In the field of psychiatry, there is a lack of biomarkers that are directly applicable in diagnostic settings or that clearly reflect illness severity. Due to this issue, it is often unclear how to best begin treatment, and patients' responses to treatment can be difficult to understand. In recent years, there has been a global effort to utilize information and communication technology, and to use machine learning to analyze Big Data collected through those means, in the hopes of finding biomarkers that can be used in the diagnosis of and severity evaluations for psychiatric illnesses. In this paper, we will introduce prior research that particularly focuses on using wearable devices to evaluate depression, as well as our efforts to attempt to use multiple modalities in the screening of and severity assessments for depression.

Image quality of dental panoramic x-ray images with complementary metal oxide semiconductor (CMOS images) is trend to be lower than that with charge coupled device (CCD images). The purpose of this study was to improve the image quality of CMOS images to be comparable to that of CCD images by applying the concept of an example-based super-resolution technique based on a dictionary. Our database consisted of pairs of CMOS images and CCD images obtained from nine volunteers. In the proposed method, a dictionary representing the relationship between CMOS patches and CCD patches was first generated by dividing CMOS images and CCD images into small regions. The virtual CCD images with high image quality were then constructed from the CMOS images by embedding optimal CCD patches selected from the dictionary in each of CMOS patches. With the proposed method, the contrast-to-noise ratio (CNR) and the signalto-noise ratio (SNR) for the virtual CCD images were 13.72±12.48, 22.17±15.65, showing a significant improvement when compared with the CMOS images (CNR : 12.20±10.80, P < .0001, SNR : 20.55±12.90, P < .0001). The virtual CCD images constructed from the CMOS images would be useful in dental image diagnosis.