Machine learning for differentiating between pancreatobiliary-type and intestinal-type periampullary carcinomas based on CT imaging and clinical findings.

PURPOSE: To develop a diagnostic model for distinguishing pancreatobiliary-type and intestinal-type periampullary adenocarcinomas using preoperative contrast-enhanced computed tomography (CT) findings combined with clinical characteristics. METHODS: This retrospective study included 140 patients with periampullary adenocarcinoma who underwent preoperative enhanced CT, including pancreaticobiliary (N = 100) and intestinal (N = 40) types. They were randomly assigned to the training or internal validation set in an 8:2 ratio. Additionally, an independent external cohort of 28 patients was enrolled. Various CT features of the periampullary region were evaluated and data from clinical and laboratory tests were collected. Five machine learning classifiers were developed to identify the histologic type of periampullary adenocarcinoma, including logistic regression, random forest, multi-layer perceptron, light gradient boosting, and eXtreme gradient boosting (XGBoost). RESULTS: All machine learning classifiers except multi-layer perceptron used achieved good performance in distinguishing pancreatobiliary-type and intestinal-type adenocarcinomas, with the area under the curve (AUC) ranging from 0.75 to 0.98. The AUC values of the XGBoost classifier in the training set, internal validation set and external validation set are 0.98, 0.89 and 0.84 respectively. The enhancement degree of tumor, the growth pattern of tumor, and carbohydrate antigen 19-9 were the most important factors in the model. CONCLUSION: Machine learning models combining CT with clinical features can serve as a noninvasive tool to differentiate the histological subtypes of periampullary adenocarcinoma, in particular using the XGBoost classifier.

Disrupted Causal Connectivity Anchored on the Right Anterior Insula in Drug-Naive First-Episode Patients With Depressive Disorder.

Object: Major depressive disorder (MDD) has been demonstrated to be associated with abnormalities in neural networks. However, few studies examined information flow in the salience network (SN). This study examined abnormalities in the causal connectivity between the SN and whole brain in drug-naive first-episode patients with MDD in the resting state. Methods: Based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) diagnostic criteria, 23 drug-naive first-episode MDD patients and 20 matched healthy individuals were recruited and underwent a resting-state magnetic resonance scan. The acquired functional image data were preprocessed using resting-state functional magnetic resonance imaging (rs-fMRI) data analysis toolkit plus (RESTplus). Then, using the data processing & analysis for brain imaging (DPABI) software and a coefficient-based general component analysis method with the right anterior insula (rAI) as the region of interest (ROI), the causal connectivity of the SN with the whole brain and its correlation with cognitive and mental performance were examined in the resting state. Results: (1) The MDD group showed a significantly higher Hamilton Depression Rating Scale total score and significantly higher scores for anxiety, cognitive disturbance, and block factors compared with normal controls. (2) Compared with control: from whole brain to the rAI, the MDD group showed a lower causal connectivity in the left inferior frontal gyrus; from the rAI to the whole brain, the MDD group showed a lower causal connectivity in the right cingulate gyrus, the right precuneus, and extending to paracentral lobule but higher causal connectivity in the left inferior and middle frontal gyrus. (3) In the MDD group, from rAI to the whole brain, the causal connectivity values for the right cingulate gyrus/precuneus were negatively correlated with the score of Stroop Color-Word Test A, B, and C as well as interference times. Conclusion: Our results indicated disrupted causal connectivity among the default mode network (DMN), the central executive network (CEN), and SN in drug-naive first-episode MDD patients. Especially, our results suggest a unique role for rAI in the ordered or hierarchical information processing, presumed to include bottom-up and top-down reciprocal influences among the three networks in MDD.

Desynchronized white matter function and structure in drug-naive first-episode major depressive disorder patients.

Background: Major depressive disorder (MDD) is a highly prevalent mental disease. Using magnetic resonance imaging (MRI), although numerous studies have revealed the alterations in structure and function of grey matter (GM), few studies focused on the synchronization of white matter (WM) structure and function in MDD. The aim of this study was to investigate whether functional and structural abnormalities of WM play an essential role in the neurobiological mechanisms of MDD. Methods: Gradient-echo imaging sequences at 3.0T were used to gather resting state functional MRI (rsfMRI) data, which were performed on 33 drug-naive first-episode MDD patients and 34 healthy controls (HCs). After data preprocessed, amplitude of low frequency fluctuation (ALFF) of WM was calculated. ALFF values in different frequency bands were analyzed, including typical (0.01-0.15 Hz) band, slow-4 (0.027-0.073 Hz) and slow-5 (0.01-0.027 Hz) bands. In addition, the fractional anisotropy (FA) values in WM in 23 patients and 26 HCs were examined using tract-based spatial statistics (TBSS) and tractography based on diffusion tensor imaging (DTI). Pearson correlation analysis was applied to analyze the relationships between ALFF values and Hamilton Depression Scale (HAMD) and Hamilton Anxiety Scale (HAMA). Results: Compared with the HCs, MDD patients showed decreased ALFF values in posterior thalamic radiation (PTR) and superior longitudinal fasciculus (SLF) in slow-5 frequency band, no significant differences of ALFF values were found in typical and slow-4 frequency bands. In addition, there were no significant differences in FA values with TBSS analysis as well as the number of fibers in PTR and SLF with tractography analysis between two groups. Further correlation analysis showed that the ALFF value in SLF was negatively correlated with HAMA-2 score (r = -0.548, p FDR = 0.037) in patients. Conclusion: Our results indicated that WM dysfunction may be associated with the pathophysiological mechanism of depression. Our study also suggested that the functional damage of the WM may precedes the structural damage in first-episode MDD patients. Furthermore, for mental disorders, slow-5 frequency band may be a more sensitive functional indicator for early detection of abnormal spontaneous brain activity in WM.