The emergence of Artificial Intelligence (AI) has had a mass transformative effect in shifting healthcare to a positive enlightenment, particularly in neuro-oncology, where it empowered healthcare professionals to diagnose patient cases more efficiently, strengthen prognosis, and enable real-time monitoring and treatment planning (Khalighi et al., 2024). This helped improve the overall clinical outcomes of cancer patients.

To understand the impact of AI in the diagnosis and treatment of a patient harbouring a brain tumour. A brief overview of what commonly entails according to the guidelines necessitates the importance of AI. At first, physical examination and radiological imaging occur. A histopathological analysis is applied after obtaining a biopsy sample or tumour resection to identify the type of tumour. There are more than 100 types of mutations, and brain cancers manifest a heterogeneous cellular origin to find the precise tumour classification, disease diagnosis, and staging. Such surgical procedures are increasing with the complex nature of the brain and limited accessibility for successive accurate imaging (Khalighi et al., 2024).

Evaluation of other bodily fluids, namely, serum and cerebrospinal fluid (CSF), frequently occurs to identify biomarkers associated with brain cancer (Xiao et al., 2020). The consequence of these assessments will facilitate the decision on the optimal therapy discussed in multidisciplinary meetings. The effectiveness of the treatment is monitored through Magnetic Resonance Imaging (MRI) and biomarker evaluation using blood and CSF (Khalighi et al., 2024). This can be halted by the genotype patterns and signals from neighbouring neural tissues (Aldape et al., 2019).

Deep Learning (DL) is a form of machine learning that comprises multilayered artificial neural networks that depict the structure of the human brain and is applied for routine medical imaging (IBM, 2025). The DL models are moderated by supervised learning from training data and/or data that is labelled, where they perform classification and regression. The larger the scale of the neural networks, the more training data is required to optimize the conditions. DL models can learn from training data and can detect any potential biases.

Most healthcare AI applications contain models that train on populations that are less diverse and present the following sociodemographic factors: age, socioeconomic status, ethnicity, and other health conditions. As a consequence, AI methods that do not generalize are applied in other countries with high diversity (Chen et al., 2023).

One of the most promising effects of AI is in multimodal medical imaging (Monsour et al., 2022). MRI assisted in optimizing workflows, providing accurate quantitative measurements of tumour samples by using automated segmentations, for instance, nnU-Net and segment anything models (SAMs) (Isensee, 2021; Ma, 2024). This surpasses previous challenges when analyzing tumour radiological images via the RECIST (response evaluation criteria in solid tumours) guidelines (Nishino et al., 2010). It requires a substantial amount of time to manually measure the tumour size and compare it with follow-up scans (Villaruz and Socinski, 2013). Such segmentation tools can monitor the burden of the disease without manual annotations by a radiologist.

AI in medical imaging offers advantages such as analyzing large datasets efficiently and detecting abnormalities with precision (Monsour et al., 2022). AI conducts detailed imaging analysis, mapping tumour boundaries and identifying molecular subtypes. This helps in grading, monitoring, and evaluating treatment response both before and after therapy. During surgery, AI assists with assessing surgical margins, providing real-time diagnoses, and enhancing precision.

According to Ligero et al. (2025), AI biomarkers support the diagnosis and treatment of brain cancers. They do this by enabling efficient histological and molecular examination of samples, distinguishing tumours from necrosis, and predicting biomarkers from histopathological imaging. These features improve time efficiency and reduce workload, making AI biomarkers advantageous over Next Generation Sequencing (NGS).

For instance, Microsatellite Instability (MSI) is a form of genetic hypermutability caused by impaired DNA mismatch repair. It is found in some cancers, predominantly endometrial, colorectal, and brain cancer, and it can increase sensitization to immunotherapy. There are several forms of MSI: MSI-H refers to high levels of instability and can influence the prognosis for treatment. MSI-L indicates low instability, where alternative treatments would be considered. MSI-I suggest that no instability may affect treatment options. The MSI status can be predicted from Whole-slide imaging (WSI) histopathological slides that are stained with Hematoxylin and Eosin (H and E). WSI is a high-resolution digital scans that provide a detailed analysis of tissue samples at multiple magnifications on the microscope. If MSI is present, the area under the curve (AUC) is estimated between 0.71 and 0.97 (Kather et al., 2019; Saillard et al., 2023).

Other examples include DL’s ability to predict other oncogenic driver mutations found in brain cancers, such as the guardian of the genome (TP53), and may include other DNA or RNA expression signatures. This helps to attenuate predictions for personalized treatment and detect cancer type early via the gender-specific characteristics (Khalighi et al., 2024).

On the other hand, the genomic application of NGS is a standardized method facilitating decision-making, but it has several considerable complications. At first, how developed the country is depends on whether hospital services offer such services as it is costly (Bayle et al., 2023). Even so, it not only depends on the geographical status but also on the socioeconomic status of the patient themselves, which may reflect on the causal precursors for inequity. In addition, the turnaround time from receiving the sample to obtaining the results is approximately 15 days (Sheffield et al., 2023). Thus, if a patient requires urgent results, the duration can negatively affect his or her results (Hanna et al., 2020).

Moreover, it is important to state that there is a distinction between AI-based cancer biomarker prediction and AI confirmatory testing. The former depends on the morphological structures of the tumour microenvironment that may not appear in confirmatory tests. Therefore, additional research is essential to determine the effectiveness of the targeted treatment and patient response (Ligero et al., 2025).

Furthermore, there are alternative AI models that integrate text and images to make it more interactive and interpretable for information and AI-based decisions. For example, PathChat is a vision-language model with a built-in multimodal integrative system. It can interact with the healthcare profession on the image content (Lu et al., 2024). Large language models (LLMs) have facilitated in improving accuracy of predictive biomarkers and electronic health records.

In addition, in clinical trials, AI involvement can help save time and resources to find eligible subjects, especially in the stages of image rescreening and care management.

On the contrary, several disadvantages of AI have come to my attention. The approval of an AI-based method or source requires regulatory certification that has a lengthy timeline. Another limitation is attributed to the ethical, legal, and social concerns. AI commonly helps to stratify patients by data patterns: clinical history, molecular biomarker, pathology, and radiology. Thus, patient privacy, informed consent, and transparency of algorithms are paramount to maintain patients’ trust (Khalighi et al., 2024).

Another limitation of AI-based technology is that not all centres have multidisciplinary tumour boards and are referred to specialized centres, which delays treatment. Overall, it is notable that AI has multiple applications in healthcare and other disciplines, but additional research is needed to solidify its stance in the diagnosis and treatment of cancers.

References

Aldape, K., Brindle, K.M., Chesler, L., Chopra, R., Gajjar, A., Gilbert, M.R., Gottardo, N., Gutmann, D.H., Hargrave, D., Holland, E.C., Jones, D.T.W., Joyce, J.A., Kearns, P., Kieran, M.W., Mellinghoff, I.K., Merchant, M., Pfister, S.M., Pollard, S.M., Ramaswamy, V. and Rich, J.N. (2019). Challenges to curing primary brain tumours. Nature Reviews Clinical Oncology, [online] 16(8), pp.509–520. doi:https://doi.org/10.1038/s41571-019-0177-5.

‌Bayle, A., Bonastre, J., Chaltiel, D., Nicola Jane Latino, Rouleau, É., Peters, S., Galotti, M., Gracemarie Bricalli, Besse, B. and Giuliani, R. (2023). ESMO study on the availability and accessibility of biomolecular technologies in oncology in Europe. Annals of Oncology, 34(10), pp.934–945. doi:https://doi.org/10.1016/j.annonc.2023.06.011.

Chen, R.J., Wang, J.J., Williamson, D., Chen, T.Y., Lipkova, J., Lu, M.Y., Sahai, S. and Mahmood, F. (2023). Algorithmic fairness in artificial intelligence for medicine and healthcare. Nature biomedical engineering, 7(6), pp.719–742. doi:https://doi.org/10.1038/s41551-023-01056-8.

Hanna, T.P., King, W.D., Thibodeau, S., Jalink, M., Paulin, G.A., Harvey-Jones, E., O’Sullivan, D.E., Booth, C.M., Sullivan, R. and Aggarwal, A. (2020). Mortality due to cancer treatment delay: systematic review and meta-analysis. BMJ, [online] 371. doi:https://doi.org/10.1136/bmj.m4087.

‌IBM (2025) What is deep learning? Available at: https://www.ibm.com/think/topics/deep-learning (Accessed: 11^th December 2025)

Isensee, F., Jaeger, P.F., Kohl, S.A.A., Petersen, J. and Maier-Hein, K.H. (2020). nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods, [online] 18(2), pp.203–211. doi:https://doi.org/10.1038/s41592-020-01008-z.

Kather, J.N., Pearson, A.T., Halama, N., Jäger, D., Krause, J., Loosen, S.H., Marx, A., Boor, P., Tacke, F., Neumann, U.P., Grabsch, H.I., Yoshikawa, T., Brenner, H., Chang-Claude, J., Hoffmeister, M., Trautwein, C. and Luedde, T. (2019). Deep learning can predict microsatellite instability directly from histology in gastrointestinal cancer. Nature Medicine, 25(7), pp.1054–1056. doi:https://doi.org/10.1038/s41591-019-0462-y.

‌Khalighi, S., Reddy, K., Midya, A., Pandav, K.B., Madabhushi, A. and Abedalthagafi, M. (2024). Artificial intelligence in neuro-oncology: advances and challenges in brain tumor diagnosis, prognosis, and precision treatment. npj Precision Oncology, [online] 8(1), pp.1–12. doi:https://doi.org/10.1038/s41698-024-00575-0.

Ligero, M., El Nahhas, O.S.M., Aldea, M. and Kather, J.N. (2025). Artificial intelligence-based biomarkers for treatment decisions in oncology. Trends in Cancer. [online] doi:https://doi.org/10.1016/j.trecan.2024.12.001.

Lu, M.Y., Chen, B., Williamson, D.F.K., Chen, R.J., Zhao, M., Chow, A.K., Ikemura, K., Kim, A., Pouli, D., Patel, A., Soliman, A., Chen, C., Ding, T., Wang, J.J., Gerber, G., Liang, I., Le, L.P., Parwani, A.V., Weishaupt, L.L. and Mahmood, F. (2024). A Multimodal Generative AI Copilot for Human Pathology. Nature, [online] pp.1–3. doi:https://doi.org/10.1038/s41586-024-07618-3.

Ma, J., He, Y., Li, F., Han, L., You, C. and Wang, B. (2024). Segment anything in medical images. Nature Communications, [online] 15(1). doi:https://doi.org/10.1038/s41467-024-44824-z.

Monsour, R., Dutta, M., Mohamed, A. Z., Borkowski, A. & Viswanadhan, N.A. (2022) Neuroimaging in the era of artificial intelligence: current applications. Federal Practitioner, 39, S14–S20.

Nishino, M., Jagannathan, J.P., Ramaiya, N.H., and Van den Abbeele, A.D. (2010). Revised RECIST Guideline Version 1.1: What Oncologists Want to Know and What Radiologists Need to Know. American Journal of Roentgenology, 195(2), pp.281–289. doi:https://doi.org/10.2214/ajr.09.4110.

‌Saillard, C., Dubois, R., Tchita, O., Loiseau, N., Garcia, T., Adriansen, A., Carpentier, S., Reyre, J., Enea, D., von Loga, K., Kamoun, A., Rossat, S., Wiscart, C., Sefta, M., Auffret, M., Guillou, L., Fouillet, A., Kather, J.N. and Svrcek, M. (2023). Validation of MSIntuit as an AI-based pre-screening tool for MSI detection from colorectal cancer histology slides. Nature Communications, [online] 14(1), p.6695. doi:https://doi.org/10.1038/s41467-023-42453-6.

Sheffield, B.S., Beharry, A., Diep, J., Perdrizet, K., Iafolla, M., Raskin, W., Shaan Dudani, Mary Anne Brett, Blerta Starova, Olsen, B. and Cheema, P. (2022). Point of Care Molecular Testing: Community-Based Rapid Next-Generation Sequencing to Support Cancer Care. 29(3), pp.1326–1334. doi:https://doi.org/10.3390/curroncol29030113.

Villaruz, L.C. and Socinski, M.A. (2013). The Clinical Viewpoint: Definitions, Limitations of RECIST, Practical Considerations of Measurement. Clinical Cancer Research, [online] 19(10), pp.2629–2636. doi:https://doi.org/10.1158/1078-0432.ccr-12-2935.

Xiao, F., Lv, S., Zong, Z, Wu, L, Tang, X, Kuang, W, Zhang, P, Li, X, Fu, J, Xiao, M, Wu, M, Wu, L, Zhu, X, Huang, K. and Guo, H. (2020) Cerebrospinal fluid biomarkers for brain tumor detection: clinical roles and current progress. American Journal of Translational Research. 12(4), pp.1379-1396.