Exploring the Potential of Quantum Machine Learning in Cancer Research

Lalasa Mukku; Jyothi Thomas

Lalasa Mukku^* and Jyothi Thomas

¹ Department of Data Science, Christ (Deemed to be University), Kengeri, Bangalore, India, Email: mlalasa2020@gmail.com

^*Correspondence: Lalasa Mukku, Department of Data Science, Christ (Deemed to be University), Kengeri, Bangalore, India, Email: mlalasa2020@gmail.com

Received: 09-Jan-2023, Manuscript No. AMHSR-23-86401; Editor assigned: 11-Jan-2023, Pre QC No. AMHSR-23-86401 (PQ); Reviewed: 25-Jan-2023 QC No. AMHSR-23-86401; Revised: 09-Mar-2023, Manuscript No. AMHSR-23-86401 (R); Published: 16-Mar-2023

Citation: Mukku L, et al. Exploring the Potential of Quantum Machine Learning in Cancer Research. Ann Med Health Sci Res. 2023;13:690-694.

This open-access article is distributed under the terms of the Creative Commons Attribution Non-Commercial License (CC BY-NC) (http://creativecommons.org/licenses/by-nc/4.0/), which permits reuse, distribution and reproduction of the article, provided that the original work is properly cited and the reuse is restricted to noncommercial purposes. For commercial reuse, contact reprints@pulsus.com

Abstract

Quantum machine learning is a rapidly emerging field that combines the principles of quantum computing with the techniques of machine learning. It uses quantum effects such as entanglement to improve the accuracy of traditional machine learning models. Quantum computing it is based on the ability of particles to exist in multiple states at the same time, allowing for the parallel processing of vast amounts of data. Quantum machine learning can be used to perform pattern recognition tasks, such as image classi ication and image segmentation, which are essential for image processing. By leveraging the power of quantum computing, quantum machine learning algorithms can process large amounts of data much more efficiently than traditional machine learning algorithms, resulting in faster and more accurate results. Additionally, quantum machine learning algorithms can improve the accuracy of image processing tasks by recognizing more subtle patterns in the data. This could potentially solve the medical image segmentation and classi ication to identify and predict cancers. This paper examines the intersection of quantum computing and machine learning models for medical image segmentation and the potential implications for advancing cancer research. The paper begins by describing the basics of quantum computing and machine learning algorithms, then goes on to explore how they can be combined to create powerful new systems for efficient cancer research. Subsequently the study presents image segmentation algorithms K-means and Q-means and further discusses the need for further research on the combination of quantum computing and machine learning, as well as potential ethical implications.

Keywords

Quantum machine; X-rays; CT scans; MRI scans; Quantum computing; Cancer research

Introduction

Traditional machine learning and deep learning algorithms have been used to segment and classify bio medical images to identify malignancies [1]. The development of Computer Aided Diagnostic (CAD) technology has advanced the cancer diagnosis, prognosis, and personalized treatment [2]. Several cancer diagnoses suffers from delay of diagnosis and holds the limitation of manual image analysis that often requires highly skilled personnel, are a scarce resource in developing countries [3,4]. The reason for this predicament is that the malignancy is dormant and asymptomatic for years during which time the cells in cervix turn malignant [5]. Early diagnosis enables timely treatment and reduces the burden of cancer. Artificial intelligence has contributed significantly to cancer diagnosis and prognosis by aiding the clinical decisions made based on image analysis. Segmentation and classification are two stage solutions for intelligent system medical image diagnosis [6].

Machine Learning (ML) is a branch of AI that deals with large amounts of data to identify patterns and predict events that may potentially occur in future like cancer in this context [7,8]. It has proven to be a reliable source for oncological diagnosis and prognosis [9]. ML can be used to help diagnose cancer by analyzing large amounts of data to identify patterns and trends that may indicate the presence of cancer. This data can include medical images, patient records, and other data sources.

Machine learning algorithms are an efficient choice identifies features in the data that are associated with cancer, and can be used to make predictions about the likelihood of a patient having cancer. Machine learning can also be used to help identify the most effective treatments for a particular type of cancer. Deep learning has been used to develop Computer-Aided Diagnosis (CAD) systems for diagnosing cancer. These systems use deep learning algorithms to analyse medical images such as X-rays, CT scans, and MRI scans (2D, 3D, 4D, 4.5D image) to detect and classify tumours. Deep learning algorithms can also be used to analyse patient data such as lab results, medical history, and lifestyle factors to identify patterns that may indicate the presence of cancer. Deep learning can also be used to develop decision support systems that can help doctors.

Lung cancer, liver cancer, brain tumours, breast cancer, oral cancer, pancreatic cancer, colon cancer, cervical cancer etc., have benefited from deep learning solutions. Singh, et al., has done a chronological review of machine and deep learning models used for cancer diagnosis and summarize the drawbacks of the current methods [10-15]. Among them, insufficient data, computational cost, efficient pipeline of methods are few of the noted shortcomings.

Quantum computing is a type of computing that uses quantum bits, or qubits, to store and process data. Unlike classical computers, which use bits that can either be a 0 or a 1, qubits can be in a state of 0, 1, or a superposition of both at the same time. This allows quantum computers to perform certain operations much faster than classical computers. One of the key principles of quantum mechanics is that particles can be in multiple states at the same time, and this is what makes quantum computing possible. Quantum computers use this principle to perform multiple calculations simultaneously, which can speed up certain types of computation. One of the main challenges of quantum computing is that qubits are highly sensitive to their environment and can be easily disturbed. This means that quantum computers must be carefully isolated from outside interference in order to function properly.

Quantum computing is an amalgamation of computer science techniques and quantum mechanics. Quantum computing was first proposed in 1980’s as a means to advance the computational modelling using quantum principles. In this study we adapt Quantum Machine Learning (QML) Technique Q-means. QML integrates quantum principles in machine learning models, in this case a K-means model. Potential increase in computational speed, needing smaller sized dataset; increased accuracy etc., is among a few advantages of using quantum learning over machine learning. In this study we propose the use of a quantum based Q-means learning algorithm to segment the bio medical images.

Quantum machine learning has the potential to revolutionize the way cancer is diagnosed. By leveraging the power of quantum computing, machine learning algorithms can be used to analyse large datasets of medical images and patient records to identify patterns and correlations that may indicate the presence of cancer [16]. This could lead to more accurate and earlier diagnosis of cancer, as well as more personalized treatments. Additionally, quantum machine learning could be used to develop new drugs and treatments for cancer, as well as to identify biomarkers that could be crucial to predict the occurrence of malignancy [17,18].

In this paper, we present an algorithm for the task of bio medical image segmentation by machine learning and quantum learning models. Subsequently their performance is analyzed and compared. Although it has been shown that the traditional ML models obtain good performance on this task, often the superior, state of the art techniques suffer from some important drawbacks, including a very long training time and limitations on the size of dataset and computational time. Recent extensions to quantum neural networks models have been developed in an attempt to address these drawbacks.

Context

Quantum technology is playing an increasingly important role in cancer research. This technology is allowing scientists to better understand the complexities of cancer and its underlying mechanisms, as well as develop new treatments and therapies. Quantum computing is providing researchers with powerful tools for studying the intricate genetic networks that underlie cancer. By leveraging quantum computing’s capacity for processing large amounts of data, researchers can identify the specific gene mutations that are associated with different types of cancer and develop more effective treatments. Quantum imaging is also being used to improve the accuracy of cancer diagnosis and staging. By providing detailed images of the cellular structure of tumours, these technologies can help doctors better understand the stage and progression of the disease.

Finally, quantum enabled sensors are being developed to detect the presence of cancer cells in the body. These sensors can detect even small trace amounts of cellular material, allowing doctors to detect cancers earlier and reduce the risk of metastasis. Understanding, diagnosing, and treating cancer. Overall, quantum technology is providing researchers with powerful tools for understanding, diagnosing, predicting, and treating cancer. This technology is helping to improve the accuracy of diagnosis, and is leading to the development of new treatments and therapies that are more effective and less harmful [19].

Materials and Methods

A series of quantum based methods are available to promote efficient working of artificial intelligence based machine and deep leaning models used in tasks of classifying, clustering, predicting, and diagnosing various types of cancers.

• Quantum machine learning: Quantum machine learning is a field of research that focuses on applying quantum computing principles and techniques to machine learning algorithms. This involves using quantum computers to develop new algorithms for solving complex classification, prediction problems and for learning from data. Quantum machine learning algorithms can be used to create more efficient, accurate, and robust machine learning models like quantum SVM for cancer detection.
• Quantum simulation: Quantum computers can simulate quantum systems [20], allowing for the study of molecular structures and processes that are difficult to study with classical computers. This can be used to develop improved treatments and drugs for cancer.
• Quantum computing: Quantum computing is a type of computing that uses quantum mechanical phenomena, such as superposition and entanglement, to perform operations on data. It is different from traditional computing, which works using bits (1s and 0s). In quantum computing, data is stored and processed using quantum bits, or qubits. Qubits can be in multiple states at once, allowing for data to be processed much faster than traditional computing methods.
• Quantum optimization: Quantum optimization algorithms can be used to optimize cancer treatment protocols, such as radiation therapy and chemotherapy.
• Quantum imaging: Quantum imaging is a field of research that focuses on the use of quantum entanglement to improve the resolution, sensitivity and accuracy of imaging techniques. It relies on the principles of quantum mechanics and quantum optics to create images with greater detail and accuracy than traditional imaging methods. The goal of quantum imaging is to create images that are more precise and contain more information than what is currently possible.
• Quantum deep learning: Quantum deep learning is a field of research that combines quantum computing with deep learning to create Artificial Intelligence (AI) algorithms with enhanced capabilities. It uses quantum computing principles to create algorithms that can solve complex problems more quickly and accurately than those created with classical computing. Quantum deep learning is still in its early stages, but it has the potential to revolutionize AI and machine learning.
• Quantum clustering: Quantum clustering is a quantum deep learning technique in which quantum algorithms are used to cluster together data points with similar characteristics. Clustering is an unsupervised learning technique, meaning that it does not require labeled data. Instead, it determines the relationships between data points, allowing for the identification of patterns and trends that would otherwise be difficult to detect.
• Quantum segmentation: Quantum segmentation is a method of segmenting or dividing a large or complex data set into smaller, more manageable and more easily analysed subsets. It uses a combination of advanced algorithms and machine learning to identify patterns in the data, enabling more accurate segmentation and better understanding of the data. This method of segmentation is especially useful in big data analytics, where the data sets are often too large and complex for traditional methods.

Role of quantum computing in cancer research

Bisarya, et al., used the amalgamation of supervised learning in quantum framework. They trained ten qubits system to learn the patterns in Wisconsin breast cancer dataset. Iyer, et al., proposed hybrid quantum mechanical system with 2 qubits operation to classify cancerous cells in skin cancer. Dabba, et al., experimented with Quantum Moth Flame Optimization Algorithm (QMFOA) to understand the gene expression data to predict cancerous patterns. More works were aimed at exploring the quantum solution. A chronological review on application of quantum dots in cancer research is given in Fang, M.

Quantum dots in quantum computing are tiny particles that can contain a single electron or a single quantum bit (qubit). These particles can be used to create systems that are much smaller and more efficient than traditional computer circuits. Quantum dots offer greater precision and control over quantum systems, and thus promise to revolutionize computing and data processing. Quantum dots have been a constant source of experimental research toward cancer biomarker identification. Tabish, et al., employed graphene quantum dots to distinguish between malignant and normal cells. Smith, et al., studied genomic data to identify nanoparticles using quantum dots. Zajac, et al., experimented with protein microarrays through quantum dots.

Not only the cancer research has been benefitted by quantum principles but it is also used in other arenas successfully. For instance, Gupta, et al., approached the post COVID health care monitoring through quantum computations. Mental health department benefitted from quantum in several angles, prediction being on the top. Swarna, et al., combined machine learning and quantum computing to predict Parkinson’s disease in a dementia patient. The quantum K means was used by Kavita, et al., to predict heart diseases.

Results

Segmentation

A K-means quantum algorithm is a type of clustering algorithm used for segmentation. It is an iterative algorithm that uses a cost function to determine the optimal number of clusters and then assigns each data point to the closest cluster. The algorithm works by first randomly assigning each data point to a cluster and then iteratively refining the clusters by minimizing the cost function.

K means algorithm

Initialize K cluster centres c₁, c₂,..., c_K randomly.

Repeat until convergence

• For each data point x_i:: • Calculate the distance between xi and the K cluster centres c₁, c₂,..., c_K.; • Assign x_i to the closest cluster centre.
• Update the cluster centres c₁, c₂,..., c_K to the mean of all data points assigned to it.

Return the final cluster centres and the data points assigned to each cluster.

Q-means

Q-means is a quantum inspired algorithm for unsupervised machine learning, used for clustering data. It is based on the principles of quantum mechanics and uses qubits to represent and manipulate data. The algorithm works by assigning each data point to a qubit and then using an optimization process to determine the best set of cluster centres that minimize the distance between each data point and its assigned cluster centre. It performs with poly logarithmic complexity; it is an efficient and accurate clustering algorithm that can be used for a variety of machine learning tasks.

Q means algorithm

Initially chose ‘k’ centroids c₁, c₂..c_k to be stored in the QRAM. Make sure t=0

Step 1: Centroid distance calculation using

equation

Step 2: Assign clusters by creating superimpositions using

equation

Calculate the minimum distance between

equation

and cluster centres.

Then assign superimposition for all data points and their labels.

Step 3:

equation

Step 4:

equation

Step 5: Until the convergence is achieved iterate steps 1-4.

In Q-means initially the input image is converted into a grey scale format. For the purpose of feature selection, the pixel intensity serves best to understand whether the pixel belongs to the region of interest ROI or not. Post which the images are compressed to 50 × 50 to apply a 5 × 5 Gaussian blurring filter to remove the noise. Since the task is to segment, the cluster (K) must have a value of 2. There after the individual pixels are each considered to be single featured vectors. In the q means, the vectors generated are stored in QRAM. Each of these vectors have different norms Q means, distance estimation is done through dot products of these generated vectors therefore the distance estimation is more efficient (Figure 1).

Figure 1: Segmentation flow of Q means algorithm.

Evaluation metrics

Swap test: A swap test is an algorithm used in quantum computing to determine whether two quantum systems are the same or different. It works by swapping the two systems and then measuring the effect of the swap. If the two systems are the same, then the swap will have no effect and the measurement will be the same before and after the swap. If the two systems are different, then the measurement will be different after the swap (Figure 2). Swap value is calculated by

Figure 2: Swap test circuit.

Discussion

In this section, the Q-means and K-means algorithms are compared in terms of methodology, computational time and resources, dataset sizes, and performance output. K-means and Q-means are two popular clustering algorithms. K-means is a classic clustering algorithm that uses a Euclidean distance metric and iterative refinement to find the optimal cluster centres. Q-means is a modified version of K-means which uses quantum bits, QRAM, circuits, also uses a different distance metric, the squared Euclidean distance metric, and a more sophisticated iterative refinement process. K-means is simpler to implement than Q-means, since it does not require any special tuning of the distance metric or refinement procedure. However, Q-means is more accurate, since it takes into account the actual distances between points when finding the cluster centres. Q-means also tends to be faster than K-means, since it only needs to compute the squared Euclidean distance once. Nevertheless, Q-means execution requires a high performing quantum computer which is a very expensive and scarce resource while K means can be implemented in with relatively much lower cost. Overall, both algorithms are useful for clustering data and can produce good results, but Q-means tends to be more accurate and faster than K-means. As a future work, we intend to explore the Quantum-CNN algorithm in comparison with traditional CNN models.

Conclusion

Quantum machine learning is a rapidly growing field of research which aims to develop algorithms and models that can take advantage of the unique properties of quantum computers to solve complex problems. In this paper, detailed methodology of quantum machine learning is presented. A comparative analysis is performed between the traditional K-means clustering algorithm and its quantum equivalent, Q-means algorithm. The findings of this study support the notion that quantum machine learning has the potential to revolutionize the way we approach computing, and to provide significant improvements in efficiency and accuracy.