TY  - JOUR
AU  - Agarwal, Sunil Kumar
AU  - Gupta, Yogesh Kumar
PY  - 2026
DA  - 2026/10/19
TI  - A Comparative Analysis of Various Transfer Learning-Based Deep Learning Frameworks Using Brain Tumor MRI
JO  - ICCK Transactions on Machine Intelligence
T2  - ICCK Transactions on Machine Intelligence
JF  - ICCK Transactions on Machine Intelligence
VL  - 2
IS  - 4
SP  - 200
EP  - 212
DO  - 10.62762/TMI.2025.919091
UR  - https://www.icck.org/article/abs/TMI.2025.919091
KW  - Brain tumor MRI classification
KW  - transfer learning
KW  - pre-trained model
KW  - VGG16
KW  - Xception
KW  - ResNet50
KW  - MobileNetV1
AB  - Brain tumors are a critical and challenging health issue worldwide, and they often reduce patient survival if they are not identified at an early stage. This study presents an automatic approach to identify and classify brain tumor magnetic resonance imaging (MRI) scans using several pre-trained deep learning frameworks and compares their relative performance. Four popular frameworks, namely VGG16, Xception, ResNet50, and MobileNetV1, were fine-tuned with a transfer learning technique on 2870 brain tumor MRI training images to predict four classes (glioma, meningioma, pituitary tumor, and no tumor). A quantitative comparison among all applied frameworks was performed using accuracy, precision, recall, and F1-score. On the held-out test set, VGG16 achieved the highest accuracy (70.1%), followed by Xception (63.7%), MobileNetV1 (53.3%), and ResNet50 (50.8%). The results indicate that VGG16 and Xception outperform the other two frameworks under the same transfer learning setting, while all models show low recall for the glioma class. The study provides a baseline comparison of pre-trained frameworks for brain tumor MRI classification and outlines directions for future research.
SN  - 3068-7403
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Agarwal2026A,
  author = {Sunil Kumar Agarwal and Yogesh Kumar Gupta},
  title = {A Comparative Analysis of Various Transfer Learning-Based Deep Learning Frameworks Using Brain Tumor MRI},
  journal = {ICCK Transactions on Machine Intelligence},
  year = {2026},
  volume = {2},
  number = {4},
  pages = {200-212},
  doi = {10.62762/TMI.2025.919091},
  url = {https://www.icck.org/article/abs/TMI.2025.919091},
  abstract = {Brain tumors are a critical and challenging health issue worldwide, and they often reduce patient survival if they are not identified at an early stage. This study presents an automatic approach to identify and classify brain tumor magnetic resonance imaging (MRI) scans using several pre-trained deep learning frameworks and compares their relative performance. Four popular frameworks, namely VGG16, Xception, ResNet50, and MobileNetV1, were fine-tuned with a transfer learning technique on 2870 brain tumor MRI training images to predict four classes (glioma, meningioma, pituitary tumor, and no tumor). A quantitative comparison among all applied frameworks was performed using accuracy, precision, recall, and F1-score. On the held-out test set, VGG16 achieved the highest accuracy (70.1\%), followed by Xception (63.7\%), MobileNetV1 (53.3\%), and ResNet50 (50.8\%). The results indicate that VGG16 and Xception outperform the other two frameworks under the same transfer learning setting, while all models show low recall for the glioma class. The study provides a baseline comparison of pre-trained frameworks for brain tumor MRI classification and outlines directions for future research.},
  keywords = {Brain tumor MRI classification, transfer learning, pre-trained model, VGG16, Xception, ResNet50, MobileNetV1},
  issn = {3068-7403},
  publisher = {Institute of Central Computation and Knowledge}
}