| 1. |
- Froberg, Gabrielle, et al.
(författare)
-
Towards clinical breakpoints for non-tuberculous mycobacteria-Determination of epidemiological cut off values for the Mycobacterium avium complex and Mycobacterium abscessus using broth microdilution
- 2023
-
Ingår i: Clinical Microbiology and Infection. - : ELSEVIER SCI LTD. - 1198-743X .- 1469-0691. ; 29:6, s. 758-764
-
Tidskriftsartikel (refereegranskat)abstract
- Objective: For non-tuberculous mycobacteria (NTM), minimum inhibitory concentration (MIC) distri-butions of wild-type isolates have not been systematically evaluated despite their importance for establishing antimicrobial susceptibility testing (AST) breakpoints.Methods: We gathered MIC distributions for drugs used against the Mycobacterium avium complex (MAC) and Mycobacterium abscessus (MAB) obtained by commercial broth microdilution (SLOMYCOI and RAPMYCOI) from 12 laboratories. Epidemiological cut-off values (ECOFFs) and tentative ECOFFs (TEC-OFFs) were determined by EUCAST methodology including quality control (QC) strains.Results: The clarithromycin ECOFF was 16 mg/L for M. avium (n = 1271) whereas TECOFFs were 8 mg/L for M. intracellulare (n = 415) and 1 mg/L for MAB (n = 1014) confirmed by analysing MAB subspecies without inducible macrolide resistance (n = 235). For amikacin, the ECOFFs were 64 mg/L for MAC and MAB. For moxifloxacin, the WT spanned >8 mg/L for both MAC and MAB. For linezolid, the ECOFF and TECOFF were 64 mg/L for M. avium and M. intracellulare, respectively. Current CLSI breakpoints for amikacin (16 mg/L), moxifloxacin (1 mg/L) and linezolid (8 mg/L) divided the corresponding WT dis-tributions. For QC M. avium and M. peregrinum, >= 95% of MIC values were well within recommended QC ranges.Conclusion: As a first step towards clinical breakpoints for NTM, (T)ECOFFs were defined for several antimicrobials against MAC and MAB. Broad wild-type MIC distributions indicate a need for further method refinement which is now under development within the EUCAST subcommittee for anti-mycobacterial drug susceptibility testing. In addition, we showed that several CLSI NTM breakpoints are not consistent in relation to the (T)ECOFFs. Gabrielle Froeuroberg, Clin Microbiol Infect 2023;29:758 (c) 2023 The Author(s). Published by Elsevier Ltd on behalf of European Society of Clinical Microbiology and Infectious Diseases. This is an open access article under the CC BY license (http://creativecommons. org/licenses/by/4.0/).
|
|
| 2. |
- Gomariz, Alvaro, et al.
(författare)
-
Unsupervised Domain Adaptation with Contrastive Learning for OCT Segmentation
- 2022
-
Ingår i: Medical Image Computing and Computer Assisted Intervention, MICCAI 2022, pt viii. - Cham : Springer Nature. - 9783031164521 - 9783031164514 ; , s. 351-361
-
Konferensbidrag (refereegranskat)abstract
- Accurate segmentation of retinal fluids in 3D Optical Coherence Tomography images is key for diagnosis and personalized treatment of eye diseases. While deep learning has been successful at this task, trained supervised models often fail for images that do not resemble labeled examples, e.g. for images acquired using different devices. We hereby propose a novel semi-supervised learning framework for segmentation of volumetric images from new unlabeled domains. We jointly use supervised and contrastive learning, also introducing a contrastive pairing scheme that leverages similarity between nearby slices in 3D. In addition, we propose channel-wise aggregation as an alternative to conventional spatial-pooling aggregation for contrastive feature map projection. We evaluate our methods for domain adaptation from a (labeled) source domain to an (unlabeled) target domain, each containing images acquired with different acquisition devices. In the target domain, our method achieves a Dice coefficient 13.8% higher than SimCLR (a state-of-the-art contrastive framework), and leads to results comparable to an upper bound with supervised training in that domain. In the source domain, our model also improves the results by 5.4% Dice, by successfully leveraging information from many unlabeled images.
|
|
