Graph machine learning-based classification of PTSD using verbal memory task-based fMRI

1443 

Abstract Submission 

Shin-Eui Park¹, Yeong-Jae Jeon^2,3, Kyungmin Nam⁴, Alex Bhogal⁴, Hyeon-Man Baek^2,3, Jong-Il Park⁵, Jong-Chul Yang⁵

¹Gachon University, Incheon, Incheon, ²Lee Gil Ya Cancer & Diabetes Institute, Gachon University, Incheon, Korea, Republic of, ³Department of Health Sciences and Technology, GAIHST, Gachon University, Incheon, Korea, Republic of, ⁴Centre for Image Sciences, Department of High Field MR, University Medical Centre Utrecth, Utrech, The Netherlands, ⁵Department of Psychiatry, Jeonuk National University Medical School, Jeonju, Jeonbuk

Shin-Eui Park  
Gachon University
Incheon, Incheon

Yeong-Jae Jeon  
Lee Gil Ya Cancer & Diabetes Institute, Gachon University|Department of Health Sciences and Technology, GAIHST, Gachon University
Incheon, Korea, Republic of|Incheon, Korea, Republic of

Kyungmin Nam  
Centre for Image Sciences, Department of High Field MR, University Medical Centre Utrecth
Utrech, The Netherlands

Alex Bhogal  
Centre for Image Sciences, Department of High Field MR, University Medical Centre Utrecth
Utrech, The Netherlands

Hyeon-Man Baek  
Lee Gil Ya Cancer & Diabetes Institute, Gachon University|Department of Health Sciences and Technology, GAIHST, Gachon University
Incheon, Korea, Republic of|Incheon, Korea, Republic of

Jong-Il Park  
Department of Psychiatry, Jeonuk National University Medical School
Jeonju, Jeonbuk

Jong-Chul Yang  
Department of Psychiatry, Jeonuk National University Medical School
Jeonju, Jeonbuk

This study aims to develop a classification model to distinguish between individuals with Post-Traumatic Stress Disorder(PTSD) and healthy controls(HC). Utilizing memory task-based fMRI data, we can seek to identify distinctive patterns in brain activity that can serve as markers for the classification of PTSD and HC individuals. The objective is to contribute to the development of a reliable and objective method for differentiating between individuals with PTSD and those without utilizing graph machine learning algorithm, potentially enhancing diagnostic precision and informing targeted interventions.

23 healthy controls and 13 individuals diagnosed with PTSD underwent verbal memory task fMRI scans (Fig. 1.a). Graph analysis was conducted using the CONN-fMRI FC toolbox (version 21a)[1], in conjunction with SPM12 software. Graph matrices, featuring 164 nodes as Regions of Interest (ROI), were constructed, and a connectivity matrix was derived through ROI-to-ROI correlations[2]. Following the acquisition of the connectivity matrix, thresholding was applied to eliminate less relevant information, resulting in an adjacency matrix as part of the preprocessing. Subsequently, low-dimensional features (specifically, 10 features) were extracted from functional brain connectivity networks using the "graph2vec" graph embedding technique[3]. Our classification approach involved employing RandomForest, XGBoost, Support Vector Classifier (SVC), Gaussian Naïve Bayes (Gaussian NB), and k-Nearest Neighbors (KNN) models to identify PTSD-related brain networks based on the aforementioned low-dimensional features (Fig. 1.b). Classification results were compared against the ground-truth to derive classification accuracy. To ensure robust prediction accuracy, we employed k-fold cross-validation with k=5.

Figure 2 illustrates the classification accuracy achieved across the rest, encoding, and retrieval datasets. In the rest dataset, XGBoost exhibited the highest performance with a 73% classification accuracy at the 0.05 threshold level, while Gaussian NB and KNN models achieved a comparable accuracy of 73% at the 0.15 threshold level. For the encoding dataset, XGBoost demonstrated the most effective classification performance, achieving a 75% accuracy at the 0.25 threshold level. Finally, in the retrieval dataset, Gaussian NB emerged as the best-performing model, attaining a 70% classification accuracy at the 0.45 threshold level.

This study distinguishes itself from prior machine learning-based classification studies because we utilize memory tasks in the fMRI dataset for PTSD. While most machine learning studies[4,5] rely on resting-state fMRI due to its stability and sufficient scan time, our approach focuses on understanding how the brain functions during memory tasks, enabling us to capture more sensitive brain activation during the memory process. Notably, verbal memory impairment is a prominent feature in PTSD[6,7]. Focusing on memory tasks helps us identify unique patterns associated with memory in people with PTSD. Our study is the first to use memory task-based fMRI along with graph machine learning algorithms to identify PTSD. The different brain functional network patterns observed during memory tasks provide valuable insights for clinical understanding and identification of PTSD.
In our results, XGBoost demonstrated strong performance in both the rest and encoding datasets, indicating its ability to effectively capture complex patterns associated with PTSD. On the other hand, Gaussian NB was outstanding in the retrieval dataset, highlighting its proficiency, particularly in the memory recall process. The result that different models succeeded in various datasets highlights importance of customizing machine learning methods for specific memory processing in PTSD. This comprehensive approach, incorporating memory tasks and graph machine learning, contributes to a detailed understanding of PTSD's neural correlates and its identification.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ²

Learning and Memory Other

Activation (eg. BOLD task-fMRI)

Classification and Predictive Modeling ¹

Machine Learning

Memory

Psychiatric Disorders