Title: Streamlining colorectal cancer diagnosis: leveraging MobileNet-V3 for efficient image classification

Authors: Artatrana Biswaprasanna Dash; Sachikanta Dash; Sasmita Padhy; Biswaranjan Mishra; Bijay Kumar Paikaray

Addresses: Department of CSE, GIET University, Gunupur, Odisha, India ' Department of CSE, GIET University, Gunupur, Odisha, India ' School of Computing Science and Engineering (SCSE), VIT Bhopal University, Madhya Pradesh, India ' Department of CSE, GIET University, Odisha, India ' Department of Computer Science and Engineering, Centre for Data Science, Siksha 'O' Anusandhan (Deemed to be) University, Bhubaneswar, Odisha, India

Abstract: Colorectal cancer ranks second in cancer-related deaths, emphasising its impact on public health. Recent advancements in medical image analysis, particularly through deep-learning, have improved cancer diagnosis. This study focuses on utilising MobileNet-V3, a streamlined convolutional neural network, to classify colorectal malignancies using medical imaging data. Transfer learning-based MobileNet model is employed, trained on publicly available histopathological images from the Kather_image_tiles dataset. The model is fine-tuned to distinguish between malignant and benign tissues. Performance assessment involves thorough validation using a distinct set of images, evaluating key metrics, delivering a comprehensive analysis of its predictive proficiency. The performance of the suggested model is compared with existing deep learning and traditional classification methods. Results show the MobileNet-V3 approach achieves a test accuracy of 90%; performance metrics achieved F1-score of 76%, demonstrating its potential for accurate and efficient colorectal cancer classification. This application can be beneficial for medical practitioners for quick analysis of colorectal malignancy type at its initial stage that can save many lives in danger. The lightweight nature of MobileNet facilitates deployment on resource-constrained devices, paving the way for real-time clinical decision support systems.

Keywords: colon malignancy; MobileNet-V3; cancer diagnosis; image classification.

DOI: 10.1504/IJIMS.2025.150839

International Journal of Internet Manufacturing and Services, 2025 Vol.11 No.4, pp.317 - 340

Received: 21 May 2024
Accepted: 14 Jul 2024
Published online: 24 Dec 2025 *

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