Cancer Genetics and Epigenetics 2024, Vol.12, No.4, 210-222 http://medscipublisher.com/index.php/cge 212 In the realm of imaging, AI models have demonstrated the ability to predict CRLM development and detect metastases with high accuracy, potentially outperforming expert radiologists (Rompianesi et al., 2022). Furthermore, AI can integrate multi-modal data, including clinical, imaging, and molecular data, to provide a comprehensive assessment of disease risk and prognosis (Qiu et al., 2022; Mansur et al., 2023). This holistic approach can lead to more precise and personalized treatment plans, ultimately improving patient outcomes. Despite the initial enthusiasm, the integration of AI in clinical practice faces several challenges, including the need for large, high-quality datasets, regulatory considerations, and ethical concerns (Qiu et al., 2022; Rompianesi et al., 2022). However, ongoing research and advancements in AI technology continue to pave the way for its broader application in CRC management. Future studies focusing on the validation and standardization of AI tools are essential to fully realize their potential in improving CRC detection, prediction, and treatment (Thakur et al., 2020; Qiu et al., 2022; Rompianesi et al., 2022). 3 Multi-Modal Data in Cancer Research 3.1 Definition and types of multi-modal data Multi-modal data in cancer research refers to the integration of various types of biological and clinical data to provide a comprehensive understanding of cancer. This approach leverages the strengths of different data types to improve the accuracy of cancer diagnosis, prognosis, and treatment. The primary types of multi-modal data include: 3.1.1 Genomic data Genomic data encompasses the complete set of DNA within an organism, including all of its genes. This data type is crucial for identifying genetic mutations and variations that may contribute to cancer development. For instance, studies have shown that integrating genomic data with other data types can enhance the prediction of cancer outcomes (Shao et al., 2020; Carrillo-Perez et al., 2022). 3.1.2 Transcriptomic data Transcriptomic data involves the study of RNA transcripts produced by the genome under specific circumstances. This data type helps in understanding gene expression patterns and their alterations in cancerous cells. The use of RNA-Seq data has been shown to improve cancer prediction models significantly (Table 1) (Xiao et al., 2018; Carrillo-Perez et al., 2022). Table 1 Accuracy Analysis of Multimodal Fusion Model for Different Data Types (Adopted from Carrillo-Perez et al., 2022) WSI RNA miRNA CNV metDNA Correct 1232 913 834 1636 821 Misclassified 159 67 70 220 62 Fusion Correct 1328 929 857 1796 838 Misclassified 63 51 47 60 45 Absolute difference in misclassified error rate (#samples(%)) 96(6.5%) 16(1.6%) 23 (2.6%) 160(8.6%) 17(2%) Table caption: Correct and misclassified samples over the whole dataset for each data type and the fusion model using all modalities. RNA, CNV, and metDNA stand for RNA-Seq, copy number variation, and DNA methylation, respectively (Adopted from Carrillo-Perez et al., 2022) 3.1.3 Epigenetic data Epigenetic data includes information about chemical modifications to DNA and histone proteins that regulate gene expression without altering the DNA sequence. DNA methylation is a common epigenetic modification studied in cancer research. Integrating epigenetic data with other modalities can provide insights into cancer progression and treatment responses (Carrillo-Perez et al., 2022). 3.1.4 Proteomic data Proteomic data involves the large-scale study of proteins, including their structures and functions. Proteins are the primary effectors of cellular functions, and their expression levels can be indicative of cancerous changes.
RkJQdWJsaXNoZXIy MjQ4ODYzNQ==