Comparative Study of Classical Image Processing Techniques for Brain Tumor Segmentation in MRI Scans

Abstract

Brain Tumor segmentation plays a vital role in the early diagnosis and treatment planning of neurological disorders. While modern deep learning approaches have shown remarkable accuracy, classical image processing techniques remain significant due to their simplicity, lower computational requirements, and interpretability. This study presents a comparative analysis of four classical segmentation methods—thresholding, edge detection, region growing, and watershed—for segmenting brain tumours from MRI images. Each technique is evaluated against ground truth masks using metrics such as Dice coefficient, Jaccard index, accuracy, sensitivity, and precision. Experimental results show that although no single method outperforms the others in all metrics, region growing and watershed methods offer better segmentation quality for complex tumour boundaries. This study emphasises the continued relevance of classical methods as lightweight and effective solutions in constrained environments.

Introduction

The text discusses image segmentation techniques used for detecting brain tumors in MRI scans. Images are represented as pixel matrices, and image processing involves enhancing or extracting useful information from these images. Medical image segmentation, particularly for brain tumors, is crucial for diagnosis and treatment planning.

The study evaluates four classical image segmentation methods—thresholding, edge detection, region growing, and watershed segmentation—on a dataset of 50 axial brain MRI scans with expert-annotated tumor masks. Preprocessing steps include grayscale conversion, noise reduction, and normalization.

Segmentation Techniques:

• Thresholding: Uses intensity values to separate tumor from non-tumor regions (Otsu’s method).

• Edge Detection: Detects boundaries but is noise-sensitive.

• Region Growing: Groups similar neighboring pixels but depends on seed selection.

• Watershed: Uses morphological gradients but prone to over-segmentation.

Evaluation Metrics:

Performance was measured using Dice Similarity Coefficient, Jaccard Index, Sensitivity, Precision, and Accuracy.

Results:

• Thresholding showed the highest sensitivity, detecting most tumor pixels but with many false positives (low precision).

• Region growing had the highest accuracy but very low sensitivity.

• Edge detection and watershed performed moderately but struggled with noise and tumor boundary clarity.

• Overall, all classical methods had low overlap scores with ground truth masks, indicating limited segmentation accuracy.

Limitations of Classical Methods:

• Sensitive to noise and intensity variations.

• Poor generalization to different tumor shapes and MRI variations.

• Tend to over- or under-segment tumors.

• Lack contextual understanding, unlike deep learning models.

• Require manual parameter tuning.

• Limited robustness to anatomical variability.

The study highlights the need for more advanced or hybrid methods, especially those incorporating machine learning, to improve brain tumor segmentation in clinical practice.

Conclusion

This study evaluated and compared four classical image segmentation techniques—thresholding, edge detection, region-based segmentation (region growing), and the watershed method—for brain tumour segmentation on MRI images. Each method was implemented using MATLAB and evaluated against manually annotated ground truth using standard performance metrics: Dice coefficient, Jaccard index, sensitivity, precision, and accuracy. Among the methods, the region-based approach demonstrated the highest segmentation accuracy and consistency, followed by watershed, edge detection, and thresholding, respectively. The region-based method particularly excelled in maintaining spatial coherence and detecting tumour boundaries accurately. Thresholding, although simple, showed poor performance due to its sensitivity to intensity variations and lack of contextual understanding. The results underscore the limitations of classical methods in handling complex tumour shapes, intensity inhomogeneities, and noise. These challenges often result in under-segmentation or over-segmentation, especially in heterogeneous MRI datasets.

References

[1] N. Otsu, “THRESHOLD SELECTION METHOD FROM GRAY-LEVEL HISTOGRAMS.,” IEEE Trans Syst Man Cybern, vol. SMC-9, no. 1, pp. 62–66, 1979, doi: 10.1109/TSMC.1979.4310076. [2] J. Canny, “A Computational Approach to Edge Detection,” IEEE Trans Pattern Anal Mach Intell, vol. PAMI-8, no. 6, pp. 679–698, 1986, doi: 10.1109/TPAMI.1986.4767851. [3] R. Adams and L. Bischof, “Seeded Region Growing,” IEEE Trans Pattern Anal Mach Intell, vol. 16, no. 6, pp. 641–647, 1994, doi: 10.1109/34.295913. [4] A. K. Jain, “Fundamentals of digital image processing. Anil K. Jain.,” Vision, no. November, p. 569, 1989. [5] “Gonzalez, R.C. and Woods, R.E. (2018) Digital Image Processing. 4th Edition, Pearson Education, New York, 1022 p. - References - Scientific Research Publishing.” Accessed: Jun. 16, 2025. [Online]. Available: https://www.scirp.org/reference/referencespapers?referenceid=3227103 [6] “(PDF) Use of Watersheds in Contour Detection.” Accessed: Jun. 16, 2025. [Online]. Available: https://www.researchgate.net/publication/230837989_Use_of_Watersheds_in_Contour_Detection [7] D. Anithadevi and K. Perumal, “A hybrid approach based segmentation technique for brain tumor in MRI Images,” Mar. 2016, Accessed: Jun. 16, 2025. [Online]. Available: https://arxiv.org/abs/1603.02447v1 [8] P. Tambe, J. Thomson, and N. Surti, “Comparative Study of Segmentation Techniques for Brain Tumor Detection,” International Journal of Engineering Research & Technology, vol. 5, no. 2, Feb. 2016, doi: 10.17577/IJERTV5IS020258. [9] S. Bauer, R. Wiest, L. P. Nolte, and M. Reyes, “A survey of MRI-based medical image analysis for brain tumor studies,” Phys Med Biol, vol. 58, no. 13, Jul. 2013, doi: 10.1088/0031-9155/58/13/R97. [10] . M. Z., .A. A. S., . S. H. S., . A. Z. J., and . A. K., “Brain Tumor Segmentation through Region-based, Supervised and Unsupervised Learning Methods: A Literature Survey Brain Tumor Segmentation through Image Processing Methods: A Literature Survey,” J Biomed Eng Med Imaging, vol. 6, no. 2, Apr. 2019, doi: 10.14738/JBEMI.62.6725. [11] A. Wadhwa, A. Bhardwaj, and V. Singh Verma, “A review on brain tumor segmentation of MRI images,” Magn Reson Imaging, vol. 61, pp. 247–259, Sep. 2019, doi: 10.1016/J.MRI.2019.05.043. [12] K. Ejaz, M. S. Mohd Rahim, M. Arif, D. Izdrui, D. M. Craciun, and O. Geman, “Review on Hybrid Segmentation Methods for Identification of Brain Tumor in MRI,” Contrast Media Mol Imaging, vol. 2022, 2022, doi: 10.1155/2022/1541980. [13] D. N. Louis et al., “The 2021 WHO Classification of Tumors of the Central Nervous System: a summary,” Neuro Oncol, vol. 23, no. 8, pp. 1231–1251, Aug. 2021, doi: 10.1093/NEUONC/NOAB106. [14] “Brain tumor dataset includes the mask and images.” Accessed: Jun. 17, 2025. [Online]. Available: https://www.kaggle.com/datasets/tinashri/brain-tumor-dataset-includes-the-mask-and-images [15] N. R. Pal and S. K. Pal, “A review on image segmentation techniques,” Pattern Recognit, vol. 26, no. 9, pp. 1277–1294, Sep. 1993, doi: 10.1016/0031-3203(93)90135-J. [16] M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: Active contour models,” Int J Comput Vis, vol. 1, no. 4, pp. 321–331, Jan. 1988, doi: 10.1007/BF00133570/METRICS. [17] S. Bauer, R. Wiest, L. P. Nolte, and M. Reyes, “A survey of MRI-based medical image analysis for brain tumor studies,” Phys Med Biol, vol. 58, no. 13, Jul. 2013, doi: 10.1088/0031-9155/58/13/R97. [18] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9351, pp. 234–241, 2015, doi: 10.1007/978-3-319-24574-4_28. [19] R. C. Gonzalez, “Support Package for Digital Image Processing,” 2018, Accessed: Jun. 17, 2025. [Online]. Available: www.ImageProcessingPlace.com [20] C. S. Rao and K. Karunakara, “A comprehensive review on brain tumor segmentation and classification of MRI images,” Multimed Tools Appl, vol. 80, no. 12, pp. 17611–17643, May 2021, doi: 10.1007/S11042-020-10443-1.

Copyright

Copyright © 2025 Mohit Kalyan, Dr. Aman Kumar Sharma. 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.