Automatic Segmentation of Ultrasound Images?

I would like to write a code that would be able to automatically go through multiple ultrasound images and segment out tendons. I am not great with coding or matlab and have never worked with image analysis. What would be the best method to go about starting to do this?

Sample image below, I would like to segment out what is circled in red.

[Mathieu NOE]

hello Christina

how would you define the criteria to select that specific area ? based on coordinates (so like cropping the original image) or based on color / brightness / contrast ... other ??

[Christina]

The location I would like to segment out depends on anatomy and placement and will vary geometry wise. I was hoping to be able to segment based off of some sort of combination of color and contrast. However, I am finding it difficult since those vary a lot in the area I am looking at so I am unsure of whether a program would be able to do this automatically.

[KALYAN ACHARJYA]

How to define ROI for those who do not know about anatomy? Are there any criteria, such as border, color, or any other features that can be considered in image analysis? May be their certain parameters can be descibe the probable region of interest.

• Write one by one on plain paper.
• Verify for other images
• Modify....
• Repeat ...

Obtain visualization or image understanding features to extract potential area of interest.

Or you can look for different image clustering algorithms.

Answer 1

[Image Analyst]

It's so noisy and ill-defined that I see deep learning using SegNet as your only hope. Otherwise look to the literature:

USC Annotated Computer Vision Bibliography