Table 2. Comparative analysis of segmentation techniques and their performance used for WBC image analysis.
A summary of various segmentation methods used by different authors, along with the corresponding performance accuracy achieved and the evaluation parameters used to measure the effectiveness of each technique.
| Author | Segmentation | Performance accuracy | Evaluation parameter |
|---|---|---|---|
| Sharma & Buksh (2019) | Hybrid segmentation using ALL-DC model | 98.02%, 70.07%, and 86.2% | Precision, recall, and F1-measure |
| Al-jaboriy et al. (2019) | ROI-based segmentation using local pixel information | 97% | Accuracy |
| Abdulhay et al. (2018) | ROI and edge detection | 95.3% and 91.66% | Accuracy, specificity, and sensitivity |
| Shahin et al. (2018) | Otsu’s thresholding | 97.6% | Performance measurement |
| Quiñones et al. (2018) | Zak algorithm | 98.88% | Counting accuracy |
| Cao, Liu & Song (2018) | SWAM&IVFS, fuzzy divergence based | 93.75% | Accuracy |
| Miao & Xiao (2018) | Marker-controlled watershed | 97.2% and 94.8% | Over/under-segmentation and fault rate |
| Nikitaev et al. (2018) | Modified watersheds/distance transformation | 82% | Accuracy |
| Di Ruberto, Loddo & Putzu (2016) | Segmentation Via SVM | 99.73% | Counting accuracy |
| Liu et al. (2015) | Nucleus mark watershed, mean shift clustering | 99%, 95.5%, and 97% | Precision, recall, and F1 score |
| Tosta et al. (2015) | Neighborhood valley-emphasis | 89.89% and 99.75% | Jaccard’s similarity coefficient and accuracy |
| Shirazi et al. (2016b) | Thresholding and mathematical morphology | 96.15% | Accuracy |
| Alomari et al. (2014) | Thresholding | 98.4% | Precision, recall, and F-measurements |