Hybrid Attention-Enhanced CNNs for Small Object Detection in Mammography, CT, and Fundus Imaging

Author First name, Last name, Institution

Hewa Majeed Zangana, Duhok Polytechnic University
Marwan Omar, Illinois Institute of Technology
Shuai Li, Oulun Yliopisto
Jamal N. Al-Karaki, Zayed University
Anik Vega Vitianingsih, Universitas Dr. Soetomo

Document Type

Article

Source of Publication

Buletin Ilmiah Sarjana Teknik Elektro

Publication Date

9-1-2025

Abstract

Early detection of subtle pathological features in medical images is critical for improving patient outcomes but remains challenging due to low contrast, small lesion size, and limited annotated data. The research contribution is a hybrid attention-enhanced CNN specifically tailored for small object detection across mammography, CT, and retinal fundus images. Our method integrates a ResNet-50 backbone with a modified Feature Pyramid Network, dilated convolutions for contextual scale expansion, and combined channel– spatial attention modules to preserve and amplify fine-grained features. We evaluate the model on public benchmarks (DDSM, LUNA16, IDRiD) using standardized preprocessing, extensive augmentation, and cross-validated training. Results show consistent gains in detection and localization: ECNN achieves an F1-score of 88.2% (95% CI: 87.4–89.0), mAP@0.5 of 86.8%, IoU of 78.6%, and a low false positives per image (FPPI = 0.12) versus baseline detectors. Ablation studies confirm the individual contributions of dilated convolutions, attention modules, and multi-scale fusion. However, these gains involve higher computational costs (≈2× training time and increased memory footprint), and limited dataset diversity suggests caution regarding generalizability. In conclusion, the proposed ECNN advances small-object sensitivity for early disease screening while highlighting the need for broader clinical validation and interpretability tools before deployment.

DOI Link

10.12928/biste.v7i3.14015

ISSN

2685-7936

Volume

7

Issue

3

First Page

595

Last Page

607

Disciplines

Computer Sciences | Medicine and Health Sciences

Keywords

Attention Mechanisms, Computational Efficiency, Dataset Bias, Dilated Convolutions, Early Disease Screening, Hybrid CNN, Multi-Scale Feature Fusion

Scopus ID

105019535333

Recommended Citation

Zangana, Hewa Majeed; Omar, Marwan; Li, Shuai; Al-Karaki, Jamal N.; and Vitianingsih, Anik Vega, "Hybrid Attention-Enhanced CNNs for Small Object Detection in Mammography, CT, and Fundus Imaging" (2025). All Works. 7687.
https://zuscholars.zu.ac.ae/works/7687

Indexed in Scopus

yes

Open Access

no