Advances in Applied Nano-Bio Technologies

Published date: Oct 27 2025

Journal Title: Advances in Applied Nano-Bio Technologies

Issue title: Advances in Applied Nano-Bio Technologies: Volume 6 Issue 3

Pages: 98 - 110

DOI: 10.18502/aanbt.v6i3.18843

Hasti Hosseinimaryamghafari5454545@gmail.comDepartment of Biomedical Engineering, Tehran Medical Sciences Branch, Islamic Azad University, Tehran

Mohadeseh Parhizkarist_m_parhizkari@azad.ac.irDepartment of Biomedical Engineering, South Tehran Branch, Islamic Azad University, Tehran

Fatemeh Abdoliabasi.fatemeh123@gmail.comDepartment of Computer Engineering, Electronic Campus, Islamic Azad University, Tehran

Accurate and early diagnosis of brain tumors plays a vital role in improving patient survival and guiding effective treatment strategies. Magnetic Resonance Imaging (MRI) is widely regarded as the gold standard for brain tumor detection due to its superior soft-tissue contrast and non-invasive nature. However, the interpretation of MRI scans remains a complex and time-consuming task that relies heavily on expert radiologists, often leading to subjectivity and inter-observer variability. In response to these challenges, artificial intelligence (AI), and in particular deep learning methods, have shown remarkable promise in automating and enhancing diagnostic accuracy. Among deep learning models, Residual Neural Networks (ResNet) have gained prominence for their ability to overcome the vanishing gradient problem and to train very deep architectures effectively. This review provides a comprehensive overview of the performance of various ResNet models-ResNet18, ResNet34, ResNet50, and their modified variants—in detecting brain tumors from MRI images. Drawing upon findings from fourteen recent peer-reviewed studies, we explore the use of ResNet in tumor classification, segmentation, and localization tasks, with reported accuracies reaching as high as 98.52%. Additionally, the integration of explainable AI techniques, especially Gradient-weighted Class Activation Mapping (Grad-CAM), provides interpretability to the deep models, enabling clinicians to visualize the regions of interest that influenced the model’s predictions. Overall, this review highlights the significant role of ResNetbased architectures in advancing automated brain tumor diagnosis and presents a critical perspective on current challenges and future research opportunities. Finally, we discuss the current limitations of ResNet applications in medical imaging, including dataset imbalance, domain shift, and reproducibility issues, and outline directions for future research.

Keywords: brain tumor detection, magnetic resonance imaging, ResNet, deep learning, interpretability

[1] Musthafa MM, Arunkumar M, Azeez NA, Alghamdi NS, Khan S, Alshamrani SS, et al. Brain tumor classification using deep transfer learning with ResNet50 and Grad-CAM. J Imaging Med Artif Intell. 2024;11(1):45-58.

[2] Aggarwal M, Gupta R, Choudhary A, Sharma V, Singh H, Kumar R, et al. Efficient deep residual network for brain tumor segmentation using MRI images. Biomed Signal Process Control. 2023;81:104276.

[3] Patil R, Kirange D. Classification of brain tumors using ensemble of CNN models. Procedia Comput Sci. 2023;219:1242-1249.

[4] Abdusalomov A, Cho YI, Whangbo TK. A robust deep learning approach for brain tumor detection using modified ResNet and YOLOv7. Sensors. 2023;23(3):1087.

[5] Kanchanamala P, Revathi KG, Belsam Jeba Ananth M. Optimization-enabled hybrid deep learning for brain tumor classification. Biomed Signal Process Control. 2023;84:104955.

[6] Mahmud M, Ishtyaq M, Mamun M, Abdelgawad A. ResNet-based brain tumor detection using deep convolutional neural networks. Int J Comput Appl. 2023;175(13):17.

[7] Wozniak M, Siemiński A, Puchalski M, Wieczorek M, Siłka J, Jastrzębski J, et al. Deep learning in brain tumor classification: ResNet with correlation learning mechanism. Neurocomputing. 2023;512:134-144.

[8] Asad M, Khan A, Ali H, Ahmed S, Yousaf M, Tariq M, et al. Brain tumor detection using CNN with stochastic gradient descent optimization. Expert Syst Appl. 2023;213:118960.

[9] Archana A, Komarasamy G. ShCNN and deep CNN for MRI-based brain tumor detection. Meas Sci Sensors. 2023;26:100598.

[10] Gayathri V, Prakash A, Subramani S, Ramesh K, Senthilkumar M. Brain tumor classification using VGG- 16 with transfer learning. Comput Biol Chem. 2023;99:107792.

[11] Haq AU, Ejaz W, Khan M, Qamar R, Hussain S, Ahmad I, et al. Improved brain tumor segmentation with CNN and conditional random fields. Appl Soft Comput. 2023;130:109621.

[12] Rahman A, Islam MR, Akter N, Chowdhury S, Hossain M, Mahmud T, et al. PDCNN: a parallel deep CNN approach for brain tumor detection. J Digit Imaging. 2023;36:1123-1134.

[13] Anaya-Isaza A, Restrepo R, Valencia F, Jaramillo A, Castaño R, Arboleda C, et al. Cross-transformer fusion with deep learning for brain tumor classification. Comput Med Imaging Graph. 2023;102:102155.

[14] Kumar S, Gupta P, Singh R, Sharma A, Verma V, Choudhary N, et al. Deep CNN-based approach for brain tumor classification. Pattern Recognit Lett. 2023;169:21-28.