1000/1000
Hot
Most Recent
Cervical cancer is regularly diagnosed in women all over the world. This cancer is the seventh most frequent cancer globally and the fourth most prevalent cancer among women. Automated and higher accuracy of cervical cancer classification methods are needed for the early diagnosis of cancer. In addition, routine Pap smears could enhance clinical outcomes by facilitating the early diagnosis of cervical cancer. Liquid-based cytology (LBC)/Pap smears for advanced cervical screening is a highly effective precancerous cell detection technology based on cell image analysis, where cells are classed as normal or abnormal. Computer-aided systems in medical imaging have benefited greatly from extraordinary developments in artificial intelligence (AI) technology.
Title | Class and Database | Methodology | Results |
---|---|---|---|
CVM-Cervix: A hybrid cervical Pap smear image classification framework using CNN, visual transformer, and multilayer perceptron [18] | Database: CRIC dataset, SIPaKMeD dataset and combination of CRIC and SIPaKMeD Class: CRIC-6 class, SIPaKMeD 5 Class |
Framework CVM-Cervix based on deep learning. Type: Machine learning—Neural network |
Effective and potential of the proposed CVM-Cervix proved. |
A Comparative Analysis of Deep Learning Models for Automated Cross-Preparation Diagnosis of Multi-Cell Liquid Pap Smear Images [19] | Database: Thin Prep Pap dataset Class: 4 Bethesda class |
An ensemble novel convolutional neural network (CNN) and a CNN with autoencoder (AE). Type: Machine learning—Neural network |
All models’ accuracy >905. The individual transfer model had high variability in performance, while CNN and AE CNN did not. ResNet101 accuracy is 92.65%. |
Multi-class nucleus detection and classification using a deep convolutional neural network with enhanced high dimensional dissimilarity translation model on cervical cells [20] | Database: Herlev dataset, SIPaKMeD dataset, CRIC dataset Class: Herlev-7 class, SIPaKMeD-5 Class, CRIC-6 class |
Segmentation—hybrid system that incorporates two binary image patches obtained by a 19-layered convolutional neural network (ConvNet) model with an enhanced deep high dimensional dissimilarity translation (HDDT). A Pre trained Resnet-50 model. T-distribution stochastic neighbour embedding (t-SNE) for down-sampled. Classification using a multi-class weighted kernel extreme learning machine (WKELM) classifier via a sparse multicanonical correlation (SMCCA) method. Type: Machine learning—Neural network |
Accuracy 99.12% Specificity 99.45% Sensitivity 99.25% Execution time 99.6248 s The proposed model is more effective compared to existing approaches. |
Hybrid Loss-Constrained Lightweight Convolutional Neural Networks for Cervical Cell Classification [21] | Database: Herlev dataset Class: Herlev-7 class |
Classification using hybrid loss function with label smoothing to improve the distinguishing power of lightweight convolutional neural networks (CNN) Type: Machine learning—Neural network |
ShufflenetV2 results:- Accuracy 96.18% Precision 96.30% Recall 96.23% Specificity 99.08% GhostnetV2 results:- Accuracy 96.39% Precision 96.42% Recall 96.39% Specificity 99.09% |
Detection of cervical cells based on improved SSD network [22] | Database: Herlev dataset Class: Herlev-7 class |
Integration of Single Shot MultiBox Detector with the positive and negative features to address the problem of insufficient sensitivity for small objects. Type: Machine learning—Neural network |
Accuracy 90.80% Mean average precision (mAP) is 81.53%, which is 7.54% and 4.92% higher than YOLO and classical SSD. |
Detection of cervical cancer cells in a complex situation based on improved YOLOv3 network [1] | Database: Herlev dataset Class: Herlev-7 class |
Detection using the YOLO algorithm. Feature extraction generalization by adding the dense block and S3Pool algorithm on the basis of the feature extraction network DarkNet-53. Clustering algorithm of improved algorithm k-means++. Type: Machine learning—Neural network |
Mean average precision (mAP) is 78.87%, which is 7.54% and 4.92% higher than YOLO (You Only Look Once) and classical SSD |
Pap smear-based cervical cancer detection using residual neural networks deep learning architecture [23] | Database: Mendeley LBC SIPaKMeD dataset Class: Mendeley LBC SIPaKMeD—4 class |
Data augmentation module of DTWCT module and convolutional neural networks (CNN). Classification using ResNet 18, which defines four classes of sources for Pap smear cell images. Type: Machine learning—Neural network |
Average Pap smear detection index (PDI) is 99%. |
Cervical cell multi-classification algorithm using global context information and attention mechanism [3] | Database: SIPaKMeD dataset Class: SIPaKMeD—5 class |
Convolutional neural network (L-PCNN) that integrates global context information and attention mechanism. Improved ResNet-50 backbone network for feature extraction. Type: Machine learning—Neural network |
Accuracy 98.89%. Sensitivity 99.9%. Specificity 99.8%. F-measure 99.89%. |
DeepCyto: a hybrid framework for cervical cancer classification using deep feature fusion of cytology images [24] | Database: Herlev dataset, SIPaKMeD dataset, LBC dataset Class: Herlev-7 class, SIPaKMeD-5 class, LBC-4 class |
Novel classification using DeepCyto. Principal component analysis and machine learning ensemble for classification of Pap smear images. Artificial neural network with feature fusion vectors as an input for classification. Type: Machine learning—Neural network |
DeepCyto is a powerful tool for precise feature extraction and Pap smear image classification. |
Classification of Cervical Cytology Overlapping Cell Images with Transfer Learning Architectures [25] | Database: Cervix93 cervical cytology image Class: 3 class |
Transfer learning using deep learning convolutional neural network. Cutting edge pretrained networks: AlexNet, ImageNet, and Places 365. Type: Machine learning—Neural network |
Accuracy 99.03%. Kappa coefficient showing perfect agreement. AlexNet proved a successful assistive tool for cervical cancer detection. |
Optimal deep convolution neural network for cervical cancer diagnosis model [8] | Database: Herlev dataset Class: Herlev-7 class |
Detection using intelligent deep convolutional neural network. Classification (IDCNN-CDC) model using biomedical Pap smear images. Noise removal using Gaussian Filter. Segmentation using the Tsallis entropy technique with the dragonfly optimization. Deep learned feature using SqueezeNet. Classification using weighted extreme learning machine (ELM). Type: Machine learning—Neural network |
Higher performance of the proposed technique in terms of sensitivity, specificity, accuracy, and F-Score. |
Modified metaheuristics with stacked sparse denoising autoencoder model for cervical cancer classification [9]. | Database: Herlev dataset Class: Herlev-7 class |
Novel Modified Firefly Optimization Algorithm with Deep Learning-enabled cervical cancer classification (MFFOA-DL3) model. Noise removal using Bilateral Filtering (BF)-based. Segmentation technique of Kapur’s entropy-based image to define affected area. Generate feature vectors using EfficientNet. Classification of the cell using MFFOA with Stacked Sparse Denoising Autoencoder (SSDA) model. Type: Machine learning—Neural network |
The findings of a comprehensive comparison investigation revealed that the MFFOA-DL3 model outperformed other recent approaches. |
Imaging based cervical cancer diagnostics using small object detection—generative adversarial networks [5] | Database: Herlev dataset, Colposcopy images, Clinical references Class: not applicable |
An effective hybrid deep learning technique using Small-Object Detection-Generative Adversarial Networks (SOD-GAN) with Fine-tuned Stacked Autoencoder (F-SAE). Generation and discrimination of the cervical cell using Region-based Convolutional Neural Network (RCNN). Type: Machine learning—Neural network |
The proposed method identifies and classifies cervical premalignant and malignant diseases based on deep characteristics without the necessity for initial classification and segmentation. |
Cervical cancer diagnosis based on modified uniform local ternary patterns and feed forward multilayer network optimized by genetic algorithm [26] | Database: Herlev dataset Class: Herlev-7 class |
Segmentation of the image using a thresholding approach. Feature extraction by applying a texture descriptor titled modified uniform local ternary patterns (MULTP). Classification of the cell using an optimized multilayer feed-forward neural network. Type: Machine learning—Neural network |
MULTP, the proposed texture descriptor, is a generic operator that may be used to characterise texture features of images in numerous computer vision issues. In addition, the suggested optimization approach may be utilised to increase performance in deep networks. |
Early cervical cancer diagnosis using Sooty tern-optimized CNN-LSTM classifier [11] | Database: Herlev dataset Class: Herlev-7 class |
Augmentation process of image enhancement, image flipping, and image rotating to reduce the number of parameters necessary. Segmentation of the cancer-affected regions with the help of kernel weighted fuzzy local information c-means clustering (KWFLICM) model. Classification using the Sooty Tern Optimization (STO) algorithm with CNN-based long short-term memory classifier (CNN-LSTM). Type: Machine learning—Neural network |
Accuracy 99.80%. Specificity 99%. Sensitivity 98.83%. F-Score 97.8. Improvement of 28.5% better than Random Forest and 19.46% better than ensemble classifier. |
Hybrid Model for Detection of Cervical Cancer Using Causal Analysis and Machine Learning Techniques [10] | Database: Private Class: not applicable |
Boruta analysis and SVM method for an efficient feature selection and prediction of the model for the cervical cell dataset. Type: Machine learning—Linear model |
Boruta analysis shows a better performance approach compared to the existing techniques available. |
Cervical Cancer Classification Using Combined Machine Learning and Deep Learning Approach [27] | Database: Herlev dataset Class: Herlev-2 class |
Feature extraction using ResNet-101. Classification using Support vector Machine (SVM). Type: Machine learning—Linear model |
Accuracy 97.30%. |
Auxiliary classification of cervical cells based on the multi-domain hybrid deep learning framework [17] | Database: Herlev dataset, SIPaKMeD dataset, BJTU dataset Class: Herlev-2&7 class, SIPaKMeD-5 Class, BJTU-7 class |
Deep features extraction using deep Convolutional Neural Network of pretrained Visual Geometry Group-19 (VGG-19). Hand-crafted images undergo the process of feature selection, clustering, and dimensionality reduction. Classification using a Support Vector Machine (SVM) classifier. Type: Machine learning-Linear model |
Accuracy 98.70%. Sensitivity 98.20%. Specificity 98.90%. The suggested novel screening methodology is promising for early cervical cancer detection, with multi-domain and hybrid characteristics proving realistic in clinical practise. |
An Evaluation of Computational Learning-based Methods for the Segmentation of Nuclei in Cervical Cancer Cells from Microscopic Images [28] | Database: Z-Stack cellular microscopy proliferation images provided by the HCS Pharma Class: not applicable |
Machine learning architecture of Random Forest, Ada Boost, and MLP algorithm. Type: Machine learning—Nonlinear model |
All machine learning architectures gave outstanding nuclei segmentation in cervical cancer cells but did not solve the overlapping nuclei and Z-stack segmentation problems. |
Prognosis of Cervical Cancer Disease by Applying Machine Learning Techniques [4] | Database: Dataset of 858 cervical cancer patients with 36 risk factors and one outcome variable Class: not applicable |
Analysis of the different supervised machine learning techniques. The classification algorithm used Artificial Neural Network, Bayesian Network, SVM, Random Tree, Logistic Tree and XG-Boost Tree. Selection algorithm for feature selection: relief rank, wrapper method, and LASSO regression. Type: Machine learning—Nonlinear model |
Maximum accuracy achieved using XG-Boost with complete features 94.94%. This approach offers much potential for clinical use and cervical cancer cell detection. |
Is the aspect ratio of cells important in deep learning? A robust comparison of deep learning methods for multi-scale cytopathology cell image classification: From convolutional neural networks to visual transformers [29] | Database-SIPaKMeD dataset Class: not applicable |
Twenty-two deep learning models were used to classify the cervical cancer cells into two categories of standard and scaled datasets. Type: Machine learning—Nonlinear model |
Deep learning models are robust to changes in the aspect ratio of cervical cells in cervical cytopathological images. |
A Fast Hybrid Classification Algorithm with Feature Reduction for Medical Images [30] | Database: Herlev dataset Class: Herlev-7 class |
A novel fast hybrid fuzzy classification algorithm with feature reduction for medical images. Integration of quantum-based grasshopper computing algorithm (QGH) with a fuzzy clustering technique for feature extraction. The second integration of the fusion technique utilises QGH with the fuzzy c-means algorithm to determine the best features. Type: Machine learning—Nonlinear model |
Established the importance of the feature selection on the accuracy of the proposed classifier |
Cervical Cancer Classification from Pap Smear Images Using Modified Fuzzy C Means, PCA, and KNN [7] | Database: Herlev dataset Class: Herlev-7 class |
Geometrical and feature extraction using a novel approach of modified fuzzy c-means. Augmentation of the images using Principal Component Analysis (PCA) to maintain the uncorrelated features and thus reduce the algorithm processing time. Classification of the Pap smear image into normal and abnormal cells using K Nearest Neighbour (KNN). Type: Machine learning—Nonlinear model |
Minimum accuracy 94.15%. Maximum accuracy 96.28%. Average accuracy 94.86%. Sensitivity 97.96%. Specificity 83.65%. F1-Score 96.87%. Precision 96.31%. |
A Semi-supervised Deep Learning Method for Cervical Cell Classification [31] | Database: Herlev dataset, SIPaKMeD dataset Class: Herlev-7 class, SIPaKMeD-5 Class |
A novel manual features and voting mechanism to achieve data expansion in semi-supervised learning. Clarity function to filter out higher-quality images, annotating a small amount of the high-quality images, and voting mechanism for balancing and training data. Type: Machine learning—Classifier |
Accuracy 91.94%. |