Authors:
K. Jayasheel Kumar, Tarunika Sharma, M. Arunadevi Thirumalraj, S. Venkatasubramanian, Bushra Rehman

Addresses:
Department of Automobile Engineering, New Horizon College of Engineering, Bengaluru, Karnataka, India. Department of Applied Sciences, New Horizon College of Engineering, Bengaluru, Karnataka, India. Department of Computer Science and Engineering, Karunya Institute of Technology and Science, Coimbatore, Tamil Nadu, India. Department of Computer Science and Business Systems, Saranathan College of Engineering, Trichy, Tamil Nadu, India. Institute of Pathology and Diagnostic Medicine, Khyber Medical University, Peshawar, Khyber Pakhtunkhwa, Pakistan.

Abstract:

Histopathological images are used to diagnose breast cancer. Interpreting human histopathology images is difficult; hence, AI- and DL-based automated methods are advisable. Characterising the edge region of normal and deformed nuclei and highlighting the structure of small-shaped nuclei are the major challenges. A contourlet-driven attention network (CDAnet) addresses these challenges with a unique attention mechanism and content-preserving sampling. A controlled signal-driven attention system could replace spatial or channel-wise attention. This approach can focus on the nucleus structure boundary or edge information to obtain fine edge details. Contourlet transforms are proposed for managing signal generation. This approach uses wavelets' multi-scale temporal-frequency localisation and strong directionality. Final categorisation is done using the Attention-based Convolutional-Modified Inception Network (AI-UNet). The Improved Wolf package method fine-tunes this network's settings. A convolutional layer with multiple kernel sizes helps the modified inception module learn effective features quickly and efficiently, enabling kernel classification while balancing computation and the presentation of deeper layers. The BCSS model examines segmentation and classifier models using several criteria.

Keywords: Histopathological Images; Contourlet-Driven Attention Network; Improved Wolf Package Algorithm; Convolutional-Modified Inception Network; Nuclei Segmentation; AI and DL Techniques.

Received on: 02/12/2024, Revised on: 10/02/2025, Accepted on: 02/04/2025, Published on: 07/09/2025

DOI: 10.69888/FTSHSL.2025.000502

FMDB Transactions on Sustainable Health Science Letters, 2025 Vol. 3 No. 3, Pages: 165-178

• Views : 204
• Downloads : 11

Download PDF