Table 10. Comparative analysis.
Summary of the previously published survey in the blood cell image segmentation and classification domain.
| Title | Focus area | Techniques analyzed | Challenges highlighted |
|---|---|---|---|
| Das et al. (2022) | WBC | SVM, KNN, ANN, CNN, RNN | Publicly accessible datasets, generalization, complexity |
| Khan et al. (2021) | WBC classification | Traditional machine learning (TML) and deep learning (DL) methods | Need for lightweight TML and DL techniques, transitioning from supervised to unsupervised learning |
| SivaRao & Rao (2023) | WBC classification | SegNet EfficientNet, and XGBoost | The conventional method is time-consuming, laborious, and potentially erroneous. |
| Bhatia et al. (2023) | RBC classification in sickle cell disease | Customized-DCNN, SHAP, and LIME for interpretability | Small unbalanced dataset, overlapping or clustered RBCs in some images |
| Bhargavi et al. (2023) | White blood cell classification | Deep-CNNs, decision tree | Traditional methods are time-consuming and less accurate |
| Anilkumar, Manoj & Sagi (2023) | Computer-aided diagnosis of leukemia | SVM, CNN, k-NN, naïve Bayes, ensemble classifiers | Lack of public datasets for chronic leukemia, Intra-observer and inter-observer variability |
| Thomas & Sreejith (2018) | White blood cells segmentation | K-means clustering, otsu thresholding, color-based segmentation | Complexity and uncertainty in microscopic blood smear images make WBCs segmentation challenging |
| Byndur et al. (2023) | Segmentation and classification WBCs | Staining techniques, datasets, preprocessing techniques, Otsu’s method, K-means clustering, CNN, k-NN | Manual identification of WBCs is prone to errors, complexity in microscopic blood smear images |
| Raina et al. (2023) | Acute leukemia detection using deep learning | Preprocessing techniques like resizing, normalization, histogram equalization | Lack of publicly available datasets, complexity in microscopic blood smear images, variability in diagnosis depending on hematologist's experience. |
| Rao & Rao (2023) | WBC segmentation and classification | Pyramid scene parsing network MobilenetV3, artificial gravitational cuckoo search, ShufflenetV2 | Challenges in preprocessing include noise, occlusion, and missing data. |
| Umamaheswari & Geetha (2019) | Machine learning in leukemia detection | Otsu's method, automatic thresholding, Watershed Algorithm, k-means clustering | Lack of standard datasets for leukemia detection, Complexity in microscopic blood smear images |
| More & Sugandhi (2023) | Leukemia detection using machine learning | Noise removal, contrast adjustment, extraction methods for shape, texture, forest, naive bayes, SVM, and logistic regression. | Complexity in microscopic blood smear images, Challenges in preprocessing like noise, occlusion, missing data. |
| Al-Dulaimi & Makki (2023) | Blood cell detection and classification in CAD systems | Preprocessing techniques like noise removal, contrast adjustment. Segmentation methods like Fuzzy C means, K-means clustering, and thresholding. | Different staining techniques affecting segmentation, Non-uniform illumination, Variation in cell maturity stages, And morphology complexities. |
| Asghar et al. (2023) | Medical image analysis for white blood cell classification | Preprocessing techniques, feature extraction, CNN, R-CNN, Fast R-CNN, GAN. | Availability of appropriate datasets, medical training of researchers for better understanding of WBC structure |
| Veeraiah, Alotaibi & Subahi (2023) | Medical image analysis for leukemia detection | Histogram threshold segmentation classifier (HTsC) | Overburdening of pathologists with large data sets, variations in illumination and staining in manual setups |
| Proposed Survey | Target both WBC and RBC images | Machine learning and deep learning | Multidimensional research question and future directions, target audience |