We propose a Deep learning-based weak label learning method for analyzing whole slide images (WSIs) of Hematoxylin and Eosin (H&E) stained tumor tissue not requiring pixel-level or tile-level annotations using Self-supervised pre-training and heterogeneity-aware deep Multiple Instance LEarning (DeepSMILE). We apply DeepSMILE to the task of Homologous recombination deficiency (HRD) and microsatellite instability (MSI) prediction. We utilize contrastive self-supervised learning to pre-train a feature extractor on histopathology tiles of cancer tissue. Additionally, we use variability-aware deep multiple instance learning to learn the tile feature aggregation function while modeling tumor heterogeneity. For MSI prediction in a tumor-annotated and color normalized subset of TCGA-CRC (n=360 patients), contrastive self-supervised learning improves the tile supervision baseline from 0.77 to 0.87 AUROC, on par with our proposed DeepSMILE method. On TCGA-BC (n=1041 patients) without any manual annotations, DeepSMILE improves HRD classification performance from 0.77 to 0.81 AUROC compared to tile supervision with either a self-supervised or ImageNet pre-trained feature extractor. Our proposed methods reach the baseline performance using only 40% of the labeled data on both datasets. These improvements suggest we can use standard self-supervised learning techniques combined with multiple instance learning in the histopathology domain to improve genomic label classification performance with fewer labeled data. 相似文献
PurposeAs protection from COVID-19 following two doses of the BNT162b2 vaccine showed a time dependent waning, a third (booster) dose was administrated. This study aims to compare the antibody response following the third dose versus the second and to evaluate post-booster seroconversion.MethodsA prospective observational study conducted in Maccabi Healthcare Services. Serial SARS-CoV-2 Spike IgG tests, 1,2,3 and 6 months following the second vaccine dose and one month following the third were obtained. Neutralizing antibody levels were measured in a subset of participants. Per individual SARS-CoV-2 Spike IgG titer ratios were calculated one month after the booster administration compared to titers one month following the second dose and prior to booster.ResultsAmong 110 participants, 56 (51%) were women. Mean age was 61.7 ± 1.9 years and 66 (60%) were immunocompromised. One month after third dose, IgG titers were induced 7.83 (95 %CI 5.25–11.67) folds and 2.40 (95 %CI 1.90–3.03) folds compared to one month after the second, in the immunocompromised and immunocompetent groups, respectively. Of the 17 immunocompromised participants who were seronegative after the second dose, 4 (24%) became seropositive following the third. Comparing the titers prior to the third dose, an increase of 50.7 (95 %CI 32.5–79.1) fold in the immunocompromised group and 25.7 (95 %CI 19.1–34.7) fold in and immunocompetent group, was observed.ConclusionA third BNT162b2 vaccine elicited robust humoral response, superior to the response observed following the second, among immunocompetent and immunocompromised individuals. 相似文献