Advances in Artificial Intelligence-Based Radiomultiomics for Assisting Cancer Diagnosis and Treatment

The advent of radiomics has significantly enhanced the role of medical imaging in the precision diagnosis and treatment of tumors. However, as the occurrence and development of tumours involve numerous factors at different levels, relying on a single histological data alone is not sufficient to advance the development of precision oncology. Therefore, radiological multi-omics is of great importance. A primary challenge in radio-multiomics is the extraction of meaningful features from complex multiomics datasets to develop predictive models. Artificial intelligence (AI) has demonstrated its potential to integrate high-throughput, high-dimensional, and large-scale multi-omics data, thereby enabling more comprehensive analyses. In this review, recent advances in integrated AI-based radio-multiomics analysis are summarized, along with an overview of the workflow involved in AI-based radio-multiomics analysis. Additionally, the application of AI-based radio-multiomics approaches is discussed in the context of early cancer screening, diagnosis, classification, prognosis, and treatment response prediction. Finally, the opportunities and challenges inherent to radio-multiomics analysis are highlighted.