In the experiments briefly presented in this abstract, we compare the performance of a generalist Swedish pre-trained languagemodel with a domain-specific Swedish pre-trained model on the downstream task of focused terminology extraction of implantterms, which are terms that indicate the presence of implants in the body of patients. The fine-tuning is identical for bothmodels. For the search strategy we rely on KD-Tree that we feed with two different lists of term seeds, one with noise and onewithout noise. Results shows that the use of a domain-specific pre-trained language model has a positive impact on focusedterminology extraction only when using term seeds without noise.