Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Transfer Learning for Medical Image Classification with SMOTE: A Comparative Study

Md. Zehan Alam, Tonmoy Roy, H. M. Nahid Kawsar, Iffat Rimi

TL;DR

This study evaluates transfer learning (TL) for multilabel medical image classification on Brain Tumor MRI and Diabetic Retinopathy datasets, comparing five pretrained CNNs. To address severe class imbalance in DR, it integrates SMOTE-based resampling with traditional ML classifiers on features extracted from TL models, achieving notable improvements in sensitivity and overall accuracy. TL excels on the balanced BT dataset, while DR benefits from the SMOTE-ensemble approach, with the Voting Classifier reaching 92.14% accuracy and high specificity. The work demonstrates a practical, low-computation pathway to improve medical image diagnostics by combining TL with resampling and ensemble learning, with potential applicability to other imaging domains.

Abstract

This paper explores and enhances the application of Transfer Learning (TL) for multilabel image classification in medical imaging, focusing on brain tumor class and diabetic retinopathy stage detection. The effectiveness of TL-using pre-trained models on the ImageNet dataset-varies due to domain-specific challenges. We evaluate five pre-trained models-MobileNet, Xception, InceptionV3, ResNet50, and DenseNet201-on two datasets: Brain Tumor MRI and APTOS 2019. Our results show that TL models excel in brain tumor classification, achieving near-optimal metrics. However, performance in diabetic retinopathy detection is hindered by class imbalance. To mitigate this, we integrate the Synthetic Minority Over-sampling Technique (SMOTE) with TL and traditional machine learning(ML) methods, which improves accuracy by 1.97%, recall (sensitivity) by 5.43%, and specificity by 0.72%. These findings underscore the need for combining TL with resampling techniques and ML methods to address data imbalance and enhance classification performance, offering a pathway to more accurate and reliable medical image analysis and improved patient outcomes with minimal extra computation powers.

Enhancing Transfer Learning for Medical Image Classification with SMOTE: A Comparative Study

TL;DR

This study evaluates transfer learning (TL) for multilabel medical image classification on Brain Tumor MRI and Diabetic Retinopathy datasets, comparing five pretrained CNNs. To address severe class imbalance in DR, it integrates SMOTE-based resampling with traditional ML classifiers on features extracted from TL models, achieving notable improvements in sensitivity and overall accuracy. TL excels on the balanced BT dataset, while DR benefits from the SMOTE-ensemble approach, with the Voting Classifier reaching 92.14% accuracy and high specificity. The work demonstrates a practical, low-computation pathway to improve medical image diagnostics by combining TL with resampling and ensemble learning, with potential applicability to other imaging domains.

Abstract

This paper explores and enhances the application of Transfer Learning (TL) for multilabel image classification in medical imaging, focusing on brain tumor class and diabetic retinopathy stage detection. The effectiveness of TL-using pre-trained models on the ImageNet dataset-varies due to domain-specific challenges. We evaluate five pre-trained models-MobileNet, Xception, InceptionV3, ResNet50, and DenseNet201-on two datasets: Brain Tumor MRI and APTOS 2019. Our results show that TL models excel in brain tumor classification, achieving near-optimal metrics. However, performance in diabetic retinopathy detection is hindered by class imbalance. To mitigate this, we integrate the Synthetic Minority Over-sampling Technique (SMOTE) with TL and traditional machine learning(ML) methods, which improves accuracy by 1.97%, recall (sensitivity) by 5.43%, and specificity by 0.72%. These findings underscore the need for combining TL with resampling techniques and ML methods to address data imbalance and enhance classification performance, offering a pathway to more accurate and reliable medical image analysis and improved patient outcomes with minimal extra computation powers.
Paper Structure (24 sections, 5 equations, 5 figures, 6 tables)

This paper contains 24 sections, 5 equations, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. CNN for Medical Image Classification
  4. Transfer Learning in Medical Image Classification
  5. Transfer Learning in Brain Tumor Detection
  6. Transfer Learning in Diabetic Retinopathy Stage Detection
  7. Research Methodology
  8. Datasets
  9. Data Preprocessing and Augmentation
  10. Convolutional Neural Networks (CNNs)
  11. Transfer Learning
  12. MobileNet
  13. Xception
  14. InceptionV3
  15. ResNet50
  16. ...and 9 more sections

Figures (5)

  • Figure 1: Sample Images from Each Class of BT Dataset
  • Figure 2: Sample Images from Each Class of DR Dataset
  • Figure 3: Workflow of the Proposed Model for Medical Image Classification
  • Figure 4: Confusion Matrix of Final Prediction on BT Dataset
  • Figure 5: Confusion Matrix for Final Prediction in DR Dataset