AAI_2025_Capstone_Chronicles_Combined

Breast Tumor Classification Using Quantum Neural Networks

3

Breast Tumor Classification Using Quantum Neural Networks

In the United States hundreds of thousands of females are diagnosed with breast cancer

yearly, while tens of thousands die (US Centers for Disease Control and Prevention, 2025).

Obviously, to effectively treat breast cancer, the possible tumor must be identified as cancerous

(malignant) or non-cancerous (benign). This can be done by examining a variety of factors

about the patient and the tumor, making it a binary classification problem. Specifically, this

project will utilize the Breast Cancer Wisconsin Diagnostic dataset available on Kaggle (H,

2021). This dataset contains 31 features over 569 records.

Binary classification itself is not especially difficult or time consuming. However, this

project aims to leverage a quantum neural network (QNN)- and this is what makes the project

more complex and hopefully opens a window to the future on the potential blend of quantum

computing and artificial intelligence. Viggiano has described the blend as “The convergence of

these two technologies may have the same civilization-altering effects as the telegraph…”

(Viggiano, 2023). Therefore, this project can be used as a potential model for more elaborate

systems as quantum technology scales.

As quantum computers are still in their infancy, a simulated quantum computer is used.

Specifically, Google’s Cirq, which is an open-source framework. A key constraint is that this

framework only allows for about 20 quantum bits (qubits) to be simulated 1 (Google, 2025). This

constricted space presents implementation challenges. Specifically, scaling down training data

to fit in this low resolution has a substantial impact on model performance. Working within these

constraints was a key challenge for the project.

1 This is because there is an exponential slowdown as we increase the number of qubits in a simulation.

256