An Explainable Multi-Model Deep Learning Framework for Breast Cancer Diagnosis from Histopathological Images
Abstract
Breast cancer is a significant cause of cancer-related deaths among women worldwide. Its early identification and screening are essential for improved patient outcomes and reduced mortality rates. Histopathological image analysis is considered as the gold standard for the diagnosis and prognosis of breast cancer. Nevertheless, the complexity of Whole-Slide Images (WSI) and their manual examination make this task time consuming, and prone to pathologist subjectivity. Recently, Deep Learning (DL) technology has achieved remarkable success in computer vision. However, their application still faces critical challenges in pathology analysis, including Region-of-Interest (RoI) scale variations, inter- and intra-class heterogeneity, diverse staining protocols, and the scarcity of annotated datasets. Furthermore, DL model’s findings are opaque and lack decision-level transparency. This study proposes a novel explainable multi-model DL framework for breast cancer classification leveraging histopathological images. The framework integrates Contrast Limited Adaptive Histogram Equalization (CLAHE) for image contrast enhancement, and diverse data augmentation to mitigate class imbalance and overfitting. Proposed architecture ensembles two branches, one employes DenseNet201 benefiting from Transfer Learning (TL) via ImageNet weights, while other utilizes a custom light weight attention based Hierarchal Feature Fusion (HFF) Network. DenseNet201 utilizes multilevel features to effectively tackle gradient vanishing issues and capture intricate feature representations, while HFF-Net, designed specifically for biomedical imaging, leverages HFF stem and multiscale feature extraction with Swish activation to enhance learning stability. Attention mechanism introduced within HFF-Net further refines the output features. Final feature vectors from the two branches are fused at the Global Average Pooling (GAP) layer, consolidating discriminative information. Experimental results on BRACS dataset demonstrate the proposed framework achieves 97.15% accuracy, 92.59% precision, 93.81% recall, and a 93.19% F1-score in screening tasks, while for grading tasks, it attains 84.08% accuracy, 83.39% precision, 83.64% recall, and 83.44% F1-score on 4391 test samples. Additionally, Gradient-Class Activation Mapping (Grad-CAM) saliency heatmap are generated for visual representation of proposed model’s choices, thereby increased transparency. The integration of these advanced techniques significantly enhances diagnostic reliability, addressing the challenges in histopathological image analysis.References
M. Xie et al., “Multi-resolution consistency semi-supervised active learning framework for histopathology image classification,” Expert Syst Appl, vol. 259, Jan. 2025, doi: 10.1016/j.eswa.2024.125266.
C. C. Ukwuoma et al., “Enhancing histopathological medical image classification for Early cancer diagnosis using deep learning and explainable AI – LIME & SHAP,” Biomed Signal Process Control, vol. 100, p. 107014, Feb. 2025, doi: 10.1016/J.BSPC.2024.107014.
World Health Organization, “Breast cancer”, Accessed: Jan. 29, 2025. [Online]. Available: https://www.who.int/news-room/fact-sheets/detail/breast-cancer
F. Yu et al., “BiaCanDet: Bioelectrical impedance analysis for breast cancer detection with space–time attention neural network,” Expert Syst Appl, vol. 269, p. 126223, Apr. 2025, doi: 10.1016/J.ESWA.2024.126223.
M. Zubair, S. Wang, and N. Ali, “Advanced Approaches to Breast Cancer Classification and Diagnosis,” Front Pharmacol, vol. 11, p. 632079, Feb. 2021, doi: 10.3389/FPHAR.2020.632079/BIBTEX.
M. Tafavvoghi et al., “Deep learning-based classification of breast cancer molecular subtypes from H&E whole-slide images,” J Pathol Inform, vol. 16, Jan. 2025, doi: 10.1016/j.jpi.2024.100410.
M. H. Ashraf, F. Jabeen, H. Alghamdi, M. S. Zia, and M. S. Almutairi, “HVD-Net: A Hybrid Vehicle Detection Network for Vision-Based Vehicle Tracking and Speed Estimation,” Journal of King Saud University - Computer and Information Sciences, vol. 35, no. 8, p. 101657, Sep. 2023, doi: 10.1016/J.JKSUCI.2023.101657.
M. H. Ashraf et al., “HIRD-Net: An Explainable CNN-Based Framework with Attention Mechanism for Diabetic Retinopathy Diagnosis Using CLAHE-D-DoG Enhanced Fundus Images,” Life, vol. 15, no. 9, p. 1411, Sep. 2025, doi: 10.3390/life15091411.
S. Singh et al., “Hybrid Models for Breast Cancer Detection via Transfer Learning Technique,” Computers, Materials and Continua, vol. 74, no. 2, pp. 3063–3083, 2023, doi: 10.32604/cmc.2023.032363.
N. Brancati et al., “BRACS: A Dataset for BReAst Carcinoma Subtyping in H&E Histology Images,” Database, vol. 2022, 2022, doi: 10.1093/database/baac093.
K. Wang, F. Zheng, D. Guan, J. Liu, and J. Qin, “Distilling heterogeneous knowledge with aligned biological entities for histological image classification,” Pattern Recognit, vol. 160, Apr. 2025, doi: 10.1016/j.patcog.2024.111173.
Q. He et al., “Global attention based GNN with Bayesian collaborative learning for glomerular lesion recognition,” Comput Biol Med, vol. 173, May 2024, doi: 10.1016/j.compbiomed.2024.108369.
M. Tafavvoghi, L. A. Bongo, N. Shvetsov, L. T. R. Busund, and K. Møllersen, “Publicly available datasets of breast histopathology H&E whole-slide images: A scoping review,” Dec. 01, 2024, Elsevier B.V. doi: 10.1016/j.jpi.2024.100363.
Y. Zheng et al., “Application of transfer learning and ensemble learning in image-level classification for breast histopathology,” Intelligent Medicine, vol. 3, no. 2, pp. 115–128, May 2023, doi: 10.1016/j.imed.2022.05.004.
V. Kumari and R. Ghosh, “A magnification-independent method for breast cancer classification using transfer learning,” Healthcare Analytics, vol. 3, Nov. 2023, doi: 10.1016/j.health.2023.100207.
J. Potsangbam and S. Shuleenda Devi, “Classification of Breast Cancer Histopathological Images Using Transfer Learning with DenseNet121,” in Procedia Computer Science, Elsevier B.V., 2024, pp. 1990–1997. doi: 10.1016/j.procs.2024.04.188.
R. Maurya, N. N. Pandey, M. K. Dutta, and M. Karnati, “FCCS-Net: Breast cancer classification using Multi-Level fully Convolutional-Channel and spatial attention-based transfer learning approach,” Biomed Signal Process Control, vol. 94, Aug. 2024, doi: 10.1016/j.bspc.2024.106258.
R. Karthik, R. Menaka, and M. V. Siddharth, “Classification of breast cancer from histopathology images using an ensemble of deep multiscale networks,” Biocybern Biomed Eng, vol. 42, no. 3, pp. 963–976, 2022, doi: 10.1016/j.bbe.2022.07.006.
S. Majumdar, P. Pramanik, and R. Sarkar, “Gamma function based ensemble of CNN models for breast cancer detection in histopathology images,” Expert Syst Appl, vol. 213, p. 119022, Dec. 2023, doi: 10.1016/J.ESWA.2022.119022.
Pramod K.B. Rangaiah, B.P. Pradeep Kumarb, and Robin Augustinea, “Histopathology-driven prostate cancer identification: A VBIR approach with CLAHE and GLCM insights,” Elsevier , vol. 182, 2024.
M. H. Ashraf and H. Alghamdi, “HFF-Net: A hybrid convolutional neural network for diabetic retinopathy screening and grading,” Biomedical Technology, vol. 8, pp. 50–64, Dec. 2024, doi: 10.1016/J.BMT.2024.09.004.
A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Adv Neural Inf Process Syst, vol. 25, 2012, [Online]. Available: http://code.google.com/p/cuda-convnet/
K. He, X. Zhang, S. Ren, and J. Sun, “Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition,” IEEE Trans Pattern Anal Mach Intell, vol. 37, no. 9, pp. 1904–1916, Dec. 2015, doi: 10.1109/TPAMI.2015.2389824.
M. E. Qureshi, M. H. Ashraf, M. W. Arshad, A. Khan, H. Ali, and Z. U. Abdeen, “Hierarchical Feature Fusion With Inception V3 for Multiclass Plant Disease Classification,” Informatica, vol. 49, no. 27, Jul. 2025, doi: 10.31449/inf.v49i27.8208.
M. H. Ashraf, M. E. Qureshi, A. Khan, and M. Ahmed, “DRD-Net: Diabetic Retinopathy Diagnosis Using A Hybrid Convolutional Neural Network,” International Journal on Robotics, Automation and Sciences, vol. 7, no. 2, p. 96, 2025, doi: 10.33093/ijoras.2025.7.2.9.
L. Solorzano, S. Robertson, B. Acs, J. Hartman, and M. Rantalainen, “Ensemble-based deep learning improves detection of invasive breast cancer in routine histopathology images,” Heliyon, vol. 10, no. 12, Jun. 2024, doi: 10.1016/j.heliyon.2024.e32892.
J. Potsangbam and S. S. Devi, “Classification of Breast Cancer Histopathological Images Using Transfer Learning with DenseNet121,” in Procedia Computer Science, Elsevier B.V., 2024, pp. 1990–1997. doi: 10.1016/j.procs.2024.04.188.
G. Huang, Z. Liu, L. van der Maaten, and K. Q. Weinberger, “Densely Connected Convolutional Networks,” 2017. [Online]. Available: https://github.com/liuzhuang13/DenseNet.
C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the Inception Architecture for Computer Vision,” 2016.
K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” 2016. Accessed: Jun. 27, 2023. [Online]. Available: http://image-net.org/challenges/LSVRC/2015/
A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Adv Neural Inf Process Syst, vol. 25, 2012, [Online]. Available: http://code.google.com/p/cuda-convnet/
A. Howard et al., “Searching for MobileNetV3.”
F. Chollet, “Xception: Deep Learning With Depthwise Separable Convolutions,” 2017.
DOI:
https://doi.org/10.31449/inf.v50i2.11993Downloads
Published
Issue
Section
License
Authors retain copyright in their work. By submitting to and publishing with Informatica, authors grant the publisher (Slovene Society Informatika) the non-exclusive right to publish, reproduce, and distribute the article and to identify itself as the original publisher.
All articles are published under the Creative Commons Attribution license CC BY 3.0. Under this license, others may share and adapt the work for any purpose, provided appropriate credit is given and changes (if any) are indicated.
Authors may deposit and share the submitted version, accepted manuscript, and published version, provided the original publication in Informatica is properly cited.
