Automatic Thyroid Ultrasound Image Classification Using Feature Fusion Network

Zhao, Xiaohui; Shen, Xueqin; Wang, Wenbo; Lu, Yuanyuan; Hu, Shidong; Xiao, Ruoxiu; Du, Xiaohui; Li, Junlai

doi:10.1109/access.2022.3156096

Cited by 24 publications

(6 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Improved ultrasound picture interpretation and quicker processing are the main reasons. Ultrasonography, FNA, and thyroid surgery now utilize deep learning (DL) and machine learning (ML) to classify thyroid nodules automatically [16] [19].…”

Section: Related and Recent Workmentioning

confidence: 99%

Enhancing Thyroid Cancer Diagnostics Through Hybrid Machine Learning and Metabolomics Approaches

Raj

2024

IJACSA

View full text Add to dashboard Cite

Thyroid cancer, a prevalent endocrine malignancy, necessitates advanced diagnostic techniques for accurate and early detection. This study introduces an innovative approach that integrates hybrid Machine Learning (ML) algorithms with metabolomics, offering a novel pathway in thyroid cancer diagnostics. Our methodology employs a range of hybrid ML models, combining the strengths of various algorithms to analyze complex metabolomic data effectively. These models include ensemble methods, neural network-based hybrids, and integrations of unsupervised and supervised learning techniques, tailored to decipher the intricate patterns within metabolic profiles associated with thyroid cancer. The study demonstrates how these hybrid ML algorithms can efficiently process and interpret metabolomic data, leading to enhanced diagnostic accuracy. By leveraging the distinct characteristics of each ML model, our approach not only improves the detection of thyroid cancer but also contributes to a deeper understanding of its metabolic underpinnings. The findings of this study pave the way for more personalized and precise medical interventions in thyroid cancer management, showcasing the potential of hybrid ML models in revolutionizing cancer diagnostics. Our system analyzes thyroid cancer metabolomic data using ensemble methods, neural network-based hybrids, and unsupervised and supervised learning integrations. The research shows hybrid ML models may revolutionize cancer diagnoses by improving accuracy. LSTM+CNN, LSTM+GRU, and CNN+GRU have high accuracy rates, helping us comprehend thyroid cancer's biochemical roots. Hybrid ML models enhance thyroid cancer diagnosis and management, enabling more tailored and accurate medical treatments. The hybrid machine learning models like LSTM+CNN, LSTM+GRU, and CNN+GRU beat CNN, VGG-19, Inception-ResNet-v2, decision support, and random forests (99.45%).

show abstract

Section: Related and Recent Workmentioning

confidence: 99%

Enhancing Thyroid Cancer Diagnostics Through Hybrid Machine Learning and Metabolomics Approaches

Raj

2024

IJACSA

View full text Add to dashboard Cite

show abstract

“…To obtain a feature model of the nodule in pictures, Zhao et al [1] created an automated thyroid ultrasound nodule diagnostic method based on CNN. The first step is to collect both positive and negative samples, normalise the images, then segment the nodule region to build an ultrasound thyroid nodule dataset.…”

Section: Literature Surveymentioning

confidence: 99%

“…Up to 68% of symptomatic individuals in the general population have thyroid nodules. 7-15% of thyroid nodules are affected by thyroid cancer, which is the disease with the highest rate of growth across all populations [1]. The thyroid gland, an endocrine gland, secretes thyroid hormones.…”

Section: Introductionmentioning

confidence: 99%

Classification of Thyroid Nodules Using Follow the Regularized Leader Optimization Based Deep Neural Networks

Beyyala,

Priya,

Choudari

et al. 2023

RIA

View full text Add to dashboard Cite

Thyroid cancer, the most common endocrine malignancy, has exhibited a steadily increasing incidence worldwide. Recently, ultrasound (US) imaging has been recognized as a potential diagnostic tool for early assessment of thyroid nodules. However, visual interpretation of nodules is subject to the radiologists' subjective evaluations. To address this, a computer-aided Diagnostic (CAD) system has been developed to differentiate between benign and malignant thyroid nodules. The efficiency of this nodule classifier is heavily dependent on the features employed in the classification process. In this study, the efficacy of the RREMI-RF approach, employing the Multi-Channel Convolutional Neural Network (MCNN) and Hybrid Feature Cropping Network (HFCN) techniques, was evaluated. An innovative Follow the Regularised Leader-based Deep Neural Network (FTRL-DNN) technique is proposed for the precise classification of thyroid nodules. In this method, the AlexNet learning-based feature extraction system was utilized to extract features during the classification process. Images from the Digital Database Thyroid Image (DDTI) dataset were classified using the Long Short Term Memory (LSTM) classifier. Performance metrics, including accuracy, sensitivity, specificity, precision, and error rate, were used to assess the effectiveness of the FTRL-DNN algorithm. Compared to the HFCN and MCNN, the FTRL-DNN-based thyroid nodule classification demonstrated superior accuracy, achieving a rate of 98.94%. This research presents a significant advancement in the automated classification of thyroid nodules, potentially improving early detection and treatment of thyroid cancer.

show abstract

“…The model used a convolutional neural network (CNN) as a feature extractor and a long short-term memory (LSTM) as a classifier. Zhao et al [ 6 ] proposed a method that combines image texture features with CNN-extracted features for thyroid classification using ultrasound images. Rehman et al [ 7 ] utilized the U-Net model to segment thyroid ultrasound images.…”

Section: Introductionmentioning

confidence: 99%

A Framework for Detecting Thyroid Cancer from Ultrasound and Histopathological Images Using Deep Learning, Meta-Heuristics, and MCDM Algorithms

Sharma,

Mahanti,

Panda

et al. 2023

J. Imaging

View full text Add to dashboard Cite

Computer-assisted diagnostic systems have been developed to aid doctors in diagnosing thyroid-related abnormalities. The aim of this research is to improve the diagnosis accuracy of thyroid abnormality detection models that can be utilized to alleviate undue pressure on healthcare professionals. In this research, we proposed deep learning, metaheuristics, and a MCDM algorithms-based framework to detect thyroid-related abnormalities from ultrasound and histopathological images. The proposed method uses three recently developed deep learning techniques (DeiT, Swin Transformer, and Mixer-MLP) to extract features from the thyroid image datasets. The feature extraction techniques are based on the Image Transformer and MLP models. There is a large number of redundant features that can overfit the classifiers and reduce the generalization capabilities of the classifiers. In order to avoid the overfitting problem, six feature transformation techniques (PCA, TSVD, FastICA, ISOMAP, LLE, and UMP) are analyzed to reduce the dimensionality of the data. There are five different classifiers (LR, NB, SVC, KNN, and RF) evaluated using the 5-fold stratified cross-validation technique on the transformed dataset. Both datasets exhibit large class imbalances and hence, the stratified cross-validation technique is used to evaluate the performance. The MEREC-TOPSIS MCDM technique is used for ranking the evaluated models at different analysis stages. In the first stage, the best feature extraction and classification techniques are chosen, whereas, in the second stage, the best dimensionality reduction method is evaluated in wrapper feature selection mode. Two best-ranked models are further selected for the weighted average ensemble learning and features selection using the recently proposed meta-heuristics FOX-optimization algorithm. The PCA+FOX optimization-based feature selection + random forest model achieved the highest TOPSIS score and performed exceptionally well with an accuracy of 99.13%, F2-score of 98.82%, and AUC-ROC score of 99.13% on the ultrasound dataset. Similarly, the model achieved an accuracy score of 90.65%, an F2-score of 92.01%, and an AUC-ROC score of 95.48% on the histopathological dataset. This study exploits the combination novelty of different algorithms in order to improve the thyroid cancer diagnosis capabilities. This proposed framework outperforms the current state-of-the-art diagnostic methods for thyroid-related abnormalities in ultrasound and histopathological datasets and can significantly aid medical professionals by reducing the excessive burden on the medical fraternity.

show abstract

Automatic Thyroid Ultrasound Image Classification Using Feature Fusion Network

Cited by 24 publications

References 28 publications

Enhancing Thyroid Cancer Diagnostics Through Hybrid Machine Learning and Metabolomics Approaches

Enhancing Thyroid Cancer Diagnostics Through Hybrid Machine Learning and Metabolomics Approaches

Classification of Thyroid Nodules Using Follow the Regularized Leader Optimization Based Deep Neural Networks

A Framework for Detecting Thyroid Cancer from Ultrasound and Histopathological Images Using Deep Learning, Meta-Heuristics, and MCDM Algorithms

Contact Info

Product

Resources

About