The use of artificial intelligence in detections, management and determining the prognosis of brain tumours; new insights

Background and aims: Brain tumors are a widespread and severe neurological problem. Theseabnormal growths can be managed if caught in the early stages. Despite all the progress in thefield, accurate segmentation, classification, and prognosis determination remains a challenge.Computational artificial intelligence (AI) models and deep-learning networks are now being assessedas new tools to help physicians overcome these challenges. In this review, we try to compilethe most recent researched methods that use artificial intelligence for better patient care in thefield of neurology and neurosurgery.Method: For this article, we searched Medline/PubMed, Web of Science, Scopus, and Embaseusing the keywords “artificial intelligence” and “brain tumor”. Each search was specified accordingto the database and special keywords (mesh, Emtree) were used.Results: AI can be used in different stages of patient care. For categorization, between computer-aided diagnoses systems, Convolutional neural networks are proven the most used magneticresonance imaging (MRI) based tool that categorizes tumors as either normal or pathologic.Support vector machines achieved ۹۸% accuracy in segmentations and analysis of brain tumordetection in MR imaging. One study suggested that using all MRI technics is the key to the bestmanagement of brain tumors. Another study suggested that AI-assisted brain tumor has betteroutcomes and less hospitalization time. Personalized chemotherapy using AI-collected molecularand genetic data from the tumor is still being studied in many institutions.Conclusion: According to the data we compiled, AI and deep-learning models still have a lot offlaws, one of which is the unexplained process of diagnosis which is the reason many physicianscan’t fully trust the results. Some studies are now working on AI models that can explain theprocess. However, despite all the shortcomings, AI and machine learning processes have shownprominent results in the early stages, but the need for future investigations remains to make thesetools usable in clinics.

کلیدواژه ها:

artificial intelligence ، medicine ، brain tumor ، cancer