I use pyradiomics in CT imaging for extracting features but I'm not sure how to use it in radiography imaging. Is it reliable as CT imaging ? Should I change any parameters ?
Conventional X-ray images (radiography) are defiantly less robust, compared to the CT. They are not calibrated with respect to the HU (like CT), so you should normalize them yourself, plus correct for different nonuniformities in the image. First, make sure that all of them were aquifer with the same protocol (exposure, kv, clinically task), look into difference between different scanners, etc. There were a lot of talks about use of radiomics for chest X-rays at the RSNA 2019, checking abstract or videos of the presentations might help. Good luck!