Detecting Skin Lesions in Dermoscopy Images
Author Information
Author(s): Mete Mutlu, Sirakov Nikolay Metodiev
Primary Institution: Texas A&M University–Commerce
Hypothesis
Can automated algorithms improve the detection of skin lesions in dermoscopy images?
Conclusion
The study successfully proposed two frameworks for delineating suspicious skin lesions, achieving an average border error of 6.6%.
Supporting Evidence
- ACM and BD-DBSCAN achieved the same average border error of 6.6%.
- ACM outperformed BD-DBSCAN in 38 out of 50 cases.
- BD-DBSCAN showed a lower standard deviation in border errors compared to ACM.
Takeaway
The researchers created two computer programs to help doctors find skin problems in pictures, and they worked really well.
Methodology
The study used Active Contour Model (ACM) and boundary-driven density-based clustering (BD-DBSCAN) algorithms on dermoscopy images.
Potential Biases
The study did not address potential biases in the selection of images or the manual delineation process.
Limitations
The algorithms may be sensitive to noise in images, and the t-test for comparison was not performed due to non-normal distribution of errors.
Participant Demographics
The images included benign lesions such as nevocellular nevi and dysplastic nevi.
Statistical Information
P-Value
0.001
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website