A Focused Study on Otsus Thresholding for Segmenting Images of Paralysis-Affected Individuals

Abstract

A crucial step in the diagnosis and treatment of neurological conditions like paralysis is medical picture segmentation. This study examines the application of Otsu\'s thresholding technique for segmenting medical images of patients with paralysis. The automatic global thresholding technique developed by Otsu is used to maximize the inter-class variance between foreground and background pixels in order to extract regions of interest. In computer vision and digital image processing, where the main goal is to divide a picture into meaningful structures, image segmentation is essential. The simplicity, effectiveness, and efficiency of Otsu\'s thresholding in differentiating foreground from background in grayscale photographs make it stand out among other segmentation techniques. This work provides a thorough examination of Otsu\'s approach, covering its algorithmic implementation, mathematical underpinnings, and empirical testing on a variety of image datasets. We investigate the method\'s shortcomings in more detail and suggest improvements, including multi-level thresholding and preprocessing stages to deal with real-world issues like noise and uneven lighting.

Introduction

Image segmentation is a core technique in computer vision used to divide images into meaningful regions for tasks like object detection and medical diagnosis. In this study, Otsu’s thresholding, a widely used global segmentation method, is applied to medical images of patients with paralysis to support clinical understanding and treatment planning.

Main Contributions

• Applied Otsu's method to segment images from a custom dataset of paralyzed patients.

• Evaluated results using the Dice Coefficient, Jaccard Index (IoU), and execution time.

• Analyzed the strengths and limitations of Otsu’s method in the context of medical image segmentation.

Related Work

• Hybrid approaches (e.g., CNN + Otsu) improve segmentation accuracy in complex medical images (Bhavani & Karunakara).

• Adaptive versions of Otsu’s method address challenges like low contrast and complex tissue structures (Wan).

• Optimization techniques (e.g., particle swarm optimization) further enhance thresholding effectiveness (Khairuzzaman & Chaudhury).

• 3D extensions of Otsu’s method improve segmentation in volumetric medical data like brain MRIs (Feng et al.).

• Foundational texts and surveys provide theoretical background and performance comparisons of thresholding methods (Gonzalez & Woods; Sezgin & Sankur).

• A paralysis-specific image database supports the development of automated diagnostic tools (Nandyal & Kausar).

Methodology

• Dataset: Real-time grayscale images collected using a 108MP mobile camera at a paralysis center.

• Preprocessing: Conversion to grayscale and pixel normalization.

• Otsu’s Algorithm:

  1. Compute histogram and normalize it.

  2. Iterate through all thresholds to calculate between-class variance.

  3. Select the threshold with the maximum variance to separate foreground and background.

Evaluation Metrics

• Dice Coefficient: Measures overlap between segmented output and ground truth.

• Jaccard Index (IoU): Evaluates intersection over union of segmentation sets.

• Execution Time: Assesses computational efficiency, impacted by algorithm complexity and system hardware.

Key Findings

• Otsu’s method performs best with bimodal histograms, achieving accurate binary segmentation.

• Limitations:

  • Reduced performance with overlapping pixel distributions or low contrast.

  • Sensitive to illumination changes and image noise.

• While efficient and simple, global thresholding may fall short in complex medical scenarios, suggesting a need for adaptive or hybrid methods.

Conclusion

Otsu\'s approach offers a dependable and effective threshold-based picture segmentation method. Images with clear foreground and background distributions work best with it. Although it is not as effective in complex situations, its hybrid forms and expansions show promise. Future studies might adapt Otsu to real-time embedded systems or use deep learning for context-aware thresholding. Otsu\'s thresholding was used and assessed in this work to separate medical photos from individuals who were paralyzed. The technique\'s low execution time and satisfactory segmentation accuracy make it suitable for clinical applications with constrained time or computational resources. In order to increase accuracy, future research will examine deep learning models and combine Otsu\'s approach with adaptive or region-based segmentation algorithms.

References

[1] H. S. Bhavani and K. Karunakara, \"Otsu Multi-level Adaptive Thresholding with Deep Convolution Neural Network for Diabetic Retinopathy Segmentation and Classification,\" Journal of Electrical Systems, vol. 20, no. 10s, pp. 1532–1541, 2024. [2] S. Wan, \"Adaptive Trapezoid Region Intercept Histogram Based Otsu Method for Brain MR Image Segmentation,\" Journal of Ambient Intelligence and Humanized Computing, vol. 12, no. 10, pp. 10047–10058, Oct. 2021 [3] A. K. M. Khairuzzaman and S. Chaudhury, \"Brain MR Image Multilevel Thresholding by Using Particle Swarm Optimization, Otsu Method, and Anisotropic Diffusion,\" International Journal of Applied Metaheuristic Computing, vol. 10, no. 3, pp. 91–106, Jul.–Sep. 2019. [4] Y. Feng, H. Zhao, X. Li, X. Zhang, and H. Li, \"A Multi-scale 3D Otsu Thresholding Algorithm for Medical Image Segmentation,\" Digital Signal Processing, vol. 60, pp. 186–199, Jan. 2017. [5] Gonzalez, R. C., & Woods, R. E. (2008). Digital Image Processing (3rd ed.). Pearson. [6] Sezgin, M., & Sankur, B. (2004). Survey over image thresholding techniques and quantitative performance evaluation. Journal of Electronic Imaging, 13(1), 146–168.. [7] Nandyal, S., & Kausar, S. (2024, May 25-27). Development of comprehensive database system of paralysis patients for automated diagnosis. In Proceedings of the International Conference on Recent Trends in AI, IoT, Machine Learning, Image Processing, and Their Applications (ICAIMI -2024). ISBN: 978-93-340-6239-7. [8] Rasi, J. S. D. (2016). A Survey on Image Segmentation Algorithms. International Journal of Computer Trends and Technology, 35(4), 170–174. [9] Sakhare, S. V., & Nasre, V. G. (2011). Design of feature extraction in content-based image retrieval (CBIR) using color and texture. International Journal of Computer Science and Informatics, 1(2), 57–61. [10] Shahabi, C., & Safar, M. (2007). An experimental study of alternative shape-based image retrieval techniques. Multimedia Tools and Applications, 32(1), 29–48. [11] Song, Y., & Yan, H. (2017). Image Segmentation Techniques Overview. Asia Modelling Symposium (AMS), 103–107. DOI: 10.1109/AMS.2017.24. [12] Zhang, D., & Lu, G. (2002). An integrated approach to shape-based image retrieval. Proceedings of the 5th Asian Conference on Computer Vision (ACCV), Melbourne, Australia [13] Karma, I. G. M., & Susanti, J. (2020). Object Detection Using Color Dissimilarity Based Segmentation Method. International Conference on Applied Science and Technology (iCAST), DOI: 10.1109/iCAST51016.2020.9557737. [14] Khan, A. M., & Ravi, S. (2013). Image Segmentation Methods: A Comparative Study. International Journal of Soft Computing and Engineering (IJSCE), 3(4), 84. [15] Kumar, R., Gupta, R., & Jaiswal, A. K. (2022). Comparative Analysis of Image Segmentation Techniques for Brain MRI in Stroke Patients. Journal of Medical Imaging and Health Informatics, 12(5), 1205-1213. [16] Kumar, S., & Chauhan, E. A. A survey on feature extraction techniques for color images. [17] Li, J., Wang, J. Z., & Wiederhold, G. (2000). Classification of Textured and Non-Textured Images Using Region Segmentation. IEEE International Conference on Image Processing (ICIP), 754–757. [18] Mandloi, G. (2014). A Survey on Feature Extraction Techniques for Color Images. International Journal of Computer Science and Information Technology, 5(3), 4615–4620.

Copyright

Copyright © 2025 Shireen Kausar, Dr. Suvarna Nandyal. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.