TumorLens: AI-Based Brain Tumor Detection Using YOLOv8

Abstract

Brain tumors represent one of the most critical and life-threatening neurological conditions, requiring early detection for improved treatment outcomes. Manual MRI analysis remains the primary diagnostic tool, but it is slow, subjective, and depen- dent on radiological expertise. To address these limitations, this paper presents TumorLens, an end-to-end intelligent diagnostic platform that integrates YOLOv8 deep learning architecture for tumor detection, MERN stack for scalable deployment, Python-based inference services, and Large Language Model (LLM)–assisted diagnostic summarization. The system enables users to upload MRI scans through a secure web interface, where real-time preprocessing, feature extraction, and detection pipelines are executed to localize tumor regions with bounding boxes and confidence scores. The LLM further interprets detec- tions into a human-understandable diagnostic report, making the analysis accessible to non-experts. Designed as a cloud- ready medical support tool, TumorLens demonstrates high us- ability, interoperability, and performance. Experimental findings indicate significant reduction in diagnostic latency, enhanced interpretability, and improved detection accuracy. This survey paper provides a comprehensive overview of TumorLens, its architecture, supporting technologies, and the associated research foundations that underline its contribution to AI-driven medical imaging.

Introduction

Brain tumors are serious neurological conditions that can be benign or malignant, and early and accurate detection is crucial for effective treatment. MRI is commonly used for diagnosis due to its strong soft-tissue imaging capability, but manual interpretation is slow, complex, and prone to human error. This has led to the development of Computer-Aided Diagnosis (CAD) systems using AI.

Deep learning models like CNNs have shown strong performance in medical image classification, but they often struggle with tumor localization and require high computational resources. More advanced approaches such as segmentation models (e.g., U-Net, Mask R-CNN) improve localization but are still computationally heavy and not always suitable for real-time use.

Object detection models, especially YOLO (You Only Look Once), address these issues by enabling fast, real-time detection with high accuracy. Earlier versions like YOLOv5 and YOLOv7 have performed well in brain tumor detection, but limitations remain in accuracy, generalization, and efficiency.

To overcome these challenges, the study proposes a system called TumorLens, based on YOLOv8. YOLOv8 improves performance through anchor-free detection, better feature extraction, and optimized training, making it suitable for real-time, accurate tumor detection. The system classifies tumors into three types: meningioma, glioma, and pituitary tumors.

The methodology involves a full pipeline: MRI acquisition, preprocessing and annotation, dataset splitting, YOLOv8 training, tumor detection/classification, and performance evaluation. The system is designed to be scalable, efficient, and suitable for real-world clinical use.

Conclusion

This research proposed TumorLens, an AI-powered tumor detector with the help of the YOLOv8 deep learning technique capable of performing analyses on MRI scans. TumorLens was proposed as an efficient and fast way of automatic tumor detection, localization, and classification into meningiomas, gliomas, and pituitary tumors. The results presented by the authors prove that the proposed solution is significantly better than alternative options re- garding detection accuracy, efficiency, and stability. YOLOv8 provides high accuracy in tumor localization using bounding boxes, which allows performing real-time inference. Further- more, the use of preprocessing methods (such as normaliza- tion) improves the efficiency and robustness of the proposed model. The results of the comparative study have demonstrated that the suggested approach has numerous advantages compared to traditional methods, especially in regard to automation, scalability, and speed. The application of YOLOv8 does not only improve the efficiency of the system but also makes its implementation in actual medical facilities possible. There are still some limitations in TumorLens that might be taken into account while conducting future studies. The proposed algorithm is heavily dependent on the quality of used MRI data, which significantly limits its efficiency. Another drawback is the fact that the solution is designed for only two-dimensional MRI analysis. In conclusion, the authors proved that TumorLens is a promising and efficient tool for detecting brain tumors on MRI scans.

References

[1] M. F. Almufareh, M. Imran, A. Khan, and M. Humayun, “Automated brain tumor segmentation and classification in MRI using YOLO-based deep learning,” IEEE Access, vol. 12, pp. 16189–16204, 2024. [2] G. Jocher, A. Stoken, J. Borovec, and A. Chaurasia, “YOLOv5: Open- source object detection system,” GitHub Repository, 2023. [Online]. Available: https://github.com/ultralytics/yolov5 [3] Ultralytics, “YOLOv8: Next-generation object detection and segmen- tation model,” Ultralytics Documentation, 2023. [Online]. Available: https://docs.ultralytics.com [4] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2016, pp. 779–788. [5] A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Op- timal speed and accuracy of object detection,” arXiv preprint arXiv:2004.10934, 2020. [6] C.-Y. Wang, A. Bochkovskiy, and H.-Y. M. Liao, “YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,”arXiv preprint arXiv:2207.02696, 2022. [7] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional net- works for biomedical image segmentation,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. (MICCAI), 2015, pp. 234–241. [8] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2016, pp. 770–778. [9] G. Litjens et al.,“Asurvey on deep learning in medical image analysis,”Med. Image Anal., vol. 42, pp. 60–88, 2017. [10] J. Amin, M. Sharif, A. Haldorai, M. Yasmin, and R. S. Nayak, “Brain tumor detection and classification using machine learning: A compre- hensive survey,” Complex Intell. Syst., vol. 8, no. 4, pp. 3161–3183, 2022. [11] M. R. Islam, M. Rahman, and M. S. Islam, “Brain tumor detection in MRI images using deep learning techniques,” IEEE Access, vol. 11, pp. 45678–45690, 2023. [12] B. Menze et al., “The multimodal brain tumor image segmentation benchmark (BraTS),” IEEE Trans. Med. Imaging, vol. 34, no. 10, pp. 1993–2024, 2018. [13] K. He, G. Gkioxari, P. Doll´ar, and R. Girshick, “Mask R-CNN,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2017, pp. 2961–2969. [14] W. Liu et al., “SSD: Single shot multibox detector,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2016, pp. 21–37. [15] S. Pan and Q. Yang,“Asurvey on transfer learning,” IEEE Trans. Knowl. Data Eng., vol. 22, no. 10, pp. 1345–1359, 2010.

Copyright

Copyright © 2026 Ashish Sanjay Phalle, Mrs. Deepali Gohil, Abhay Shivaji Walunj, Suraj Rajendra Jadhav, Pruthviraj Ashokrao Bayas. 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.