Deep Learning Descriptor Hybridization with Feature Reduction for Accurate Cervical Cancer Colposcopy Image Classification

2024-05-01

arXiv.org

AI Summary

Low diagnostic accuracy in early-stage cervical cancer detection remains a critical clinical challenge. This paper proposes a multi-scale deep learning framework for colposcopic image analysis. Our method introduces a novel collaborative descriptor that fuses hierarchical features from ResNet50, ResNet101, and ResNet152, jointly modeling low-level (e.g., edges, color) and high-level (e.g., shape, texture) semantic representations. To enhance generalization, we systematically incorporate Linear Discriminant Analysis (LDA)-driven feature dimensionality reduction, coupled with Min-Max normalization to mitigate overfitting—first applied in this domain. Evaluated on the WHO/IARC standard colposcopy dataset—rigorously preprocessed via image segmentation and class balancing—our model achieves 97%–100% classification accuracy, substantially surpassing prior state-of-the-art methods (81%–91%). It establishes new benchmarks for both binary classification (normal vs. abnormal) and fine-grained lesion subtyping tasks.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• Security and Privacy: Large-scale security measurements

Abstract

Cervical cancer stands as a predominant cause of female mortality, underscoring the need for regular screenings to enable early diagnosis and preemptive treatment of pre-cancerous conditions. The transformation zone in the cervix, where cellular differentiation occurs, plays a critical role in the detection of abnormalities. Colposcopy has emerged as a pivotal tool in cervical cancer prevention since it provides a meticulous examination of cervical abnormalities. However, challenges in visual evaluation necessitate the development of Computer Aided Diagnosis (CAD) systems. We propose a novel CAD system that combines the strengths of various deep-learning descriptors (ResNet50, ResNet101, and ResNet152) with appropriate feature normalization (min-max) as well as feature reduction technique (LDA). The combination of different descriptors ensures that all the features (low-level like edges and colour, high-level like shape and texture) are captured, feature normalization prevents biased learning, and feature reduction avoids overfitting. We do experiments on the IARC dataset provided by WHO. The dataset is initially segmented and balanced. Our approach achieves exceptional performance in the range of 97%-100% for both the normal-abnormal and the type classification. A competitive approach for type classification on the same dataset achieved 81%-91% performance.

Problem

• Improving cervical cancer classification accuracy using ensemble models
• Combining detailed and abstract features from multiple ResNet blocks
• Introducing explainability to identify cancerous regions in images

Innovation

• Ensemble pooling of ResNet features
• Cascaded Block Fused ResNet variants
• Novel SHAP+LIME explainability method