Application of Autoencoder in Depression Diagnosis

Abstract

Abstract. Major depressive disorder (MDD) is a mental disorder characterized by at least two weeks of low mood which is present across most situations. Diagnosis of MDD using rest-state functional magnetic resonance imaging (fMRI) data faces many challenges due to the high dimensionality, small samples, noisy and individual variability. No method can automatically extract discriminative features from the origin time series in fMRI images for MDD diagnosis. In this study, we proposed a new method for feature extraction and a workflow which can make an automatic feature extraction and classification without a prior knowledge. An autoencoder was used to learn pre-training parameters of a dimensionality reduction process using 3-D convolution network. Through comparison with the other three feature extraction methods, our method achieved the best classification performance. This method can be used not only in MDD diagnosis, but also other similar disorders.

Deep learning in mental health outcome research: a scoping review

“…As reported in the previous study 10 , CNNs perform well in processing raw neuroimage data. Among the studies reviewed in this study, three 29,30,37 reported to involve CNN layers and achieved desirable performances.…”

“…The use of such tools mitigates the overfitting risk of DL models. As reported in some selected studies 28,31,35,37 , the DL models can benefit from feature engineering techniques and have been shown to outperform the traditional ML models in the prediction of multiple conditions such as depression, schizophrenia, and ADHD. However, such tools extract features relying on prior knowledge; hence may omit some information that is meaningful for mental outcome research but unknown yet.…”

“…Moreover, the use of DL in neuroimages also targeted at addressing other mental health disorders. Geng et al 37 proposed to use CNN and autoencoder to acquire meaningful features from the original time series of fMRI data for predicting depression. Two studies 31,38 integrated the fMRI and sMRI data modalities to develop predictive models for ASDs.…”

Neurobiological Perspective and Personalized Treatment in Schizophrenia

“…Finally, the findings suggest that machine learning technologies offer clinicians credible responses when making judgments concerning SZ patients. In the field of biological psychiatry, there is a rising interest in using AI and machine learning [116][117][118][119]. Researchers in machine learning and artificial intelligence (ML and AI) use mathematical models to extract attributes or features from signals and pictures to establish links between the characteristics and brain state to evaluate if the brain is normal [120].…”