Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Histopathological Image Classification via Integrated HOG and Deep Features with Robust Noise Performance

Ifeanyi Ezuma, Ugochukwu Ugwu

TL;DR

This work tackles automated histopathology image classification under challenging imaging conditions by evaluating a fine-tuned InceptionResNet-v2 as both a classifier and a feature extractor on the LC25000 dataset. It advances the field by proposing a fusion of traditional HOG features with deep representations and by assessing robustness to signal-to-noise ratio variations using multiple ML models, including Neural Networks, GBM, KNN, and SVM. The key findings show that deep-feature pipelines outperform the plain pre-trained network, with the Neural Network achieving an AUC of 99.99% and accuracy of 99.84%, while HOG integration provides additional gains for several models; however, the benefits are nuanced under higher noise. Overall, the integrated feature strategy delivers high accuracy and resilience, supporting more reliable automated histopathology classification in real-world, noisy imaging environments.

Abstract

The era of digital pathology has advanced histopathological examinations, making automated image analysis essential in clinical practice. This study evaluates the classification performance of machine learning and deep learning models on the LC25000 dataset, which includes five classes of histopathological images. We used the fine-tuned InceptionResNet-v2 network both as a classifier and for feature extraction. Our results show that the fine-tuned InceptionResNet-v2 achieved a classification accuracy of 96.01\% and an average AUC of 96.8\%. Models trained on deep features from InceptionResNet-v2 outperformed those using only the pre-trained network, with the Neural Network model achieving an AUC of 99.99\% and accuracy of 99.84\%. Evaluating model robustness under varying SNR conditions revealed that models using deep features exhibited greater resilience, particularly GBM and KNN. The combination of HOG and deep features showed enhanced performance, however, less so in noisy environments.

Enhancing Histopathological Image Classification via Integrated HOG and Deep Features with Robust Noise Performance

TL;DR

This work tackles automated histopathology image classification under challenging imaging conditions by evaluating a fine-tuned InceptionResNet-v2 as both a classifier and a feature extractor on the LC25000 dataset. It advances the field by proposing a fusion of traditional HOG features with deep representations and by assessing robustness to signal-to-noise ratio variations using multiple ML models, including Neural Networks, GBM, KNN, and SVM. The key findings show that deep-feature pipelines outperform the plain pre-trained network, with the Neural Network achieving an AUC of 99.99% and accuracy of 99.84%, while HOG integration provides additional gains for several models; however, the benefits are nuanced under higher noise. Overall, the integrated feature strategy delivers high accuracy and resilience, supporting more reliable automated histopathology classification in real-world, noisy imaging environments.

Abstract

The era of digital pathology has advanced histopathological examinations, making automated image analysis essential in clinical practice. This study evaluates the classification performance of machine learning and deep learning models on the LC25000 dataset, which includes five classes of histopathological images. We used the fine-tuned InceptionResNet-v2 network both as a classifier and for feature extraction. Our results show that the fine-tuned InceptionResNet-v2 achieved a classification accuracy of 96.01\% and an average AUC of 96.8\%. Models trained on deep features from InceptionResNet-v2 outperformed those using only the pre-trained network, with the Neural Network model achieving an AUC of 99.99\% and accuracy of 99.84\%. Evaluating model robustness under varying SNR conditions revealed that models using deep features exhibited greater resilience, particularly GBM and KNN. The combination of HOG and deep features showed enhanced performance, however, less so in noisy environments.
Paper Structure (10 sections, 5 figures, 3 tables)

This paper contains 10 sections, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Contributions
  4. Material and Methods
  5. Dataset
  6. Fine-Tuned Pre-trained CNN as Classifier
  7. Integration of Hog Features with Features Extracted from Pre-Trained CNN
  8. Results
  9. Evaluating the Impact of Signal-to-Noise Ratio (SNR) on Model Performance
  10. Conclusion

Figures (5)

  • Figure 1: Dataset LC25000. From left to right: benign lung tissue (Lung_n), lung adenocarcinoma (Lung_aca), lung squamous cell carcinoma tissue (Lung_scc), benign colonic tissue (Colon_n) and colon adenocarcinoma (Colon_aca) [30].
  • Figure 2: Performance Evaluation using the Fine-tuned Pre-trained InceptionResNet Model.
  • Figure 3: ROC curves obtained by the models. From left to right, the worst and the best performing model (a) Feature extraction using Decision Tree (b) Feature extraction using Neural Network.
  • Figure 4: Images with different levels of SNR applied. From left to right representing low, moderate and high level of background noise.
  • Figure 5: Accuracy of Various Machine Learning Models Across Different SNR Levels Under different methods (a) Features extracted from InceptionResNet-V2 and (b) Combined HOG and Deep Features extracted from InceptionResNet-V2.