Aphasia Classification Research Articles

Clinical tools accuracy for classification of primary progressive aphasia: a systematic review protocol

Clinical tools accuracy for classification of primary progressive aphasia: a systematic review protocol

Explainable machine learning radiomics model for Primary Progressive Aphasia classification.

Primary Progressive Aphasia (PPA) is a neurodegenerative disease characterized by linguistic impairment. The two main clinical subtypes are semantic (svPPA) and non-fluent/agrammatic (nfvPPA) variants. Diagnosing and classifying PPA patients represents a complex challenge that requires the integration of multimodal information, including clinical, biological, and radiological features. Structural neuroimaging can play a crucial role in aiding the differential diagnosis of PPA and constructing diagnostic support systems. In this study, we conducted a white matter texture analysis on T1-weighted images, including 56 patients with PPA (31 svPPA and 25 nfvPPA), and 53 age- and sex-matched controls. We trained a tree-based algorithm over combined clinical/radiomics measures and used Shapley Additive Explanations (SHAP) model to extract the greater impactful measures in distinguishing svPPA and nfvPPA patients from controls and each other. Radiomics-integrated classification models demonstrated an accuracy of 95% in distinguishing svPPA patients from controls and of 93.7% in distinguishing svPPA from nfvPPA. An accuracy of 93.7% was observed in differentiating nfvPPA patients from controls. Moreover, Shapley values showed the strong involvement of the white matter near left entorhinal cortex in patients classification models. Our study provides new evidence for the usefulness of radiomics features in classifying patients with svPPA and nfvPPA, demonstrating the effectiveness of an explainable machine learning approach in extracting the most impactful features for assessing PPA.

The role of cognitive control and naming in aphasia

The classical aphasia literature has placed considerable emphasis on the language-centered understanding of aphasia and failed to consider the role of executive functions (EFs) regarding different aspects of patients’ performance. Many current studies suggest deficits in EFs in individuals with aphasia, however, the available data is still limited. Here, our aim was to investigate the impairment of EFs and its potential negative effects on naming (slower performance, increased reaction time and/or decreased accuracy). We sought to determine whether the poor performance observed in word fluency task correlated with similar outcomes in naming. Our study involved five Hungarian post stroke aphasic patients (2 males and 3 females) between the ages of 60 and 70, as well as a control group matched for age and gender. The participants were diagnosed with different types of aphasia (global, Wernicke’s, anomic and conduction). This study employed various neuropsychological and linguistic batteries. By comparing the patients’ performance to that of the control group, we aimed to investigate the impacts of stroke. Within the aphasia group, we observed difficulties in following complex commands and a connection between general slowness and reduced accuracy in naming. We concluded that impairment of executive functions may have a negative impact on naming, comprehension, and fluency. Therefore, it is important to consider functional variations in neural networks, and to base our interpretations on the available psychophysiological data in literature. Our findings provide an alternative perspective to the traditional assessment of aphasia and highlight the importance of considering the role of executive functions.

Assessment tools accuracy for classification and diagnosis of Primary Progressive Aphasia: A systematic review and meta-analysis protocol.

Introduction. Primary Progressive Aphasia (PPA) is a syndrome characterized by progressive decline in language function. There are three main PPA syndromes, each one features different language profiles and neuropathologic substrates. Although there are current clinical diagnostic criteria for PPA categorization, the utility of these requires evaluation(s) by specialized staff and the administration of extensive cognitive batteries. A diagnostic tool for PPA is not currently standardized, though some batteries have been developed and/or validated exclusively for PPA categorization. We aim to describe which cognitive/aphasia diagnostic tool has the best accuracy for PPA diagnosis and categorization. Methods and Analysis. MEDLINE (PubMed), EMBASE and Web of Science databases will be searched using adequate search strategies. Studies including original data of possible, probable, and definite PPA cases according to current clinical diagnostic criteria for PPA will be included. Inclusion criteria will be 1) Studies describing data of a cognitive/aphasia clinical battery including at least one test measure (e.g., specificity, positive predictive values, etc.) and 2) PPA diagnosis according to current clinical criteria as the reference standard. Two reviewers will perform the screening and data extraction. Quality assessment will be performed according to the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) guidelines. This systematic review protocol is reported as stated by with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocol (PRISMA-P) 2015 statement. Dissemination. Findings of this systematic review protocol will be disseminated through a publication in a peer-reviewed journal. Results will be helpful to improve the diagnosis and classification of PPA syndromes.

Perspective of Speech and Language Therapists in Turkey on Primary Progressive Aphasia

Primary progressive aphasia (PPA) is a neurodegenerative disease characterized by the insidious and progressive loss of speech and language skills. However, there is no standardized intervention method applied in patients with PPA. It was conducted a virtual survey of SLTs across Turkey, to collect information about the assessment, intervention and management of patients with PPA. A total of 285 SLTs participated the survey; 76.8% ( n = 219) of the respondents were female, 58.9% ( n = 168) were 1 to 2 years post graduation, 62.8% ( n = 179) of them were master’s degree, and 46% ( n = 131) were working in a rehabilitation center. All 285 participants included in the study stated that they did not create a therapy process for their clients with PPA and they performed the therapy processes applied in the classical aphasia treatment. Two hundred sixty-eight participants stated that they had more than six sessions with their clients, and 280 respondents. It is very important for all participants to plan functional interventions that require joint participation with the patient and their family/caregiver or friends, which will support communication and language and communication-oriented intervention. However, there is a need to develop specific evidence-based care pathways for people with PPA so that more efficient interventions can be made. Raising awareness among SLTs and other clinical service providers that rehabilitation is effective can increase the accessibility of interventions for individuals with PPA and lead to a more robust support system for individuals affected by PPA.

The Moroccan Arabic Bedside Western Aphasia Battery-Revised: Linguistic and Psychometric Properties.

The objectives of this study were (a) to linguistically and culturally adapt the English bedside version of the Western Aphasia Battery-Revised (Bedside WAB-R) into Moroccan Arabic (MA) and (b) to assess its psychometric properties. The Bedside MA-WAB-R was piloted on a population of 20 participants after a process of linguistic adaptation from English to MA. To verify the psychometric properties of the test, a group of healthy controls (n = 106) and a group of persons with chronic aphasia (n = 52) completed the Bedside MA-WAB-R. The test's content validity, construct validity, and the concurrent validity of the aphasia classification system were examined. In addition, the interrater reliability, intrarater reliability, test-retest reliability, and internal consistency of the instrument were evaluated. The results indicated that the Bedside MA-WAB-R overall met standard criteria for excellent psychometric properties, as evinced by high content and construct validity and concurrent validity with independent measures of aphasia diagnosis (speech-language pathologists' impression and lesion location) as well as high internal consistency, inter- and intrarater reliability, and test-retest reliability. The test was also found to have very high sensitivity and specificity for the detection of aphasia, as revealed by the performance of controls. By controlling for age and educational level effects, specific cutoff values were determined to optimize the aphasia diagnosis. The Bedside MA-WAB-R is the first standardized quick aphasia assessment tool with associated psychometric properties for clinical use with an MA-speaking population, particularly within the 26- to 58-year age range. It meets standard criteria for a valid and reliable measure than can have a variety of clinical and research applications.

Structural correlates of language processing in primary progressive aphasia.

Understanding the relationships between brain structure and language behaviour in primary progressive aphasia provides crucial information about these diseases' pathomechanisms. However, previous investigations have been limited from providing a statistically reliable view of broad language abilities by sample size, variant focus and task focus. In this study, the authors aimed to determine the relationship between brain structure and language behaviour in primary progressive aphasia, to determine the degree to which task-associated regions were atrophied across disease variants and to determine the degree to which task-related atrophy overlaps across disease variants. Participants were 118 primary progressive aphasia patients and 61 healthy, age-matched controls tested from 2011 to 2018 in the German Consortium for Frontotemporal Lobar Degeneration cohort. Diagnosis of primary progressive aphasia required progressive deterioration of mainly speech and language for ≥ 2 years, and variant was diagnosed by the criteria of Gorno-Tempini et al. (Classification of primary progressive aphasia and its variants. Neurology. 2011;76(11):1006-1014). Twenty-one participants not fulfilling a specific subtype were classified as mixed-variant and excluded. Language tasks of interest included the Boston naming test, a German adaptation of the Repeat and Point task, phonemic and category fluency tasks and the reading/writing subtest of the Aachen Aphasia Test. Brain structure was measured by cortical thickness. We observed networks of language task-associated temporal, frontal and parietal cortex. Overlapping task-associated atrophy was observed in the left lateral, ventral and medial temporal lobes, middle and superior frontal gyri, supramarginal gyrus and insula. Some regions, primarily in the perisylvian region, were associated with language behaviour despite showing no significant atrophy. The results crucially extend less powerful studies associating brain and language measures in primary progressive aphasia. Cross-variant atrophy in task-associated regions suggests partially shared underlying deficits, whereas unique atrophy reinforces variant-specific deficits. Language task-related regions that are not obviously atrophied suggest regions of future network disruption and encourage understanding of task deficits beyond clearly atrophied cortex. These results may pave the way for new treatment approaches.

Evaluating Fluency in Aphasia: Fluency Scales, Trichotomous Judgements, or Machine Learning

ABSTRACT Background Speech-language pathologists (SLPs) and other clinicians often use aphasia batteries, such as the Western Aphasia Battery-Revised (WAB-R), to evaluate both severity and classification of aphasia. However, the fluency scale on the WAB-R is not entirely objective and has been found to have less than ideal inter-rater reliability, due to variability in weighing the importance of one dimension (e.g. articulatory effort or grammaticality) over another. This limitation has implications for aphasia classification. The subjectivity might be mitigated through the implementation of machine learning to identify fluent and non-fluent speech. Aims We hypothesized that two models consisting of convolutional and recurrent neural networks can be used to identify fluent and non-fluent aphasia as judged by SLPs, with greater reliability than use of the WAB-R fluency scale. Methods & Procedures The training and testing dataset for the networks was collected from the public domain, and the validation dataset was collected from participants in post-stroke aphasia studies. We used Kappa scores to evaluate inter-rater reliability among SLPs, and between the networks and SLPs. Outcome and Results Using public domain samples, the model for detecting non-fluent aphasia achieved high accuracy on the training dataset after 10 epochs (i.e., when algorithm scans the entire dataset) and 81% testing accuracy using public domain samples. The model for detecting fluent speech had high training accuracy and 83% testing. Across samples, using the WAB-R fluency scale, there was poor to perfect agreement among SLPs on the precise WAB-R fluency score, but substantial agreement on non-fluent (score 0-4) versus fluent (score of 5-9). The agreement between the model and the SLPs was moderate for identifying non-fluent speech and substantial fpr identifying fluent speech. When SLPs were asked to identify each sample as fluent, non-fluent, or mixed (without using the fluency scale), the agreement between SLPs was almost perfect (Kappa 0.94). The agreement between the SLPs’ trichotomous judgement and the models was fair for detecting non-fluent speech and substantial for detecting fluent speech. Conclusions Results indicate that neither the WAB-R fluency scale nor the machine learning algorithms were as useful (reliable and valid) as a simple trichotomous judgement of fluent, non-fluent, or mixed by SLPs. These results, together with data from the literature, indicate that it is time to re-consider use of the WAB-R fluency scale for classification of aphasia. It is also premature, at present, to rely on machine learning to rate spoken language fluency.

Profile of Language Disorders in Clients of Motor Transcortical Aphasia

Aphasia is a language disorder that occurs due to an injury to the brain. The classification of aphasia is divided into fluent aphasia and nonfluent aphasia. One of the non-fluent aphasia is transcortical motor aphasia. The characteristics that distinguish Broca's aphasia from other aphasia are disorders in receptive language, difficulty in answering questions, difficulty understanding spoken language, difficulty understanding written language, difficulty writing, accompanied by linguistic symptoms such as ecolalia. The implementation of this research starts from the assessment (observation, interview, and test). The purpose of this research is to find out the language disorder that exists in the client. To see the language disorder that exists in clients, interviews with families and clients are carried out. Then an observation of the client's activity is carried out and a test is carried out to determine the language disorder in the client. The conclusion of this study is that clients experience problems in understanding spoken and written language, clients also have difficulty starting to talk to other people. So the authors recommend that clients get treatment for language disorders.

Comprehensive qualitative characterization of linguistic performance profiles in primary progressive aphasia: a multivariate study with FDG-PET

Primary progressive aphasia (PPA) classification relies on profile characterization of quantitatively impaired/spared performance in language tasks. In this study, we coextracted 8 qualitative types of errors in 67 PPA patients submitted to a comprehensive language assessment. Canonical correlation analysis was applied to simultaneously correlate qualitative errors and brain metabolism, collected with FDG-PET. Results showed the contribution of semantic, syntactic and working memory errors associated with specific correlates of regional metabolic changes. Reduced metabolism in the left fusiform gyrus, anterior-middle and inferior-temporal gyri and middle-temporal pole correlated with an increase of semantic errors. Hypometabolism in the left inferior, middle, and superior frontal gyri, insula and right middle-occipital gyrus was related to syntactic errors. Higher metabolism in the bilateral pallidum, putamen, and left thalamus, as well as hypometabolism in the left angular and supramarginal gyri, inferior-parietal lobule, posterior-middle and inferior-temporal gyri and posterior cingulum predicted the increase of working memory errors. A relevant role of working memory subcomponents was associated with distinct neural systems. Patients’ profiles are easily represented in a qualitative multidimensional space, in which mixed PPA overlapped with different phenotypes.

Post-stroke aphasia management – from classical approaches to modern therapies

Post-stroke aphasia is an important topic in clinical rehabilitation neurology, affecting a significant number of stroke patients worldwide each year. Although the anatomical basis of aphasia has been explored for over a century, a revision of disease classification has led to a broader understanding of the subject and has stimulated the field of drug development to yield novel therapies. In this context, the authors commence the current narrative review by presenting the widely accepted aphasia classifications and the corresponding management practices. Subsequently, speech-language therapies are presented, with a focus on the widely preferred patient-centered approach. We conclude with a brief discussion regarding adjunctive treatment options such as pharmacological intervention, non-invasive brain stimulation techniques, and virtual reality rehabilitation.

Linguistic profiles of variants of primary progressive aphasia

BackgroundSeveral subtypes of primary progressive aphasia (PPA) have been proposed. Most reports use small samples, and few have included Spanish-speaking participants. AimTo analyze the language profile and nonlinguistic deficits in a large sample of PPA Spanish monolingual participants. Method177 individuals were diagnosed with PPA in a sample consisting of 69 men and 108 women (Mage = 66.40 years, SD = 9.30). The participants were assessed using the Spanish versions of the Western Aphasia Battery Revised (SWAB-R) and the Boston Diagnostic Aphasia Examination (SBDAE). Non-verbal reasoning was evaluated with the Raven's Colored Progressive Matrices. Results41.8% of the sample met the criteria for the logopenic variant (lvPPA), while 28.2% met the criteria for semantic (svPPA), 15.3% for lexical (lxvPPA), and 14.7% for nonfluent/agrammatic (nfvPPA) variants. Language difficulties were similar in all variants except for lxvPPA. Scores on Spontaneous Language, Auditory Comprehension, Repetition, and Naming were significantly higher for the lxvPPA group. Raven's Colored Progressive Matrices scores were significantly lower in lvPPA. Years of education correlated with all test scores, while age was negatively associated with naming. When the PPA variants were classified according to the traditional aphasia classification, discrepancies were evident. Furthermore, the most frequent type of aphasia was Amnesic, while the least frequent was Wernicke's aphasia. ConclusionThe SWAB-R is useful in describing the clinical characteristics of aphasia for each variant of PPA, but quantitative scores from this battery are not capable of distinguishing between variants of PPA, with the exception of lxvPPA.

Unclassified fluent variants of primary progressive aphasia: distinction from semantic and logopenic variants.

Primary progressive aphasia, a neurodegenerative syndrome, presents mainly with language impairment. Both semantic and logopenic variants are fluent variants of primary progressive aphasia. Before the research criteria of primary progressive aphasia were proposed, progressive fluent aphasias, such as progressive anomic aphasia, transcortical sensory aphasia and Wernicke’s aphasia, were reported as classical progressive fluent aphasias seen in Alzheimer’s disease. However, since the research criteria of primary progressive aphasia were established, classical fluent variants (other than semantic and logopenic variants) have been neglected and have not been included in the current classification of primary progressive aphasia. This study aimed to determine whether unclassified fluent variants (other than semantic and logopenic variants) can be manifestations of primary progressive aphasia. This study also reconfirmed the characteristics of classical progressive fluent aphasia, such as progressive anomic aphasia, progressive transcortical sensory aphasia and progressive Wernicke’s aphasia as unclassified fluent variants of primary progressive aphasia, using comparison with the current model of primary progressive aphasia. Twelve consecutive patients with an unclassified fluent variant other than semantic or logopenic variant underwent language, neurological, neuropsychological and neuroimaging (MRI and single-photon emission computed tomography) testing. Based on comprehensive language tests, we redefined the diagnoses as primary progressive anomic aphasia (n = 8), primary progressive transcortical sensory aphasia (n = 3) and primary progressive Wernicke’s aphasia (n = 1). Anomic aphasia was characterized by anomia but preserved repetition and comprehension; transcortical sensory aphasia by relatively preserved repetition but poor word comprehension; and Wernicke’s aphasia by poor repetition and word comprehension. In patients with anomic aphasia, voxel-based morphometry of MRI data revealed cortical atrophy, which was most prominent in the temporoparietal lobes, with no obvious lateralization; in two-thirds of patients with transcortical sensory aphasia and in one patient with Wernicke’s aphasia, it revealed atrophy, predominantly in the left temporoparietal lobe. Statistical analysis of single-photon emission computed tomography using three-dimensional stereotactic surface projections revealed patterns of left-sided hypoperfusion in the majority of patients. The temporal and parietal lobes were involved in all cases; the degree of hypoperfusion was higher in patients with transcortical sensory aphasia or Wernicke’s aphasia than in patients with anomic aphasia. The present study demonstrated the clinical and imaging features of 12 patients with an unclassified fluent variant of primary progressive aphasia, which we redefined as primary progressive anomic aphasia, primary progressive transcortical sensory aphasia and primary progressive Wernicke’s aphasia. Classical fluent variants other than semantic and logopenic variants can be found in primary progressive aphasia.

Chapter 3 - Vascular syndromes: Revisiting classification of poststroke aphasia

Over 150 years have passed since the first formal description of aphasia associated with localized neurologic damage. In the years since that time, a significant amount of research has been conducted to identify/explain the locations and functions of the brain regions responsible for (or associated with) language as well as to describe the various types of aphasia resulting from injury to these locations. Many of these attempts to associate somewhat predictable patterns of language deficits with damage to specific structures have been confounded by atypical reports and considerable variability in either the behavioral presentation and/or structural damage that directly contradict/oppose some of the proposed theories. However, considering the aphasias as vascular syndromes, or a collection of symptoms associated with damage to various structures supplied by a specific artery, accounts for both the predictability and the variability seen. This chapter presents a brief history of aphasia classification, the vascular territories commonly associated with aphasia, the different aphasic vascular syndromes, and the typical recovery/evolution of aphasia presentation over time.

Enhancing the classification of aphasia: a statistical analysis using connected speech

ABSTRACT Background: Large-shared databases and automated language analyses allow for the application of new data analysis techniques that can shed new light on the connected speech of people with aphasia (PWA). Aims: To identify coherent clusters of PWA based on language output using unsupervised statistical algorithms and to identify features that are most strongly associated with those clusters. Methods & Procedures: Clustering and classification methods were applied to language production data from 168 PWA. Language samples were from a standard discourse protocol tapping four genres: free speech personal narratives, picture descriptions, Cinderella storytelling, and procedural discourse. Outcomes & Results: Seven distinct clusters of PWA were identified by the K-means algorithm. Using the random forest algorithm, a classification tree was proposed and validated, showing 91% agreement with the cluster assignments. This representative tree used only two variables to divide the data into distinct groups: total words from free speech tasks and total closed-class words from the Cinderella storytelling task. Conclusion: Connected speech data can be used to distinguish PWA into coherent groups, providing insight into traditional aphasia classifications, factors that may guide discourse research and clinical work.

A data-driven approach to post-stroke aphasia classification and lesion-based prediction.

Aphasia is an acquired impairment in the production or comprehension of language, typically caused by left hemisphere stroke. The subtyping framework used in clinical aphasiology today is based on the Wernicke-Lichtheim model of aphasia formulated in the late 19th century, which emphasizes the distinction between language production and comprehension. The current study used a data-driven approach that combined modern statistical, machine learning, and neuroimaging tools to examine behavioural deficit profiles and their lesion correlates and predictors in a large cohort of individuals with post-stroke aphasia. First, individuals with aphasia were clustered based on their behavioural deficit profiles using community detection analysis (CDA) and these clusters were compared with the traditional aphasia subtypes. Random forest classifiers were built to evaluate how well individual lesion profiles predict cluster membership. The results of the CDA analyses did not align with the traditional model of aphasia in either behavioural or neuroanatomical patterns. Instead, the results suggested that the primary distinction in aphasia (after severity) is between phonological and semantic processing rather than between production and comprehension. Further, lesion-based classification reached 75% accuracy for the CDA-based categories and only 60% for categories based on the traditional fluent/non-fluent aphasia distinction. The results of this study provide a data-driven basis for a new approach to classification of post-stroke aphasia subtypes in both research and clinical settings.

A Brief Executive Language Screen for Frontal Aphasia.

Aphasia assessment tools have primarily focused on classical aphasia type and severity, with minimal incorporation of recent findings that suggest a significant role of executive control operations in language generation. Assessment of the interface between language and executive functions is needed to improve detection of spontaneous speech difficulties. In this study we develop a new Brief Executive Language Screen (BELS), a brief tool specifically designed to assess core language and executive functions shown to be involved in spontaneous generation of language. Similar to other measures of aphasia, the BELS assesses articulation and core language skills (repetition, naming and comprehension). Unique additions to the BELS include assessments of spontaneous connected speech, word fluency (phonemic/semantic) and sentence completion (verbal initiation, inhibition and selection). One-hundred and eight healthy controls and 136 stroke patients were recruited. Confirmatory factor analysis was used to determine construct validity and logistic regression was used to evaluate the discriminative validity, informing the final version of the BELS. The results showed that the BELS is sensitive for articulation and nominal language deficits, and it measures executive aspects of spontaneous language generation, which is a hallmark of frontal dynamic aphasia. The results have encouraging theoretical and practical implications.

AFASIA PRAGMÁTICA: CÉREBRO, LINGUAGEM E COMUNICAÇÃO

This article describes the linguistic behavior of individuals diagnosed with mild Frontal Lobe Syndrome or Pragmatic Aphasia, a neurological condition resulting from brain damage that affects linguistic areas and the semantic processes of understanding. Resorting to the pragmatic conception of language, in the analysis of the linguistic traits that characterize the pathology of brain-injured subjects who present alterations and contextual communicative difficulties. The theoretical framework for the analysis is composed of the functional model of the brain, which allows the classification of aphasia from a linguistic point of view and, to address the pragmatic and cognitive aspects of the Frontal Lobe Syndrome, the relationships between communication and context are used in the Relevance Theory (SPERBER; WILSON, 2001).

Data-Driven, Visual Framework for the Characterization of Aphasias Across Stroke, Post-resective, and Neurodegenerative Disorders Over Time.

Aphasia classifications and specialized language batteries differ across the fields of neurodegenerative disorders and lesional brain injuries, resulting in difficult comparisons of language deficits across etiologies. In this study, we present a simplified framework, in which a widely-used aphasia battery captures clinical clusters across disease etiologies and provides a quantitative and visual method to characterize and track patients over time. The framework is used to evaluate populations representing three disease etiologies: stroke, primary progressive aphasia (PPA), and post-operative aphasia. A total of 330 patients across three populations with cerebral injury leading to aphasia were investigated, including 76 patients with stroke, 107 patients meeting criteria for PPA, and 147 patients following left hemispheric resective surgery. Western Aphasia Battery (WAB) measures (Information Content, Fluency, answering Yes/No questions, Auditory Word Recognition, Sequential Commands, and Repetition) were collected across the three populations and analyzed to develop a multi-dimensional aphasia model using dimensionality reduction techniques. Two orthogonal dimensions were found to explain 87% of the variance across aphasia phenotypes and three disease etiologies. The first dimension reflects shared weighting across aphasia subscores and correlated with aphasia severity. The second dimension incorporates fluency and comprehension, thereby separating Wernicke's from Broca's aphasia, and the non-fluent/agrammatic from semantic PPA variants. Clusters representing clinical classifications, including late PPA presentations, were preserved within the two-dimensional space. Early PPA presentations were not classifiable, as specialized batteries are needed for phenotyping. Longitudinal data was further used to visualize the trajectory of aphasias during recovery or disease progression, including the rapid recovery of post-operative aphasic patients. This method has implications for the conceptualization of aphasia as a spectrum disorder across different disease etiology and may serve as a framework to track the trajectories of aphasia progression and recovery.

Support vector machine based aphasia classification of transcranial magnetic stimulation language mapping in brain tumor patients

Repetitive TMS (rTMS) allows for non-invasive and transient disruption of local neuronal functioning. We used machine learning approaches to assess whether brain tumor patients can be accurately classified into aphasic and non-aphasic groups using their rTMS language mapping results as input features. Given that each tumor affects the subject-specific language networks differently, resulting in heterogenous rTMS functional mappings, we propose the use of machine learning strategies to classify potential patterns of rTMS language mapping results. We retrospectively included 90 patients with left perisylvian world health organization (WHO) grade II-IV gliomas that underwent presurgical navigated rTMS language mapping. Within our cohort, 29 of 90 (32.2%) patients suffered from at least mild aphasia as shown in the Aachen Aphasia Test based Berlin Aphasia Score (BAS). After spatial normalization to MNI 152 of all rTMS spots, we calculated the error rate (ER) in each stimulated cortical area (28 regions of interest, ROI) by automated anatomical labeling parcellation (AAL3) and IIT. We used a support vector machine (SVM) to classify significant areas in relation to aphasia. After feeding the ROIs into the SVM model, it revealed that in addition to age (w = 2.98), the ERs of the left supramarginal gyrus (w = 3.64), left inferior parietal gyrus (w = 2.28) and right pars triangularis (w = 1.34) contributed more than other features to the model. The model’s sensitivity was 86.2%, the specificity was 82.0%, the overall accuracy was 85.5% and the AUC was 89.3%. Our results demonstrate an increased vulnerability of right inferior pars triangularis to rTMS in aphasic patients due to left perisylvian gliomas. This finding points towards a functional relevant involvement of the right pars triangularis in response to aphasia. The tumor location feature, specified by calculating overlaps with white and grey matter atlases, did not affect the SVM model. The left supramarginal gyrus as a feature improved our SVM model the most. Additionally, our results could point towards a decreasing potential for neuroplasticity with age.