2021年 · 第01卷第03期

The aim of this article is to review recent progress in the application of artificial intelligence to chest medical image analysis. The lungs, bone, and mediastinum were included in terms of anatomy, while X-ray and computed tomography (CT), with and without contrast enhancement, were considered regarding imaging modalities. Four key components of deep learning were summarized, namely, network architectures, learning strategies, optimization methods, and vision tasks. Disease-specific applications were discussed in detail with respect to the dimension of the data input, network architecture, and modality: lung cancer, pneumonia, tuberculosis, pulmonary embolism, chronic obstructive pulmonary disease, and interstitial lung disease for lung; traumatic fractures, osteoporosis, osteoporotic fractures, and bone metastases for bone; and coronary artery calcification and aortic dissection for vascular diseases. Finally, five promising research directions and possible solutions were presented for future work.

Medical image reconstruction aims to acquire high-quality medical images for clinical usage at minimal cost and risk to the patients. Deep learning and its applications in medical imaging, especially in image reconstruction have received considerable attention in the literature in recent years. This study reviews records obtained electronically through the leading scientific databases (Magnetic Resonance Imaging journal, Google Scholar, Scopus, Science Direct, Elsevier, and from other journal publications) searched using three sets of keywords: (1) Deep learning, image reconstruction, medical imaging; (2) Medical imaging, Deep learning, Image reconstruction; (3) Open science, Open imaging data, Open software. The articles reviewed revealed that deep learning-based reconstruction methods improve the quality of reconstructed images qualitatively and quantitatively. However, deep learning techniques are generally computationally expensive, require large amounts of training datasets, lack decent theory to explain why the algorithms work, and have issues of generalization and robustness. The challenge of lack of enough training datasets is currently being addressed by using transfer learning techniques.

Developments in genetics and genomics are progressing at an unprecedented speed. Twenty years ago, the human genome project provided the first glimpses into the human genome sequence and launched a new era of human genetics. The emerging of next-generation sequencing (NGS) in 2005 then made possible comprehensive genetic testing such as exome sequencing and genome sequencing. Meanwhile, great efforts have been put into the optimization of bioinformatic pipelines to make increasingly speedy and accurate variant analyses based on NGS data. These advances in sequencing technologies and analytical methods have revolutionized the diagnostic odyssey of suspected hereditary diseases. More recently, the genotype-phenotype relationship and polygenic risk scores (PRSs) generated from genome-wide association studies have expanded our horizon from rare genetic mutations to a genomic landscape implicated by the combined effect of both rare variants and polymorphisms. At the same time, clinicians and genetic counselors are facing huge challenges conferred by overwhelming genomic knowledge and long sheets of testing reports for comprehensive genomic sequencing. The path toward the "next-generation" clinical genetics and genomics may underlie semiautomatic pipelines assisted by artificial intelligence techniques.

Adapting systems and technology for an aging population has benefits for older people, the consumer market industry itself and all of society. To promote knowledge sharing on innovations for healthy ageing and digital inclusion of older people in the Western Pacific Region, a hybrid conference on "Digital inclusion of older people: harnessing digital technologies to promote healthy ageing in the Western Pacific Region" was held on 23 June 2021 by China Academy of Information and Communications Technology, a WHO Collaborating Centre for Digital Health. Barriers from demand side include: (1) unaffordability; (2) poor Information and Communication technology (ICT) knowledge and skills for navigation; and (3) low self-efficacy and motivation. Supply barriers include: (1) youth-centred design; (2) ageism; and (3) anti-facilitative environment including infrastructure and age-biased technology. Existing practices to overcome digital inclusion barriers were shared: (1) landmark initiatives related to the health and social welfare; (2) laws and policies to improve aged care services, strengthen social services, enrich spiritual and cultural life for older people; (3) ICT infrastructure and residential care facilities based on the philosophy of family care and supported by community care; (4) affordable digital application and adaptive feature design to better enable and motivate their desire to use digital technology; and (5) community activities such as trainings and tutorials to enhance digital capacity and literacy of older people. Main principles highlighted include market motivation, human-centered design, creating enabling environments, and multi-stakeholder collaborations to provide guidance to customize strategy under context of different regions and countries, instead of a one-size-fits-all solution, to avoid the risk of exacerbating inequalities experienced by older people, caused by accelerated ICT innovation, and advocate for more affordable products in the silver market.