Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Computer-Aided Osteoporosis Diagnosis Using Transfer Learning with Enhanced Features from Stacked Deep Learning Modules

Ayesha Siddiqua, Rakibul Hasan, Anichur Rahman, Abu Saleh Musa Miah

TL;DR

The paper tackles automated knee osteoporosis diagnosis from X-ray images, addressing reliance on manual feature extraction and limited generalization in existing methods. It proposes a computer-aided diagnosis system that combines transfer learning with a stacked feature enhancement mechanism built on a ResNet50 backbone, followed by a five-block Conv-RELU-MaxPooling stack and dense classification layers. The approach achieves high accuracy across three public datasets and a combined set (up to 98%), outperforming or matching state-of-the-art baselines in binary and multiclass tasks. This method promises robust, scalable, and efficient screening support for clinicians, with potential extensions to real-time deployment and explainable AI to improve trust and adoption in practice.

Abstract

Knee osteoporosis weakens the bone tissue in the knee joint, increasing fracture risk. Early detection through X-ray images enables timely intervention and improved patient outcomes. While some researchers have focused on diagnosing knee osteoporosis through manual radiology evaluation and traditional machine learning using hand-crafted features, these methods often struggle with performance and efficiency due to reliance on manual feature extraction and subjective interpretation. In this study, we propose a computer-aided diagnosis (CAD) system for knee osteoporosis, combining transfer learning with stacked feature enhancement deep learning blocks. Initially, knee X-ray images are preprocessed, and features are extracted using a pre-trained Convolutional Neural Network (CNN). These features are then enhanced through five sequential Conv-RELU-MaxPooling blocks. The Conv2D layers detect low-level features, while the ReLU activations introduce non-linearity, allowing the network to learn complex patterns. MaxPooling layers down-sample the features, retaining the most important spatial information. This sequential processing enables the model to capture complex, high-level features related to bone structure, joint deformation, and osteoporotic markers. The enhanced features are passed through a classification module to differentiate between healthy and osteoporotic knee conditions. Extensive experiments on three individual datasets and a combined dataset demonstrate that our model achieves 97.32%, 98.24%, 97.27%, and 98.00% accuracy for OKX Kaggle Binary, KXO-Mendeley Multi-Class, OKX Kaggle Multi-Class, and the combined dataset, respectively, showing an improvement of around 2% over existing methods.

Computer-Aided Osteoporosis Diagnosis Using Transfer Learning with Enhanced Features from Stacked Deep Learning Modules

TL;DR

The paper tackles automated knee osteoporosis diagnosis from X-ray images, addressing reliance on manual feature extraction and limited generalization in existing methods. It proposes a computer-aided diagnosis system that combines transfer learning with a stacked feature enhancement mechanism built on a ResNet50 backbone, followed by a five-block Conv-RELU-MaxPooling stack and dense classification layers. The approach achieves high accuracy across three public datasets and a combined set (up to 98%), outperforming or matching state-of-the-art baselines in binary and multiclass tasks. This method promises robust, scalable, and efficient screening support for clinicians, with potential extensions to real-time deployment and explainable AI to improve trust and adoption in practice.

Abstract

Knee osteoporosis weakens the bone tissue in the knee joint, increasing fracture risk. Early detection through X-ray images enables timely intervention and improved patient outcomes. While some researchers have focused on diagnosing knee osteoporosis through manual radiology evaluation and traditional machine learning using hand-crafted features, these methods often struggle with performance and efficiency due to reliance on manual feature extraction and subjective interpretation. In this study, we propose a computer-aided diagnosis (CAD) system for knee osteoporosis, combining transfer learning with stacked feature enhancement deep learning blocks. Initially, knee X-ray images are preprocessed, and features are extracted using a pre-trained Convolutional Neural Network (CNN). These features are then enhanced through five sequential Conv-RELU-MaxPooling blocks. The Conv2D layers detect low-level features, while the ReLU activations introduce non-linearity, allowing the network to learn complex patterns. MaxPooling layers down-sample the features, retaining the most important spatial information. This sequential processing enables the model to capture complex, high-level features related to bone structure, joint deformation, and osteoporotic markers. The enhanced features are passed through a classification module to differentiate between healthy and osteoporotic knee conditions. Extensive experiments on three individual datasets and a combined dataset demonstrate that our model achieves 97.32%, 98.24%, 97.27%, and 98.00% accuracy for OKX Kaggle Binary, KXO-Mendeley Multi-Class, OKX Kaggle Multi-Class, and the combined dataset, respectively, showing an improvement of around 2% over existing methods.

Paper Structure

This paper contains 45 sections, 26 equations, 10 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Challenges in Current Knee Osteoporosis Disease Detection Work
  3. Motivation
  4. The Goal and Scope of the Study
  5. Literature Review
  6. Traditional Diagnostic Methods
  7. Alternative Diagnostic Approaches
  8. Machine Learning in Osteoporosis Diagnosis
  9. Advancements in Medical Imaging and AI
  10. Dataset Description
  11. Osteoporosis Knee X-ray Binary Dataset (OKX Kaggle Binary Dataset)
  12. Knee X-ray Osteoporosis Mendeley Dataset (KXO-Mendely Multi-Class Dataset)
  13. Osteoporosis knee x-ray Kaggle multiclass dataset (OKX Kaggle Multiclass Dataset)
  14. Proposed Methodology
  15. Data Preprocessing
  16. ...and 30 more sections

Figures (10)

  • Figure 1: Visual Information of Knee X-rays: Healthy bone and affected by Osteoporosis from OKX Kaggle Binary Dataset
  • Figure 2: Visual Information of Knee X-rays: Healthy bone and affected by Osteoporosis from KXO-Mendely Multi-Class Dataset and OKX Kaggle Multiclass Dataset
  • Figure 3: Architecture of the Proposed Methodology
  • Figure 4: Model Architecture
  • Figure 5: (a) Training and Validation Accuracy curve, and (b) Training and Validation Loss curve for OKX Kaggle Binary Classification
  • ...and 5 more figures