ABSTRACT
It is noteworthy that only 5% of more than 7000 described rare diseases are treated. In the era of big data, there is ever-increasing data for understanding biomedicine. The need for efficient and rapid data collection, analyses and characterization methods is pressing. Rare diseases can particularly benefit from artificial intelligence (AI) application. AI, with an emphasis on machine learning, creates a path for such efforts and is being applied to diagnosis and treatment. AI has demonstrated its potential to learn and analyze data from different sources with results in prediction。Presently, there are AI-driven technologies applied for rare diseases and this review aims to summarize these advances. Moreover, this review scrutinizes the limitation and identifies the pitfalls of AI applications in the diagnosis and treatment of rare diseases.
ABSTRACT
The tasks of artificial intelligence (AI) in tumor histopathology include image segmentation and classification. Currently, the specific contents including lymph node metastasis, pathological classification, grade and prognostic evaluation of malignant diseases, such as breast cancer, lung cancer and prostate cancer, have been studied by AI. Evaluation of sentinel lymph node metastasis of breast cancer has been the most mature application of AI technology, whose level can be analogous to the excellent pathologists. In the future, with the close cooperation of pathologists and engineers, the research of AI will be focus on improving the technology of simple and repetitive clinical diagnosis and differential diagnosis, such as the diagnosis of sentinel lymph node metastasis of breast cancer from biopsy frozen section and the judgment of incisal margin. Ultimately, AI will help us to precisely diagnose the tumor.