significance only for ASD. Our results suggest that deep learning of neuroimaging data is a promising tool for the classification of individual psychiatric patients. However, it is not yet used to its full potential: most studies use pre-engineered features, whereas one of the main advantages of DL is its ability to learn representations of minimally processed data. Our current evaluation is limited by minimal reporting of performance measures to enable quantitative comparisons, and the restriction to ADHD, SZ and ASD as current research focusses on large publicly available datasets. To truly uncover the added value of DL, we need carefully designed comparisons of SML and DL models which are yet rarely performed.

Translational application of neuroimaging in major depressive disorder: a review of psychoradiological studies

Major depressive disorder (MDD) causes great decrements in health and quality of life with increments in healthcare costs, but the causes and pathogenesis of depression remain largely unknown, which greatly prevent its early detection and effective treatment. With the advancement of neuroimaging approaches, numerous functional and structural alterations in the brain have been detected in MDD and more recently attempts have been made to apply these findings to clinical practice. In this review, we provide an updated summary of the progress in translational application of psychoradiological findings in MDD with a specified focus on potential clinical usage. The foreseeable clinical applications for different MRI modalities were introduced according to their role in disorder classification, subtyping, and prediction. While evidence of cerebral structural and functional changes associated with MDD classification and subtyping was heterogeneous and/or sparse, the ACC and hippocampus have been consistently suggested to be important biomarkers in predicting treatment selection and treatment response. These findings underlined the potential utility of brain biomarkers for clinical practice.

Promises and pitfalls of deep neural networks in neuroimaging-based psychiatric research

By promising more accurate diagnostics and individual treatment recommendations, deep neural networks and in particular convolutional neural networks have advanced to a powerful tool in medical imaging. Here, we first give an introduction into methodological key concepts and resulting methodological promises including representation and transfer learning, as well as modelling domain-specific priors. After reviewing recent applications within neuroimaging-based psychiatric research, such as the diagnosis of psychiatric diseases, delineation of disease subtypes, normative modeling, and the development of neuroimaging biomarkers, we discuss current challenges. This includes for example the difficulty of training models on small, heterogeneous and biased data sets, the lack of validity of clinical labels, algorithmic bias, and the influence of confounding variables.

Deep learning encodes robust discriminative neuroimaging representations to outperform standard machine learning

Recent critical commentaries unfavorably compare deep learning (DL) with standard machine learning (SML) approaches for brain imaging data analysis. However, their conclusions are often based on pre-engineered features depriving DL of its main advantage — representation learning. We conduct a large-scale systematic comparison profiled in multiple classification and regression tasks on structural MRI images and show the importance of representation learning for DL. Results show that if trained following prevalent DL practices, DL methods have the potential to scale particularly well and substantially improve compared to SML methods, while also presenting a lower asymptotic complexity in relative computational time, despite being more complex. We also demonstrate that DL embeddings span comprehensible task-specific projection spectra and that DL consistently localizes task-discriminative brain biomarkers. Our findings highlight the presence of nonlinearities in neuroimaging data that DL can exploit to generate superior task-discriminative representations for characterizing the human brain.

Recent critical commentaries unfavorably compare deep learning (DL) with standard machine learning (SML) for brain imaging data analysis. Here, the authors show that if trained following prevalent DL practices, DL methods substantially improve compared to SML methods by encoding robust discriminative brain representations.

Identification of Major Psychiatric Disorders From Resting-State Electroencephalography Using a Machine Learning Approach

We aimed to develop a machine learning (ML) classifier to detect and compare major psychiatric disorders using electroencephalography (EEG). We retrospectively collected data from medical records, intelligence quotient (IQ) scores from psychological assessments, and quantitative EEG (QEEG) at resting-state assessments from 945 subjects [850 patients with major psychiatric disorders (six large-categorical and nine specific disorders) and 95 healthy controls (HCs)]. A combination of QEEG parameters including power spectrum density (PSD) and functional connectivity (FC) at frequency bands was used to establish models for the binary classification between patients with each disorder and HCs. The support vector machine, random forest, and elastic net ML methods were applied, and prediction performances were compared. The elastic net model with IQ adjustment showed the highest accuracy. The best feature combinations and classification accuracies for discrimination between patients and HCs with adjusted IQ were as follows: schizophrenia = alpha PSD, 93.83%; trauma and stress-related disorders = beta FC, 91.21%; anxiety disorders = whole band PSD, 91.03%; mood disorders = theta FC, 89.26%; addictive disorders = theta PSD, 85.66%; and obsessive-compulsive disorder = gamma FC, 74.52%. Our findings suggest that ML in EEG may predict major psychiatric disorders and provide an objective index of psychiatric disorders.

Deep learning for small and big data in psychiatry

Psychiatry today must gain a better understanding of the common and distinct pathophysiological mechanisms underlying psychiatric disorders in order to deliver more effective, person-tailored treatments. To this end, it appears that the analysis of 'small' experimental samples using conventional statistical approaches has largely failed to capture the heterogeneity underlying psychiatric phenotypes. Modern algorithms and approaches from machine learning, particularly deep learning, provide new hope to address these issues given their outstanding prediction performance in other disciplines. The strength of deep learning algorithms is that they can implement very complicated, and in principle arbitrary predictor-response mappings efficiently. This power comes at a cost, the need for large training (and test) samples to infer the (sometimes over millions of) model parameters. This appears to be at odds with the as yet rather 'small' samples available in psychiatric human research to date (n < 10,000), and the ambition of predicting treatment at the single subject level (n = 1). Here, we aim at giving a comprehensive overview on how we can yet use such models for prediction in psychiatry. We review how machine learning approaches compare to more traditional statistical hypothesis-driven approaches, how their complexity relates to the need of large sample sizes, and what we can do to optimally use these powerful techniques in psychiatric neuroscience.

Individual prediction of symptomatic converters in youth offspring of bipolar parents using proton magnetic resonance spectroscopy

Children of individuals with bipolar disorder (bipolar offspring) are at increased risk for developing mood disorders, but strategies to predict mood episodes are unavailable. In this study, we used support vector machine (SVM) to characterize the potential of proton magnetic resonance spectroscopy (¹H-MRS) in predicting the first mood episode in youth bipolar offspring. From a longitudinal neuroimaging study, 19 at-risk youth who developed their first mood episode (converters), and 19 without mood episodes during follow-up (non-converters) were selected and matched for age, sex and follow-up time. Baseline ¹H-MRS data were obtained from anterior cingulate cortex (ACC) and bilateral ventrolateral prefrontal cortex (VLPFC). Glutamate (Glu), myo-inositol (mI), choline (Cho), N-acetyl aspartate (NAA), and phosphocreatine plus creatine (PCr + Cr) levels were calculated. SVM with a linear kernel was adopted to classify converters and non-converters based on their baseline metabolites. SVM allowed the significant classification of converters and non-converters across all regions for Cho (accuracy = 76.0%), but not for other metabolites. Considering all metabolites within each region, SVM allowed the significant classification of converters and non-converters for left VLPFC (accuracy = 76.5%), but not for right VLPFC or ACC. The combined mI, PCr + Cr, and Cho from left VLPFC achieved the highest accuracy differentiating converters from non-converters (79.0%). Our findings from this exploratory study suggested that ¹H-MRS levels of mI, Cho, and PCr + Cr from left VLPFC might be useful to predict the development of first mood episode in youth bipolar offspring using machine learning. Future studies that prospectively examine and validate these metabolites as predictors of mood episodes in high-risk individuals are necessary.

Disrupted Regional Homogeneity in Melancholic and Non-melancholic Major Depressive Disorder at Rest

Background: Melancholic depression has been viewed as one severe subtype of major depressive disorder (MDD). However, it is unclear whether melancholic depression has distinct changes in brain imaging. We aimed to explore specific or distinctive alterations in melancholic MDD and whether the alterations could be used to separate melancholic MDD from non-melancholic MDD or healthy controls. Materials and Methods: Thirty-one outpatients with melancholic MDD and thirty-three outpatients with non-melancholic MDD and thirty-two age- and gender-matched healthy controls were recruited. All participants were scanned by resting-state functional magnetic resonance imaging (fMRI). Imaging data were analyzed with the regional homogeneity (ReHo) and support vector machine (SVM) methods. Results: Melancholic MDD patients exhibited lower ReHo in the right superior occipital gyrus/middle occipital gyrus than non-melancholic MDD patients and healthy controls. Merely for non-melancholic MDD patients, decreased ReHo in the right middle frontal gyrus was negatively correlated with the total HRSD-17 scores. SVM analysis results showed that a combination of abnormal ReHo in the right fusiform gyrus/cerebellum Crus I and the right superior occipital gyrus/middle occipital gyrus exhibited the highest accuracy of 83.05% (49/59), with a sensitivity of 90.32% (28/31), and a specificity of 75.00% (21/28) for discriminating patients with melancholic MDD from patients with non-melancholic MDD. And a combination of abnormal ReHo in the right fusiform gyrus/cerebellum VI and left postcentral gyrus/precentral gyrus exhibited the highest accuracy of 98.41% (62/63), with a sensitivity of 96.77% (30/31), and a specificity of 100.00%(32/32) for separating patients with melancholic MDD from healthy controls. Conclusion: Our findings showed the distinctive ReHo pattern in patients with melancholic MDD and found brain area that may be associated with the pathophysiology of non-melancholic MDD. Potential imaging markers for discriminating melancholic MDD from non-melancholic MDD or healthy controls were reported.

Autism classified by magnetic resonance imaging: A pilot study of a potential diagnostic tool

OBJECTIVES

Individual anatomical biomarkers have limited power for the classification of autism. The present study introduces a multivariate classification approach using structural magnetic resonance imaging data from individuals with and without autism.

METHODS

The classifier utilizes z-normalization, parameter weighting, and interindividual comparison on brain segmentation data, for estimation of an individual summed total index (TI). The TI indicates whether the gross morphological pattern of each individual's brain is in the direction of cases or controls.

RESULTS

Morphometric analysis found significant differences within subcortical gray matter structures and limbic areas. There was no significant difference in total brain volume. A case-control pilot-study of TIs in normally intelligent individuals with autism (24) and without (21) yielded a maximal accuracy of 78.9% following cross-validation. It showed a high accuracy compared with machine learning methods when tested on the same dataset. The TI correlated well with the autism quotient (R = 0.51) across groups.

CONCLUSION

These results are on par with studies on autism using machine learning. The main contributions are its transparency and simplicity. The possibility of including additional neuroimaging data further increases the potential of the classifier as a diagnostic aid for neuropsychiatric disorders, as well as a research tool for neuroscientific investigations.

Uncovering the prognostic gene signatures for the improvement of risk stratification in cancers by using deep learning algorithm coupled with wavelet transform

Background

The aim of gene expression-based clinical modelling in tumorigenesis is not only to accurately predict the clinical endpoints, but also to reveal the genome characteristics for downstream analysis for the purpose of understanding the mechanisms of cancers. Most of the conventional machine learning methods involved a gene filtering step, in which tens of thousands of genes were firstly filtered based on the gene expression levels by a statistical method with an arbitrary cutoff. Although gene filtering procedure helps to reduce the feature dimension and avoid overfitting, there is a risk that some pathogenic genes important to the disease will be ignored.

Results

In this study, we proposed a novel deep learning approach by combining a convolutional neural network with stationary wavelet transform (SWT-CNN) for stratifying cancer patients and predicting their clinical outcomes without gene filtering based on tumor genomic profiles. The proposed SWT-CNN overperformed the state-of-art algorithms, including support vector machine (SVM) and logistic regression (LR), and produced comparable prediction performance to random forest (RF). Furthermore, for all the cancer types, we firstly proposed a method to weight the genes with the scores, which took advantage of the representative features in the hidden layer of convolutional neural network, and then selected the prognostic genes for the Cox proportional-hazards regression. The results showed that risk stratifications can be effectively improved by using the identified prognostic genes as feature, indicating that the representative features generated by SWT-CNN can well correlate the genes with prognostic risk in cancers and be helpful for selecting the prognostic gene signatures.

Conclusions

Our results indicated that gene expression-based SWT-CNN model can be an excellent tool for stratifying the prognostic risk for cancer patients. In addition, the representative features of SWT-CNN were validated to be useful for evaluating the importance of the genes in the risk stratification and can be further used to identify the prognostic gene signatures.

Convolutional Support Vector Models: Prediction of Coronavirus Disease Using Chest X-rays

The disease caused by the new coronavirus (COVID-19) has been plaguing the world for months and the number of cases are growing more rapidly as the days go by. Therefore, finding a way to identify who has the causative virus is impressive, in order to find a way to stop its proliferation. In this paper, a complete and applied study of convolutional support machines will be presented to classify patients infected with COVID-19 using X-ray data and comparing them with traditional convolutional neural network (CNN). Based on the fitted models, it was possible to observe that the convolutional support vector machine with the polynomial kernel (CSVMPol) has a better predictive performance. In addition to the results obtained based on real images, the behavior of the models studied was observed through simulated images, where it was possible to observe the advantages of support vector machine (SVM) models.

Machine-learning classification of texture features of portable chest X-ray accurately classifies COVID-19 lung infection

BACKGROUND

The large volume and suboptimal image quality of portable chest X-rays (CXRs) as a result of the COVID-19 pandemic could post significant challenges for radiologists and frontline physicians. Deep-learning artificial intelligent (AI) methods have the potential to help improve diagnostic efficiency and accuracy for reading portable CXRs.

PURPOSE

The study aimed at developing an AI imaging analysis tool to classify COVID-19 lung infection based on portable CXRs.

MATERIALS AND METHODS

Public datasets of COVID-19 (N = 130), bacterial pneumonia (N = 145), non-COVID-19 viral pneumonia (N = 145), and normal (N = 138) CXRs were analyzed. Texture and morphological features were extracted. Five supervised machine-learning AI algorithms were used to classify COVID-19 from other conditions. Two-class and multi-class classification were performed. Statistical analysis was done using unpaired two-tailed t tests with unequal variance between groups. Performance of classification models used the receiver-operating characteristic (ROC) curve analysis.

RESULTS

For the two-class classification, the accuracy, sensitivity and specificity were, respectively, 100%, 100%, and 100% for COVID-19 vs normal; 96.34%, 95.35% and 97.44% for COVID-19 vs bacterial pneumonia; and 97.56%, 97.44% and 97.67% for COVID-19 vs non-COVID-19 viral pneumonia. For the multi-class classification, the combined accuracy and AUC were 79.52% and 0.87, respectively.

CONCLUSION

AI classification of texture and morphological features of portable CXRs accurately distinguishes COVID-19 lung infection in patients in multi-class datasets. Deep-learning methods have the potential to improve diagnostic efficiency and accuracy for portable CXRs.

Support vector machine-based classification of schizophrenia patients and healthy controls using structural magnetic resonance imaging from two independent sites

Structural brain alterations have been repeatedly reported in schizophrenia; however, the pathophysiology of its alterations remains unclear. Multivariate pattern recognition analysis such as support vector machines can classify patients and healthy controls by detecting subtle and spatially distributed patterns of structural alterations. We aimed to use a support vector machine to distinguish patients with schizophrenia from control participants on the basis of structural magnetic resonance imaging data and delineate the patterns of structural alterations that significantly contributed to the classification performance. We used independent datasets from different sites with different magnetic resonance imaging scanners, protocols and clinical characteristics of the patient group to achieve a more accurate estimate of the classification performance of support vector machines. We developed a support vector machine classifier using the dataset from one site (101 participants) and evaluated the performance of the trained support vector machine using a dataset from the other site (97 participants) and vice versa. We assessed the performance of the trained support vector machines in each support vector machine classifier. Both support vector machine classifiers attained a classification accuracy of >70% with two independent datasets indicating a consistently high performance of support vector machines even when used to classify data from different sites, scanners and different acquisition protocols. The regions contributing to the classification accuracy included the bilateral medial frontal cortex, superior temporal cortex, insula, occipital cortex, cerebellum, and thalamus, which have been reported to be related to the pathogenesis of schizophrenia. These results indicated that the support vector machine could detect subtle structural brain alterations and might aid our understanding of the pathophysiology of these changes in schizophrenia, which could be one of the diagnostic findings of schizophrenia.

Functional Connectivity Combined With a Machine Learning Algorithm Can Classify High-Risk First-Degree Relatives of Patients With Schizophrenia and Identify Correlates of Cognitive Impairments

Schizophrenia (SCZ) is an inherited disease, with the familial risk being among the most important factors when evaluating an individual's risk for SCZ. However, robust imaging biomarkers for the disease that can be used for diagnosis and determination of the prognosis are lacking. Here, we explore the potential of functional connectivity (FC) for use as a biomarker for the early detection of high-risk first-degree relatives (FDRs). Thirty-eight first-episode SCZ patients, 38 healthy controls (HCs), and 33 FDRs were scanned using resting-state functional magnetic resonance imaging. The subjects' brains were parcellated into 200 regions using the Craddock atlas, and the FC between each pair of regions was used as a classification feature. Multivariate pattern analysis using leave-one-out cross-validation achieved a correct classification rate of 88.15% [sensitivity 84.06%, specificity 92.18%, and area under the receiver operating characteristic curve (AUC) 0.93] for differentiating SCZ patients from HCs. FC located within the default mode, frontal-parietal, auditory, and sensorimotor networks contributed mostly to the accurate classification. The FC patterns of each FDR were input into each classification model as test data to obtain a corresponding prediction label (a total of 76 individual classification scores), and the averaged individual classification score was then used as a robust measure to characterize whether each FDR showed an SCZ-type or HC-type FC pattern. A significant negative correlation was found between the average classification scores of the FDRs and their semantic fluency scores. These findings suggest that FC combined with a machine learning algorithm could help to predict whether FDRs are likely to show an SCZ-specific or HC-specific FC pattern.

Denouements of machine learning and multimodal diagnostic classification of Alzheimer's disease

Alzheimer's disease (AD) is the most common type of dementia. The exact cause and treatment of the disease are still unknown. Different neuroimaging modalities, such as magnetic resonance imaging (MRI), positron emission tomography, and single-photon emission computed tomography, have played a significant role in the study of AD. However, the effective diagnosis of AD, as well as mild cognitive impairment (MCI), has recently drawn large attention. Various technological advancements, such as robots, global positioning system technology, sensors, and machine learning (ML) algorithms, have helped improve the diagnostic process of AD. This study aimed to determine the influence of implementing different ML classifiers in MRI and analyze the use of support vector machines with various multimodal scans for classifying patients with AD/MCI and healthy controls. Conclusions have been drawn in terms of employing different classifier techniques and presenting the optimal multimodal paradigm for the classification of AD.

Dysfunctional Architecture Underlies White Matter Hyperintensities with and without Cognitive Impairment

BACKGROUND

White matter hyperintensities (WMH) are common in older adults and are associated with cognitive decline. However, little is known about the functional changes underlying cognitive decline in WMH subjects.

OBJECTIVES

To investigate whole-brain functional connectivity (FC) underpinnings of cognitive decline in WMH subjects using univariate and multivariate analyses.

METHODS

Twenty-three WMH subjects with mild cognitive impairment (WMH-MCI), 43 WMH subjects with no cognitive impairment (WMH-nCI), and 55 healthy controls underwent resting-state functional MRI scans. Whole-brain FC was calculated using the fine-grained human Brainnetome Atlas, followed by performance of between-group comparisons and FC-cognition correlation analysis. A multivariate analysis using support vector machine (SVM) was performed to classify WMH-MCI and WMH-nCI subjects based on FC.

RESULTS

Both the WMH-MCI and WMH-nCI subjects exhibited characteristic impaired FC patterns. Markedly reduced FC involving subcortical nuclei and cortical hub regions of cognitive networks, especially the cingulate cortex, was identified in the WMH-MCI patients. In the WMH-MCI group, several connections involving the cingulate cortex were associated with cognitive decline. The exploratory mediation analyses indicated that FC alterations could partially explain the association between WMH and cognition. Furthermore, an SVM classifier based on FC distinguished WMH-MCI and WMH-nCI subjects with 78.8% accuracy. Connections that contributed most to the classification showed a similar distribution as the connections identified in the univariate analysis.

CONCLUSIONS

This study provides a new window into the pathophysiology of cognitive impairment in WMH subjects and offer a novel and potential approach for early detection of the cognitive impairment in WMH subjects at the individual level.

Three Dimensional Analysis of SPECT Images for Diagnosing Early Parkinson's Disease using Radial Basis Function Kernel - Extreme Learning Machine

BACKGROUND

Parkinson's Disease (PD) is caused by the deficiency of dopamine, the neurotransmitter that has an effect on specific uptake region of the substantia nigra. Identification of PD is quite tough at an early stage.

OBJECTIVE

The present work proposes an expert system for three dimensional Single-Photon Emission Computed Tomography (SPECT) image to diagnose the early PD.

METHODS

The transaxial image slices are selected on the basis of their high specific uptake region. The processing techniques like preprocessing, segmentation and feature extraction are implemented to extract the quantification parameters like Intensity, correlation, entropy, skewness and kurtosis of the images. The Support Vector Machine (SVM) and Extreme Learning Machine (ELM) classifiers using Radial Basis Function kernel (RBF) are implemented and their results are compared in order to achieve better performance of the system. The performance of the system is evaluated in terms of sensitivity, specificity analysis, accuracy, Receiver Operating Curve (ROC) and Area Under the Curve (AUC).

RESULTS

It is found that RBF-ELM provides high accuracy of 98.2% in diagnosing early PD. In addition, the similarity among the features is found out using K-means clustering algorithm to compute the threshold level for early PD. The computed threshold level is validated using Analysis of Variance (ANOVA).

CONCLUSION

The proposed system has a great potential to assist the clinicians in the early diagnosis process of PD.

Neuroharmony: A new tool for harmonizing volumetric MRI data from unseen scanners

•We present Neuroharmony, a harmonization tool for images from unseen scanners. •We developed Neuroharmony using a total of 15,026 sMRI images. •The tool was able to reduce scanner-related bias from unseen scans. •Neuroharmony represents a significant step towards imaging-based clinical tools. •Neuroharmony is available at https://github.com/garciadias/Neuroharmony .

Enriching CBT by Neuroscience: Novel Avenues to Achieve Personalized Treatments

Although cognitive behavioral therapy (CBT) is an established and efficient treatment for a variety of common mental disorders, a considerable number of patients do not respond to treatment or relapse after successful CBT. Recent findings and approaches from neuroscience could pave the way for clinical developments to enhance the outcome of CBT. Herein, we will present how neuroscience can offer novel perspectives to better understand (a) the biological underpinnings of CBT, (b) how we can enrich CBT with neuroscience-informed techniques (augmentation of CBT), and (c) why some patients may respond better to CBT than others (predictors of therapy outcomes), thus paving the way for more personalized and effective treatments. We will introduce some key topics and describe a selection of findings from CBT-related research using tools from neuroscience, with the hope that this will provide clinicians and clinical researchers with a brief and comprehensible overview of the field.

The Development of a Short Version of the SIMS Using Machine Learning to Detect Feigning in Forensic Assessment

In the present study, we applied machine learning techniques to evaluate whether the Structured Inventory of Malingered Symptomatology (SIMS) can be reduced in length yet maintain accurate discrimination between consistent participants (i.e., presumed truth tellers) and symptom producers. We applied machine learning item selection techniques on data from Mazza et al. (2019c) to identify the minimum number of original SIMS items that could accurately distinguish between consistent participants, symptom accentuators, and symptom producers in real personal injury cases. Subjects were personal injury claimants who had undergone forensic assessment, which is known to incentivize malingering and symptom accentuation. Item selection yielded short versions of the scale with as few as 8 items (to differentiate between consistent participants and symptom producers) and as many as 10 items (to differentiate between consistent and inconsistent participants). The scales had higher classification accuracy than the original SIMS and did not show the bias that was originally reported between false positives and false negatives.

More stringent criteria are needed for diagnosing internet gaming disorder: Evidence from regional brain features and whole-brain functional connectivity multivariate pattern analyses

BACKGROUND

Internet gaming disorder (IGD) is included in the DSM-5 as a provisional diagnosis. Whether IGD should be regarded as a disorder and, if so, how it should be defined and thresholded have generated considerable debate.

METHODS

In the current study, machine learning was used, based on regional and interregional brain features. Resting-state data from 374 subjects (including 148 IGD subjects with DSM-5 scores ≥5 and 93 IGD subjects with DSM-5 scores ≥6) were collected, and multivariate pattern analysis (MVPA) was employed to classify IGD from recreational game use (RGU) subjects based on regional brain features (ReHo) and communication between brain regions (functional connectivity; FC). Permutation tests were used to assess classifier performance.

RESULTS

The results demonstrated that when using DSM-5 scores ≥5 as the inclusion criteria for IGD subjects, MVPA could not differentiate IGD subjects from RGU, whether based on ReHo or FC features or by using different templates. MVPA could differentiate IGD subjects from RGU better than expected by chance when using DSM-5 scores ≥6 with both ReHo and FC features. The brain regions involved in the default mode network and executive control network and the cerebellum exhibited high discriminative power during classification.

DISCUSSION

The current findings challenge the current IGD diagnostic criteria thresholding proposed in the DSM-5, suggesting that more stringent criteria may be needed for diagnosing IGD. The findings suggest that brain regions involved in the default mode network and executive control network relate importantly to the core criteria for IGD.

Structural neuroimaging biomarkers for obsessive-compulsive disorder in the ENIGMA-OCD consortium: medication matters

No diagnostic biomarkers are available for obsessive-compulsive disorder (OCD). Here, we aimed to identify magnetic resonance imaging (MRI) biomarkers for OCD, using 46 data sets with 2304 OCD patients and 2068 healthy controls from the ENIGMA consortium. We performed machine learning analysis of regional measures of cortical thickness, surface area and subcortical volume and tested classification performance using cross-validation. Classification performance for OCD vs. controls using the complete sample with different classifiers and cross-validation strategies was poor. When models were validated on data from other sites, model performance did not exceed chance-level. In contrast, fair classification performance was achieved when patients were grouped according to their medication status. These results indicate that medication use is associated with substantial differences in brain anatomy that are widely distributed, and indicate that clinical heterogeneity contributes to the poor performance of structural MRI as a disease marker.

Predicting the progression of Parkinson's disease using conventional MRI and machine learning: An application of radiomic biomarkers in whole-brain white matter

PURPOSE

This study aimed to develop and validate a radiomics model based on whole-brain white matter and clinical features to predict the progression of Parkinson disease (PD).

METHODS

PD patient data from the Parkinson's Progress Markers Initiative (PPMI) database was evaluated. Seventy-two PD patients with disease progression, as measured by the Hoehn-Yahr Scale (HYS) (stage 1-5), and 72 PD patients with stable PD were matched by sex, age, and category of HYS and included in the current study. Each individual's T₁ -weighted MRI scans at the baseline timepoint were segmented to isolate whole-brain white matter for radiomics feature extraction. The total dataset was divided into a training and test set according to subject serial number. The size of the training dataset was reduced using the maximum relevance minimum redundancy (mRMR) algorithm to construct a radiomics signature using machine learning. Finally, a joint model was constructed by incorporating the radiomics signature and clinical progression scores. The test data were then used to validate the prediction models, which were evaluated based on discrimination, calibration, and clinical utility.

RESULTS

Based on the overall data, the areas under curve (AUCs) of the joint model, signature and Unified Parkinson Disease Rating Scale III PD rating score were 0.836, 0.795, and 0.550, respectively. Furthermore, the sensitivities were 0.805, 0.875, and 0.292, respectively, and the specificities were 0.722, 0.697, and 0.861, respectively. In addition, the predictive accuracy of the model was 0.827, the sensitivity was 0.829 and the specificity was 0.702 for stage-1 PD. For stage-2 PD, the predictive accuracy of the model was 0.854, the sensitivity was 0.960, and the specificity was 0.600.

CONCLUSION

Our results provide evidence that conventional structural MRI can predict the progression of PD. This work also supports the use of a simple radiomics signature built from whole-brain white matter features as a useful tool for the assessment and monitoring of PD progression.

Image-Based Artificial Intelligence Methods for Product Control of Tablet Coating Quality

Mimicking the human decision-making process is challenging. Especially, many process control situations during the manufacturing of pharmaceuticals are based on visual observations and related experience-based actions. The aim of the present work was to investigate the use of image analysis to classify the quality of coated tablets. Tablets with an increasing amount of coating solution were imaged by fast scanning using a conventional office scanner. A segmentation routine was implemented to the images, allowing the extraction of numeric image-based information from individual tablets. The image preprocessing was performed prior to utilization of four different classification techniques for the individual tablet images. The support vector machine (SVM) technique performed superior compared to a convolutional neural network (CNN) in relation to computational time, and this approach was also slightly better at classifying the tablets correctly. The fastest multivariate method was partial least squares (PLS) regression, but this method was hampered by the inferior classification accuracy of the tablets. Finally, it was possible to create a numerical threshold classification model with an accuracy comparable to the SVM approach, so it is evident that there exist multiple valid options for classifying coated tablets.

Machine Learning in Neuroimaging: A New Approach to Understand Acupuncture for Neuroplasticity

The effects of acupuncture facilitating neural plasticity for treating diseases have been identified by clinical and experimental studies. In the last two decades, the application of neuroimaging techniques in acupuncture research provided visualized evidence for acupuncture promoting neuroplasticity. Recently, the integration of machine learning (ML) and neuroimaging techniques becomes a focus in neuroscience and brings a new and promising approach to understand the facilitation of acupuncture on neuroplasticity at the individual level. This review is aimed at providing an overview of this rapidly growing field by introducing the commonly used ML algorithms in neuroimaging studies briefly and analyzing the characteristics of the acupuncture studies based on ML and neuroimaging, so as to provide references for future research.

Diagnostic Classification of ADHD Versus Control: Support Vector Machine Classification Using Brief Neuropsychological Assessment

Objective: Common methods for clinical diagnosis include clinical interview, behavioral questionnaires, and neuropsychological assessment. These methods rely on clinical interpretation and have variable reliability, sensitivity, and specificity. The goal of this study was to evaluate the utility of machine learning in the prediction and classification of children with ADHD-Combined presentation (ADHD-C) using brief neuropsychological measures (d2 Test of Attention, Children with ADHD-C and typically developing control children completed semi-structured clinical interviews and measures of attention/concentration and parents completed symptom severity questionnaires. Method: We used a forward feature selection method to identify the most informative neuropsychological features for support vector machine (SVM) classification and a decision tree model to derive a rule-based model. Results: The SVM model yielded excellent classification accuracy (100%) of individual children with and without ADHD (1.0). Decision tree algorithms identified individuals with and without ADHD-C with 100% sensitivity and specificity. Conclusion:This study observed highly accurate statistical diagnostic classification, at the individual level, in a sample of children with ADHD-C. The findings suggest data-driven behavioral algorithms based on brief neuropsychological data may present an efficient and accurate diagnostic tool for clinicians.

Machine-learning classification using neuroimaging data in schizophrenia, autism, ultra-high risk and first-episode psychosis

Neuropsychiatric disorders are diagnosed based on behavioral criteria, which makes the diagnosis challenging. Objective biomarkers such as neuroimaging are needed, and when coupled with machine learning, can assist the diagnostic decision and increase its reliability. Sixty-four schizophrenia, 36 autism spectrum disorder (ASD), and 106 typically developing individuals were analyzed. FreeSurfer was used to obtain the data from the participant's brain scans. Six classifiers were utilized to classify the subjects. Subsequently, 26 ultra-high risk for psychosis (UHR) and 17 first-episode psychosis (FEP) subjects were run through the trained classifiers. Lastly, the classifiers' output of the patient groups was correlated with their clinical severity. All six classifiers performed relatively well to distinguish the subject groups, especially support vector machine (SVM) and Logistic regression (LR). Cortical thickness and subcortical volume feature groups were most useful for the classification. LR and SVM were highly consistent with clinical indices of ASD. When UHR and FEP groups were run with the trained classifiers, majority of the cases were classified as schizophrenia, none as ASD. Overall, SVM and LR were the best performing classifiers. Cortical thickness and subcortical volume were most useful for the classification, compared to surface area. LR, SVM, and DT's output were clinically informative. The trained classifiers were able to help predict the diagnostic category of both UHR and FEP Individuals.

Towards a brain-based predictome of mental illness

Neuroimaging-based approaches have been extensively applied to study mental illness in recent years and have deepened our understanding of both cognitively healthy and disordered brain structure and function. Recent advancements in machine learning techniques have shown promising outcomes for individualized prediction and characterization of patients with psychiatric disorders. Studies have utilized features from a variety of neuroimaging modalities, including structural, functional, and diffusion magnetic resonance imaging data, as well as jointly estimated features from multiple modalities, to assess patients with heterogeneous mental disorders, such as schizophrenia and autism. We use the term "predictome" to describe the use of multivariate brain network features from one or more neuroimaging modalities to predict mental illness. In the predictome, multiple brain network-based features (either from the same modality or multiple modalities) are incorporated into a predictive model to jointly estimate features that are unique to a disorder and predict subjects accordingly. To date, more than 650 studies have been published on subject-level prediction focusing on psychiatric disorders. We have surveyed about 250 studies including schizophrenia, major depression, bipolar disorder, autism spectrum disorder, attention-deficit hyperactivity disorder, obsessive-compulsive disorder, social anxiety disorder, posttraumatic stress disorder, and substance dependence. In this review, we present a comprehensive review of recent neuroimaging-based predictomic approaches, current trends, and common shortcomings and share our vision for future directions.

Effects of immunization with heat-killed Mycobacterium vaccae on autism spectrum disorder-like behavior and epileptogenesis in a rat model of comorbid autism and epilepsy

Autism spectrum disorders (ASDs) and epilepsy are often comorbid. The basis for this co-occurrence remains unknown; however, inflammatory stressors during development are a shared risk factor. To explore this association, we tested the effect of repeated immunizations using a heat-killed preparation of the stress-protective immunoregulatory microbe Mycobacterium vaccae NCTC 11,659 (M. vaccae) on the behavioral and epileptogenic consequences of the combined stress-terbutaline (ST) rat model of ASD-like behavior/epilepsy. Repeated immunization of the dam with M. vaccae during pregnancy, followed by immunization of the pups after terbutaline injections, prevented the expression of ASD-like behavior but did not appear to protect against, and may have even enhanced, the spontaneous epileptogenic effects of ST. Maternal M. vaccae injections transferred an anti-inflammatory immunophenotype to offspring, and repeated injections across development prevented ST-induced increases in microglial density at early developmental time points in a region-specific manner. Despite epidemiological comorbidity between ASD/epileptic conditions and shared environmental risk factors, our results suggest that the expression of ASD-like behaviors, but perhaps not epileptogenesis, is sensitive to early anti-inflammatory intervention. These data provide support for the exploration of immunoregulatory strategies to prevent the negative neurodevelopmental behavioral effects of stressors during early critical periods.

Detecting schizophrenia at the level of the individual: relative diagnostic value of whole-brain images, connectome-wide functional connectivity and graph-based metrics

BACKGROUND

Previous studies using resting-state functional neuroimaging have revealed alterations in whole-brain images, connectome-wide functional connectivity and graph-based metrics in groups of patients with schizophrenia relative to groups of healthy controls. However, it is unclear which of these measures best captures the neural correlates of this disorder at the level of the individual patient.

METHODS

Here we investigated the relative diagnostic value of these measures. A total of 295 patients with schizophrenia and 452 healthy controls were investigated using resting-state functional Magnetic Resonance Imaging at five research centres. Connectome-wide functional networks were constructed by thresholding correlation matrices of 90 brain regions, and their topological properties were analyzed using graph theory-based methods. Single-subject classification was performed using three machine learning (ML) approaches associated with varying degrees of complexity and abstraction, namely logistic regression, support vector machine and deep learning technology.

RESULTS

Connectome-wide functional connectivity allowed single-subject classification of patients and controls with higher accuracy (average: 81%) than both whole-brain images (average: 53%) and graph-based metrics (average: 69%). Classification based on connectome-wide functional connectivity was driven by a distributed bilateral network including the thalamus and temporal regions.

CONCLUSION

These results were replicated across the three employed ML approaches. Connectome-wide functional connectivity permits differentiation of patients with schizophrenia from healthy controls at single-subject level with greater accuracy; this pattern of results is consistent with the 'dysconnectivity hypothesis' of schizophrenia, which states that the neural basis of the disorder is best understood in terms of system-level functional connectivity alterations.

Toward Robust Anxiety Biomarkers: A Machine Learning Approach in a Large-Scale Sample

BACKGROUND

The field of psychiatry has long sought biomarkers that can objectively diagnose patients, predict treatment response, or identify individuals at risk of illness onset. However, reliable psychiatric biomarkers have yet to emerge. The recent application of machine learning techniques to develop neuroimaging-based biomarkers has yielded promising preliminary results. However, much of the work in this domain has not met best practice standards from the field of machine learning. This is especially true for studies of anxiety, creating uncertainty about the potential for anxiety biomarker development.

METHODS

We applied machine learning tools to predict trait anxiety from neuroimaging measurements in humans. Using publicly available data from the Brain Genomics Superstruct Project, we compared a suite of neuroimaging-based machine learning models predicting anxiety within a discovery sample (n = 531, 307 women) via k-fold cross-validation, and we tested the final model (a stacked model incorporating region-to-region functional connectivity, amygdala seed-to-voxel connectivity, and volumetric and cortical thickness data) in a held-out, unseen test sample (n = 348, 209 women).

RESULTS

Though the best model was able to predict anxiety within the discovery sample (cross-validated R2 of .06, permutation test p < .001), the generalization test within the holdout sample failed (R2 of -.04, permutation test p > .05).

CONCLUSIONS

In this study, we did not find evidence of a generalizable anxiety biomarker. However, we encourage other researchers to investigate this topic, utilizing large samples and proper methodology, to clarify the potential of neuroimaging-based anxiety biomarkers.

Fractures in the framework: limitations of classification systems in psychiatry

This article examines the limitations of existing classification systems from the historical, cultural, political, and legal perspectives. It covers the evolution of classification systems with particular emphasis on the DSM and ICD systems. While pointing out the inherent Western bias in these systems, it highlights the potential of misuse of these systems to subserve other agendas. It raises concerns about the reliability, validity, comorbidity, and heterogeneity within diagnostic categories of contemporary classification systems. Finally, it postulates future directions in alternative methods of diagnosis and classification factoring in advances in artificial intelligence, machine learning, genetic testing, and brain imaging. In conclusion, it emphasizes the need to go beyond the limitations inherent in classifications systems to provide more relevant diagnoses and effective treatments.