Artificial intelligence has been emerging as an increasingly important aspect of our daily lives and is widely applied in medical science. One major application of artificial intelligence in medical science is medical imaging. As a major component of artificial intelligence, many machine learning models are applied in medical diagnosis and treatment with the advancement of technology and medical imaging facilities. The popularity of convolutional neural network in dental, oral and craniofacial imaging is heightening, as it has been continually applied to a broader spectrum of scientific studies. Our manuscript reviews the fundamental principles and rationales behind machine learning, and summarizes its research progress and its recent applications specifically in dental, oral and craniofacial imaging. It also reviews the problems that remain to be resolved and evaluates the prospect of the future development of this field of scientific study.

The radiation-induced bystander effect (RIBE) is a destructive reaction in nonirradiated cells and is one primary factor in determining the efficacy and success of radiation therapy in the field of cancer treatment. Previously reported studies have shown that the RIBE can be mediated by exosomes that carry miRNA components within. Exosomes, which are one type of cell-derived vesicle, exist in different biological conditions and serve as an important additional pathway for signal exchange between cells. In addition, exosome-derived miRNAs are confirmed to play an important role in RIBE, activating the bystander effect and genomic instability after radiotherapy. After investigating the field of RIBE, it is important to understand the mechanisms and consequences of biological effects as well as the role of exosomes and exosomal miRNAs therein, from different sources and under different circumstances, respectively. More discoveries could help to establish early interventions against RIBE while improving the efficacy of radiotherapy. Meanwhile, measures that would alleviate or even inhibit RIBE to some extent may exist in the near future.

Radiotherapy and chemotherapy are the most effective nonsurgical treatments for cancer treatment. They usually induce regulated cell death by increasing the level of reactive oxygen species (ROS) in tumour cells. However, as intracellular ROS concentration increases, many antioxidant pathways are concurrently upregulated by cancer cells to inhibit ROS production, ultimately leading to drug resistance. Understanding the mechanism of antioxidant stress in tumour cells provides a new research direction for overcoming therapeutic resistance. In this review, we address (1) how radiotherapy and chemotherapy kill tumour cells by increasing the level of ROS, (2) the mechanism by which ROS activate antioxidant pathways and the subsequent cellular mitigation of ROS in radiotherapy and chemotherapy treatments, and (3) the potential research direction for targeted treatment to overcome therapeutic resistance.

The extracellular matrix interacts with cancer cells and is a key factor in the development of cancer. Traditional two-dimensional models cannot mimic the natural in situ environment of cancer tissues, whereas three-dimensional (3D) models such as spherical culture, bioprinting, and microfluidic approaches can achieve in vitro reproduction of certain structures and components of the tumor microenvironment, including simulation of the hypoxic environment of tumor tissue. However, the lack of a perfusable vascular network is a limitation of most 3D models. Solid tumor growth and metastasis require angiogenesis, and tumor models with microvascular networks have been developed to better understand underlying mechanisms. Tumor-on-a-chip technology combines the advantages of microfluidics and 3D cell culture technology for the simulation of tumor tissue complexity and characteristics. In this review, we summarize progress in constructing tumor-on-a-chip models with efficiently perfused vascular networks. We also discuss the applications of tumor-on-a-chip technology to studying the tumor microenvironment and drug development. Finally, we describe the creation of several common tumor models based on this technology to provide a deeper understanding and new insights into the design of vascularized cancer models. We believe that the tumor-on-a-chip approach is an important development that will provide further contributions to the field.