|
Category | Merits | Demerits |
|
Edge-based segmentation methods | Works well when an edge is prominent | Sensitivity to noise |
Reduces overall contrast in mammograms |
Easy to find locally edge orientation |
Produce unsatisfactory results when it detects fake and weak edges in mammograms |
Not suitable for mammogram images having smooth edges |
Threshold-based segmentation methods | Simple and easy to implement | It is not applicable if the tumour area ratio is unknown |
Sensitive to noise in mammograms |
Faster |
Gives poor results when mammograms have low contrast |
Inexpensive |
Difficulties to fix the threshold value if the number of regions increases |
Not easy to process the mammogram whose histograms are nearly unimodal |
Region-based segmentation | Connected regions are guaranteed | Causes over segmentation if mammograms are noisy |
Multiple criterion and gives good results with less noise | Cannot distinguish the shading of the real mammograms |
Time consuming due to the high resolution of mammograms |
Not suitable for noisy mammograms |
Seed point must be selected |
|
Unsupervised machine learning methods | Few data are required | Number of clusters must be defined |
Easy to implement |
Prior information required |
Automatic segment masses |
Supervised machine learning methods | Easy to detect error | Knowledge about the mammogram to be segmented is required |
Require lab data |
|
Deep learning methods | Solve complex tasks | Limited annotated data |
Required unlabeled data | Time consuming during training |
Expensive because it requires higher computational machines |
Produce accurate results |
|