Hybrid classical and quantum computing for enhanced glioma tumor classification using TCGA data | JoVE Visualize

Area of Science:

Neuro-oncology
Quantum Computing
Artificial Intelligence

Background:

Gliomas, the most common primary brain tumors, pose diagnostic challenges due to significant differences in survival and treatment response between low-grade (LGGs) and high-grade (HGGs) subtypes.
Accurate classification of gliomas is critical for determining appropriate treatment strategies and predicting patient prognosis.
Classical AI methods face limitations with large, noisy medical datasets and complex data structures, hindering optimal glioma classification.

Purpose of the Study:

To develop and evaluate a novel hybrid classical and quantum computing model for distinguishing between LGGs and HGGs.
To leverage quantum computing's potential for enhanced data processing and analysis in medical diagnostics.
To identify key molecular and clinical features that differentiate LGG from HGG using a hybrid approach.

Main Methods:

A hybrid model combining classical ensemble feature selection with variational quantum classifiers (VQCs) was developed using The Cancer Genome Atlas (TCGA) data.
An ensemble method identified informative molecular and clinical features from the TCGA dataset.
Six VQC models with different hyperparameters were trained and evaluated for their ability to classify LGGs from HGGs, with the AQCD optimization method used.

Main Results:

The VQC-1 model, utilizing specific quantum gates and the AQCD optimization, achieved the highest classification accuracy of 0.74.
The most important features identified by VQC-1 for distinguishing LGGs from HGGs were IDH1, age at diagnosis, PTEN, EGFR, and ATRX.
In five-fold cross-validation, VQC-1 performance was comparable to XGBoost and GBM, outperforming KNN, SVC, DTC, and RFC.

Conclusions:

Hybrid classical and quantum computing models offer a promising approach for complex medical classification tasks like glioma grading.
The developed VQC-1 model demonstrates the potential of quantum AI in improving the accuracy and efficiency of brain tumor classification.
This study highlights the significance of integrating quantum computational methods with classical machine learning for advancing neuro-oncology diagnostics.