Revolutionizing brain tumor treatment: the rise of AI in neuro-oncology

Trending 1 week ago
ARTICLE AD BOX

A caller reappraisal article successful npj Precision Oncology summarizes nan existent authorities of knowledge astir nan domiciled of artificial intelligence (AI) successful nan diagnosis, treatment, and prognosis of encephalon tumors.

 metamorworks/Shutterstock.comStudy: Artificial intelligence successful neuro-oncology: advances and challenges successful encephalon tumor diagnosis, prognosis, and precision treatment. Image Credit: metamorworks/Shutterstock.com

Background

Brain tumors, though uncommon, airs a important wellness situation globally, pinch astir 250,000 caller cases each year. In nan United States alone, complete 96,000 encephalon tumor cases were reported successful 2022, pinch astir 26,600 of these being cancerous.

Glioblastoma is nan astir often diagnosed type of encephalon tumor and has a peculiarly mediocre prognosis, pinch only a 7% endurance complaint 5 years aft diagnosis.

This highlights nan urgent request for improved methods of diagnosing, treating, and forecasting nan progression of encephalon tumors.

Challenges successful managing encephalon tumors

Diffuse midline glioma (DMG) successful children and glioblastoma successful adults are among nan toughest encephalon tumors to dainty and are often considered incurable pinch existent aesculapian approaches.

Tailored treatments guidelines nan champion chance of providing a cure pinch nan slightest imaginable harm. However, nan situation is that accusation connected diagnosing and treating encephalon tumors is scattered and difficult to travel by.

Only a prime number of aesculapian centers person entree to nan latest curen techniques. Moreover, overmuch of nan disposable information connected these treatments is originated from conscionable 1 aliases a fewer institutions, limiting nan breadth of knowledge and accessibility for many.

Management approaches and diagnostic criteria based connected specified information are unfastened to a deficiency of demographic information and whitethorn not beryllium generalizable globally.

Socioeconomic inequity besides contributes to precocious diagnosis, therapeutic challenges, and reduced endurance by restricting entree to immoderate cardinal tests and reducing nan likelihood of operation therapies. This includes 06-Methylguanine-DNA-methyltransferase (MGMT) testing for glioblastoma.

The request for precise diagnosis, staging, and curen monitoring is difficult to meet successful galore cases.

Taking into relationship nan publication of tumor genotype to nan prognosis, constricted accessibility for imaging and biopsy, intratumor heterogeneity, and poorly reliable biomarkers to show nan advancement of therapy, location are important obstacles to nan optimal attraction of these patients.

The encephalon tumor paradigm

In astir cases, a suspected encephalon tumor is diagnosed, opening pinch a beingness introspection and neuroimaging. A biopsy follows this. If possible, nan tumor and different biomarkers are removed and subjected to histologic and molecular analysis.

The prime of therapy depends connected disposable and recommended attraction practices, objective tests that are presently going on, nan patient’s aesculapian status, and toxicity risks. Magnetic resonance imaging (MRI) is nan follow-up modality of choice, sometimes supplemented pinch cerebrospinal fluid (CSF) aliases humor tests.

“Decisions regarding encephalon tumor curen often impact multidisciplinary meetings betwixt neuro-oncologists, neurosurgeons, neuroradiologists, molecular pathologists, and neuropathologists, underscoring nan complexity of these decisions.”

The advantages of AI

AI includes instrumentality learning (ML) and heavy learning (DL) techniques, machine imagination (CV), and nan integration of these arsenic Computational Biology. ML excels astatine shape nickname and DL successful extracting elaborate features. CV improves ocular mentation of imaging information to supply aesculapian data.

Computational biology uses each these methods to parse biologic data, helping to understand tumor genetics and molecular biology.

This study intends to uncover AI-assisted tumor radiology, pathology, and genomics advancements. AI contributes synergistically to each these domains to amended their domiciled arsenic a mixed dataset successful encephalon tumor management.

AI whitethorn thief clinicians navigate tumor guidance decisions by improving MRI imaging accuracy and enhancing nan velocity astatine which results are available.

It offers accrued sensitivity to anomalies picked up connected imaging, elaborate image analysis, optimized workflows, broad information study from aggregate sources, and detecting patterns that could beryllium missed by nan quality observer.

AI algorithms thief localize tumors much efficiently, avoiding quality error. The nnU-Net algorithm excels astatine tumor segmentation, reducing radiation aliases surgical harm.

This enables AI to thief diagnose and people nan tumor, find nan prognosis, and scheme curen while mounting up a monitoring framework.

AI whitethorn go portion of caller objective trials, exploring nan feasibility of personalized therapy by leveraging its expertise to grip ample volumes of data.

AI uses various information types, including imaging information from MRI and computerized tomography (CT), radiomics, histopathologic data, genomics, molecular biomarkers from tumor cells, and objective data.

Neuroimaging often uses pre- and post-contrast T1-weighted, T2-weighted, fluid-attenuated inversion betterment (FLAIR), diffusion-weighted (DWI), and susceptibility-weighted imaging (SWI), arsenic good as, successful specialized centers, MR spectroscopy and perfusion imaging.

Molecular biomarkers see IDH mutations for astrocytomas and oligodendrogliomas, TERT promoter mutations for glioblastomas, EGFR amplification for glioblastomas, summation of chromosome 7 and nonaccomplishment of chromosome 10 for glioblastomas, and MGMT promoter methylation for glioblastomas.

Non-invasive circulating tumor DNA (ctDNA) study is simply a newer method for diagnosing specified tumors.

AI platforms

3D U-Net, DeepMedic, and V-Net are AI architectures that thief preprocess tumor images, making nan study much robust and precise. Methylome profiling is useful successful classifying encephalon tumors utilizing AI/MI and systems for illustration DeepGlioma. This uses stimulated Raman histology (SRH) to connection results connected GMB molecular test wrong 90 seconds.

Other systems to foretell IDH and different mutations based connected radiomics information from MRI perfusion scans aliases 18F-FET PET/CT scans are being explored, specified arsenic a heavy learning imaging signature (DLIS) and Terahertz spectroscopy.

‘Sturgeon’ is different DL method to categorize encephalon tumors intraoperatively utilizing nanopore-sequenced methylation array data. Its 40-minute turnaround time, pinch >70% accuracy, helps surgical decision-making.

Prognostic thief is being provided from imaging information to foretell wide endurance and progression-free survival, 2 cardinal objective and investigation metrics.

Combined pinch histology and molecular biology, exceptional predictive capacity has been demonstrated.

Integrated approaches

Multimodal information fusion approaches could thief execute a little invasive and much meticulous knowing of encephalon tumors utilizing aggregate information sources. This will yet thief tailor guidance to nan patient.

The situation is to widen and diversify nan information postulation scope to different populations and tumor types pinch standardized features to guarantee reproducibility and generalizability.

The take of AI should not worsen healthcare and societal inequities, emphasizing nan request to region biases, supply ineligible support, pass nan scope and benefits pinch transparency, specify responsibilities and support patients safe.

Conclusions

“AI has nan imaginable to empower patients by providing personalized accusation and enabling shared decision-making. However, nan equitable entree and affordability of AI-driven healthcare request to beryllium addressed to debar exacerbating existing disparities.”

Journal reference:

  • Khalighi, S., Reddy, K., Midya, A., et al. (2024) Artificial intelligence successful neuro-oncology: advances and challenges successful encephalon tumor diagnosis, prognosis, and precision treatment. npj Precision Oncology. doi: https://doi.org/10.1038/s41698-024-00575-0. https://www.nature.com/articles/s41698-024-00575-0