 Radiomics has been shown to be a powerful tool for the diagnosis, prognosis, and prediction of treatment outcomes in cancer patients. It uses machine learning techniques such as feature engineering and statistical modeling to extract meaningful features from medical images and build predictive models. Despite its promise, there are still many technical challenges that need to be addressed before it can become widely used in clinical practice. These include issues related to feature engineering, data imbalance, multimodality fusion, stability, reproducibility, and interpretability. Solutions to these problems will enable more accurate and reliable predictions, leading to better patient care and improved health outcomes. This article was authored by Yuan Peng Zhong, Xin Yun Zhong, Yu Ting Cheng, and others.