How Well Do Self-Supervised Models Transfer to Medical Imaging?

Self-supervised learning approaches have seen success transferring between similar medical imaging datasets, however there has been no large scale attempt to compare the transferability of self-supervised models against each other on medical images. In this study, we compare the generalisability of seven self-supervised models, two of which were trained in-domain, against supervised baselines across eight different medical datasets. We find that ImageNet pretrained self-supervised models are more generalisable than their supervised counterparts, scoring up to 10% better on medical classification tasks. The two in-domain pretrained models outperformed other models by over 20% on in-domain tasks, however they suffered significant loss of accuracy on all other tasks. Our investigation of the feature representations suggests that this trend may be due to the models learning to focus too heavily on specific areas.

Effect of COVID-19 Vaccines on Thyroid Function and Autoimmunity and Effect of Thyroid Autoimmunity on Antibody Response.

CONTEXT: There are concerns for COVID-19 vaccination in triggering thyroid autoimmunity and causing thyroid dysfunction. Also, data on the effect of preexisting thyroid autoimmunity on the efficacy of COVID-19 vaccination are limited. OBJECTIVES: We evaluated the effect of COVID-19 vaccination on thyroid function and antibodies, and the influence of preexisting thyroid autoimmunity on neutralizing antibody (NAb) responses. METHODS: Adults without a history of COVID-19/thyroid disorders who received the COVID-19 vaccination during June to August 2021 were recruited. All received 2 doses of vaccines. Thyrotropin (TSH), free thyroxine (fT4), free 3,5,3'-triiodothyronine (fT3), antithyroid peroxidase (anti-TPO), and antithyroglobulin (anti-Tg) antibodies were measured at baseline and 8 weeks post vaccination. NAb against SARS-CoV-2 receptor-binding domain was measured. RESULTS: A total of 215 individuals were included (129 [60%] BNT162b2; 86 [40%] CoronaVac recipients): mean age 49.6 years, 37.2% men, and 12.1% anti-TPO/Tg positive at baseline. After vaccination, TSH did not change (P = .225), but fT4 slightly increased (from 12.0 ±â€…1.1 to 12.2 ±â€…1.2 pmol/L [from 0.93 ±â€…0.09 to 0.95 ±â€…0.09 ng/dL], P < .001) and fT3 slightly decreased (from 4.1 ±â€…0.4 to 4.0 ±â€…0.4 pmol/L [from 2.67 ±â€…0.26 to 2.60 ±â€…0.26 pg/mL], P < .001). Only 3 patients (1.4%) had abnormal thyroid function post vaccination, none clinically overt. Anti-TPO and anti-Tg titers increased modestly after vaccination (P < .001), without statistically significant changes in anti-TPO/Tg positivity. Changes in thyroid function and antithyroid antibodies were consistent between BNT162b2 and CoronaVac recipients, except for greater anti-TPO titer increase post BNT162b2 (P < .001). NAb responses were similar between individuals with and without preexisting thyroid autoimmunity (P = .855). CONCLUSION: COVID-19 vaccination was associated with a modest increase in antithyroid antibody titers. Anti-TPO increase was greater among BNT162b2 recipients. However, there was no clinically significant thyroid dysfunction post vaccination. NAb responses were not influenced by preexisting thyroid autoimmunity. Our results provide important reassurance for people to receive the COVID-19 vaccination.