From: A pap-smear analysis tool (PAT) for detection of cervical cancer from pap-smear images
Author | Paper | Datasets | Features | Pre-processing | Segmentation | Classification | Results |
---|---|---|---|---|---|---|---|
Su et al. [19] | Automatic detection of cervical cancer cells by a two-level cascade classification system | Liquid-based cytology slides | 20 Morphological and 8 texture features | Histogram equalization and Median filter | Adaptive threshold | C4.5 and Logical Regression classifiers | Recognition rates of 95.6% achieved |
Sharma et al. [20] | Classification of clinical dataset of cervical cancer using KNN | Single cells data sets from Fortis Hospital, India | 7 morphological features | Gaussian filter and histogram equalization | Min–max and edge detection | K-nearest neighbour | Accuracy of 82.9% with fivefold cross-validation |
Kumar et al. [21] | Detection and classification of cancer from microscopic biopsy images using clinically significant features | Histology image dataset (histology DS2828) | 125 Nucleus and cytoplasm morphologic features | Contrast limited adaptive histogram equalization | K-means segmentation algorithm | K-NN, fuzzy KNN, SVM and random forest-based classifiers | Accuracy, specificity and sensitivity of 92%, 94% and 81% |
Chankong et al. [22] | Automatic cervical cell segmentation and classification in Pap smears | Herlev dataset | Morphological features | Median filter | Patch-based fuzzy C-means and FCM | Fuzzy C-means | Accuracies of 93.78% and 99.27% for 7 and 2-class classifications |
Talukdar et al. [23] | Fuzzy clustering based image segmentation of pap smear images of cervical cancer cell using FCM algorithm | Colour image | Morphometric, densitometry, colorimetric and textural feature | Adaptive histogram equalization with Otsu’s method | Chaos theory corresponding to R, G and B value | Pixel-level classification and shape analysis | Preserves the colour of the images and data loss is minimal |
Sreedevi et al. [24] | Pap smear image-based detection of cervical cancer, | Herlev dataset | Nucleus features | Colour conversions and contrast enhancement | Iterative thresholding method | Based on the area of the nucleus | A sensitivity of 100% and specificity of 90% achieved |
Ampazis et al. [25] | Pap-smear classification using efficient second-order neural network | Herlev University Hospital | 20 morphological features | Contrast enhancement | Neural networks | LMAM and OLMAM algorithms | Classification accuracy of 98.86% was obtained |