Article: An intelligent cancer image classification framework using region-vision transformer-based adaptive multiscale EfficientNetB7 with gated recurrent unit layer Journal: International Journal of Signal and Imaging Systems Engineering (IJSISE) 2026 Vol.14 No.3 pp.200 - 228 Abstract: Cancer is a harmful disorder, so detecting cancer in the beginning stages helps to save human life. But, detecting the type of cancer by manual methods is difficult. Conventional techniques perform separate procedures for feature learning and dimensionality reduction. However, these processes need more time. An adaptive deep learning approach is developed to resolve the limitations of the traditional cancer image classification process. The required images for performing the classification process are gathered. After collecting the images, they are directly passed to the Region-Vision Transformer-based Adaptive Multi-scale EfficientNetB7 with Gated Recurrent Unit Layer (RViT-AMEB7-GRU) for cancer image classification, in which the attributes of the RViT-AMEB7-GRU are tuned using the Updated Random Function-aware Good, the Bad, and the Ugly Optimiser (URF-GBUO). The developed model is adopted by clinicians for detecting cancer at an earlier stage. Finally, the experimentation is done on the proposed model to prove effectiveness. Inderscience Publishers - linking academia, business and industry through research

Title: An intelligent cancer image classification framework using region-vision transformer-based adaptive multiscale EfficientNetB7 with gated recurrent unit layer

Authors: Vipul Gajjar; V.N. Kamalesh; Kavitha Rani Balmuri

Addresses: Department of Computer Engineering, Gandhinagar Institute of Technology, Gandhinagar University, Kalol, Gujarat, 382721, India ' Gandhinagar University, Moti Bhoyan, Kalol, Gujarat, 382721, India ' Computer Science & Engineering, School of Engineering & Technology, Kaveri University, Gowraram (V), Wargal (M), Siddipet (D), 502279 Telangana, India

Abstract: Cancer is a harmful disorder, so detecting cancer in the beginning stages helps to save human life. But, detecting the type of cancer by manual methods is difficult. Conventional techniques perform separate procedures for feature learning and dimensionality reduction. However, these processes need more time. An adaptive deep learning approach is developed to resolve the limitations of the traditional cancer image classification process. The required images for performing the classification process are gathered. After collecting the images, they are directly passed to the Region-Vision Transformer-based Adaptive Multi-scale EfficientNetB7 with Gated Recurrent Unit Layer (RViT-AMEB7-GRU) for cancer image classification, in which the attributes of the RViT-AMEB7-GRU are tuned using the Updated Random Function-aware Good, the Bad, and the Ugly Optimiser (URF-GBUO). The developed model is adopted by clinicians for detecting cancer at an earlier stage. Finally, the experimentation is done on the proposed model to prove effectiveness.

Keywords: cancer classification; medical images; region-vision transformer-based adaptive multiscale EfficientNetB7 with gated recurrent unit layer.

DOI: 10.1504/IJSISE.2026.152514

International Journal of Signal and Imaging Systems Engineering, 2026 Vol.14 No.3, pp.200 - 228

Received: 13 Nov 2024
Accepted: 26 Aug 2025
Published online: 24 Mar 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: An intelligent cancer image classification framework using region-vision transformer-based adaptive multiscale EfficientNetB7 with gated recurrent unit layer

Keep up-to-date