Pancreatic Cancer Survival Prediction: A Survey of the State-of-the-Art

Table 2

Advantages and disadvantages of different segmentation techniques.

Segmentation technique	Advantage	Disadvantage
Region-based [16, 23, 24]	(i) High speed of operation; (ii) Efficient for object and background with high contrast; (iii) Easier to classify and implement; (iv) Best when easy to define region similarities; (v) Less sensitive to noise compared to edge detection	(i) Poor segments if there is low greyscale; (ii) Are by nature sequential and quite expensive both in computational time and memory; (iii) Region growing has inherent dependence on the selection of seed region and the order in which pixels and regions are examined
Fuzzy theory-based [25–27]	(i) The single fuzzy rule applied to stress the importance attached to feature-based and spatial information in the image; (ii) Structure of the membership functions and associated parameters automatically derived	(i) Sensitive to noise; (ii) Computationally expensive; (iii) The determination of fuzzy membership is not very easy
Artificial network-based [15]	(i) Simple programming; (ii) Make use of neural net parallelism	(i) Long training time; (ii) Initialization could influence the outcome
Generalized PCA (principle component analysis) [28, 29]	(i) Low noise sensitivity; (ii) Lack of redundancy of data; (iii) Increased efficiency; (iv) Reduced overfitting	(i) Independent variables become less interpretable; (ii) Data standardization is a must before PCA; (iii) Information loss