Breast Cancer Detection with Quanvolutional Neural Networks

Area of Science:

• Quantum Computing
• Medical Imaging
• Machine Learning

Background:

• Classical machine learning methods struggle with complex patterns in medical data.
• Quantum machine learning (QML) offers novel approaches for enhanced data analysis.
• Accurate cancer detection from medical images is crucial for effective treatment.

Purpose of the Study:

• To evaluate the efficacy of quantum convolutional layers in classifying ultrasound breast images for cancer detection.
• To compare the performance of a Quantum Convolutional Neural Network (QCNN) against a classical Convolutional Neural Network (CNN).

Main Methods:

• Developed a QCNN utilizing two quantum circuits as convolutional layers.
• Employed angle embedding for encoding classical data into quantum states.
• Designed a 9-qubit quantum circuit incorporating an SU(4) gate for robust entanglement.

Main Results:

• The QCNN achieved a peak training accuracy of 76.66% and a validation accuracy of 87.17%.
• The classical CNN achieved a training accuracy of 77.52% and a validation accuracy of 83.33%.
• The QCNN demonstrated superior performance on the validation set, particularly with limited data.

Conclusions:

• Quantum circuits can function as effective convolutional layers for image classification tasks.
• QML presents a promising avenue for improving cancer diagnostic imaging.
• The QCNN approach shows potential for feature extraction, especially in scenarios with small datasets.