Hello,
I would like to compare the ultrasound echo intensity of different patients and research participants, but those images are taken from different ultrasound machines and with different configuration (brightness, contrast etc). Is there a way to compare echo intensity between those studies, or echo intensity values are always relative to images taken within the same sample using the same configuration?
I would highly recommend against doing that if this data is not the raw RF data. Clinical scanners have various post-processing tools that can be specific to the scanners (i.e. log compression). Additionally, things like the time gain compensation, and the placement of the focus could affect the intensity values which are dependent on sonographer taking that data.
If you have access to the RF data, you should be able to beamform it into an image and then look at those intensity values.
You could instead look at the structure of the tissues (texture analysis, fascicle length), and this might be invariant to the ultrasound machine configurations.
Dear Dr. Georgios Mavropalias
I agreed with Dr. David Le . In the the other words, the best way is accessing to RF signals of different ultrasound systems.
However, the configurations of ultrasound systems are predefined approaches and they can be estimated using their definition and statistics of the signal before and signal after applying the configurations. if they were invariant during the imaging, your idea (compare echo-intensity in images with different brightness and contrast) can be done by using the below approach(we supposed that the imaged sample is invariant):
1- Define one or more region of interest(ROI) in ultrasound images.
2. Do rigid-register of ROI from different ultrasound systems using geometric-based rigid-registration methods.
2- extract some applicable features from each ROI in each image. some useful features are: spatial features, co-occurrence texture features, and non-heuristic texture features.
3- Estimate the relationships among features (from different ultrasound system) using nonparametric regression approaches.
I hope it can be useful.
Regards
- Badges
- Science topic
- Similar topics
- Automotive Engineering
- Machines