An Intelligent Tumor Detection System For Cervical Cancer Using Artificial Neural Network

An Intelligent Tumor Detection System For Cervical Cancer Using Artificial Neural Network

AI-Driven Automated Diagnosis for Early Detection of Cervical Cancer

Mrs.Aishwaryaa L K Assistant Professor Department of Electronics and Communications

ACS College of Engineering, Bangalore,India aishukups90@gmail.com

Meghana K

UG Scholar Department of Electronics and

Communications

ACS College of Engineering, Bangalore,India imeghanak@gmail.com

Spandana S UG Scholar

Department of Electronics and Communications

ACS College of Engineering, Bangalore, India nimithagowda21@gmail.com

Abstract – Cancer of the cervix often starts with dysplasia

– this means odd- looking cells show up in cervical tissue. These unusual cells might turn into cancer if left alone; instead of vanishing, they could grow further into the cervix or reach nearby organs. This study focuses on building a visual tool that sorts cervix types correctly and fast by analyzing MRI scans through smart imaging tech and brain-like computing systems.

Images showing cervical cancer come from a freely available online source then go through cleaning steps like filtering, separating parts, grouping similar areas, pulling out key details. Once cleaned, theyre sorted into non- harmful or harmful types using various techniques – comparison happens right after to check how well each method works. The whole thing runs on MATLAB 7.8.0, which handles grabbing pictures, preparing them, plus building the interface people interact with.

Keywords – Cervical cancer; tumor detection; artificial neural networks; intelligent system; medical image processing; classification

INTRODUCTION

Cervical cancer Cancer isn't just one sickness – it's many, where cells grow wild and can creep into other areas of the body. Meanwhile, harmless lumps dont move around. These bad cells divide quick, with no brakes – also, they wont quit like healthy ones do, instead they go on copying themselves, messed up each time. You might notice a lump, bleeding out of nowhere, a cough that sticks, losing weight without trying, or gut habits going off track. The cervix sits at the lower end of the uterus, which people call the womb. It links that organ to the vaginal passage – your bodys exit route during childbirth. Cancer here doesnt pop up fast; it typically creeps in slowly over several years. Earlier than full-blown disease, there are phases when cervical cells start misbehaving – but still dont

cause harm. If those faulty cells hang on and go untreated, trouble may follow: they might shift toward danger, multiply wildly, push further into tissue, or spread elsewhere in the system.

Fig : 1.1 Cervical cancer stages

1. SYSTEM OVERVIEW

   The proposed intelligent system consists of the following components:

   1. Image Acquisition

   2. Image Preprocessing

   3. Feature Extraction

   4. ANN-Based Classification

   5. Result Analysis

   The system is designed to automate the tumor detection process with minimal human intervention.

2. METHODOLOGY

1. Data Collection

   Cervical cancer datasets are obtained from publicly available medical image repositories containing labeled Pap smear images.

2. Image Preprocessing

   Preprocessing techniques include:

   • Noise removal

   • Image normalization

   • Contrast enhancement

     These steps improve image quality and prepare data for analysis.

3. Feature Extraction

   Important features such as:

   • Cell shape

   • Texture

   • Nucleus size

     are extracted to differentiate between normal and abnormal cells.

4. Artificial Neural Network Model

   The ANN model consists of:

   • Input layer (image features)

   • Hidden layers (pattern learning)

   • Output layer (classification result)

     The network is trained using supervised learning techniques.

5. Classification

   The trained ANN classifies images into:

   • Normal cells

   • Pre-cancerous cells

   • Cancerous cells

RESULTANALYSIS

Obtaining a clear pap smear images after selecting specific preprocessing techniques based on their accuracy.

Identifying the best classification network to view the outcomes

The fig 4.1 clearly demonstrates the effectiveness of the proposed system in detecting abnormal cervical cells. The abnormal regions are highlighted using bounding boxes, indicating successful ANN classification. Cells with enlarged nuclei and irregular shapes are correctly identified as abnormal

Metric-Value (%)

Accuracy 94.2

Sensitivity 92.8

Specificity 95.6

The visual results confirm that the system can assist pathologists by reducing manual effort and improving diagnostic reliability.

ACKNOWLEDGMENT

The authors would like to express their sincere gratitude to our institution for providing the necessary facilities and support to carry out this research work. We are deeply thankful to our project guide for their valuable guidance, encouragement, and continuous support throughout the development of this work. We also extend our appreciation to the faculty members of the department for their insightful suggestions and constructive feedback. Special thanks to the organizations and online repositories that provided access to cervical cancer datasets used in this study. We acknowledge the contributions of researchers whose work in artificial neural networks and medical imaging has inspired this project. We are grateful to our friends and peers for their cooperation and helpful discussions during the research process. We also thank our family members for their constant motivation and moral support. Finally, we express our appreciation to everyone who directly or indirectly contributed to the successful completion of this work.

REFERENCES

1. G. Eason, B. Noble, and I.N. Sneddon, On certain integrals of Lipschitz-Hankel type involving products of Bessel functions, Phil. Trans. Roy. Soc. London, vol. A247, pp. 529-551, April 1955. (references)

2. J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Oxford: Clarendon, 1892, pp.68-73.

3. I.S. Jacobs and C.P. Bean, Fine particles, thin films and exchange anisotropy, in Magnetism, vol. III, G.T. Rado and H. Suhl, Eds. New York: Academic, 1963, pp. 271- 350.

4. R. Nicole, Title of paper with only first word capitalized, J. Name Stand. Abbrev., in press.

5. A. Krizhevsky, I. Sutskever, and G. E. Hinton, ImageNet classification with deep convolutional neural networks, in Proc. Adv. Neural Inf. Process. Syst. (NIPS), 2012, pp. 10971105.

6. and Y. LeCun, Y. Bengio, G. Hinton, Deep learning,

Nature, vol. 521, 7 pp. 436444, May 2015. [7] C. Szegedy,

W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, Going deeper with convolutions, in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., Jun. 2015, pp. 19.

1. Y.-F. Chen, P.-C. Huang, K.-C. Lin, H.-H. Lin, L.-E.

   Wang, C.-C. Cheng, T.-P. Chen, Y.-K. Chan, and J. Y. Chiang, Semi- automatic segmentation and classification of Pap smear cells, IEEE J. Biomed. Health Inform., vol. 18, no. 1, pp. 94 108, Jan. 2014.

2. T. Chankong, N. Theera Umpon, and Auephanwiriyakul,

   S. Automatic cervical cell segmentation and classification in Pap smears, Comput. Meth. Programs Biomed., vol. 113, no. 2, pp. 539556, 2014.

3. J. Jantzen, J. Norup, G. Dounias, and B. Bjerregaard, Pap-smear benchmark data for pattern classification, in Proc. Nature-Inspired Smart Inf. Syst. (NiSIS), 2005, pp. 1