I have two images of the same person that underwent sclerotherapy for periphery blood vessel aesthetic treatment. Although the images clearly demonstrate that the treatment works, I need to quantify the total length /or area covered by the blood vessel before and after treatment. How can I perform this? By pixel area coverage, or manually by length?
Perhaps FracLac for ImageJ. You might find success with the local connected fractal dimension method, lacunarity, and box counting dimension. Here is a link to an article on retinopathy, which I assume is related to the level of inspection:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2698675/
You might also try fractal analysis to describe the nature of the patterns. I've attached an image comparing the patterns extracted from the image you uploaded and their respective fractal dimensions. Of course, the images I used were of very low quality and I extracted a binary filled pattern whereas you would probably want something more well-traced, so I can't attest to the validity of the result - its just to show what you get with the general method. In general, fractal dimension increases with pathological state.
A straightforward solution to your problem might be to transform your image into HSI (HSV, or HSB) space and then to segment vessel area manually based on Hue information:
remove the skin-color area first (maybe you can use "color threshold"), and then estimate the area of the region covered by vessels.
You may also calculate the average/min/max hue value of the vessel region to quantify the similarity between vessel region and normal skin.
Both ImageJ and PhotoShop are applicable to this case.
Does anyone knows how to apply the fractal analysis to this? Maybe a tutorial?
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Medical Pharmacology
- Clinical Pharmacology