Medizinische Universität Graz - Research portal
Selected Publication:
Claeys, S.
An innovative approach to deep Learning used as an augmenting tool in the field of oncology
Humanmedizin; [ Diplomarbeit ] Medizinische Universität Graz; 2023. pp. 52
[OPEN ACCESS]
FullText
- Authors Med Uni Graz:
- Advisor:
- Enko Dietmar
- Meinitzer Andreas
- Altmetrics:
- Abstract:
- Background: Digitalization has revolutionized our modern world, including the field of medicine. Unfortunately, the exponential growth of medical data forms an increasing problem, making it almost impossible to manually analyse and extract meaningful information. Deep learning (DL) is capable of detecting features and seeking recurring patterns in high-dimensional datasets. The field of oncology is, therefore, the perfect environment for the integration of DL to detect and characterize cancer cells and uncover molecular mechanisms. Aim and objective: The aim of this thesis consists of comprehensive research and a summary of novel insights into DL as an augmenting tool in oncology. Hence, creating an overview of developments in the fields of radiomics, computational pathology and multi-omics. Methodology: A comprehensive literature search for all published articles regarding the objective of this thesis was performed using PubMed and Google Scholar as search engines. Complimentary information about DL and tumorigenesis was mostly obtained from textbooks and papers. Results and discussion: This novel technology marks suspicious lesions, predicts malignancies and monitors aberrant tissue. Furthermore, DL is outstanding at discovering driver oncogenic pathways and identifying cancer vulnerabilities, which could lead to the development of targeted therapy. The strength of this revolutionizing technology to quantify interconnected mechanisms of tumorigenesis and uncover patient-specific characteristics boosts the transition towards precision medicine. The patient-centric approach allows the tailoring of medical treatment to individual variabilities and enables the maximization of treatment response and clinical outcomes while minimizing side effects. Conclusion: The data-driven characteristics of DL appear to be an excellent tool to augment physicians in different domains of their clinical routine. The automation of repetitive and administrative tasks will bring back the emphasis on healthcare again and will strengthen the human connection between physicians and patients.