Annotation-efficient training of medical image segmentation network based on scribble guidance in difficult areas

PURPOSE

The training of deep medical image segmentation networks usually requires a large amount of human-annotated data. To alleviate the burden of human labor, many semi- or non-supervised methods have been developed. However, due to the complexity of clinical scenario, insufficient training labels still causes inaccurate segmentation in some difficult local areas such as heterogeneous tumors and fuzzy boundaries.

METHODS

We propose an annotation-efficient training approach, which only requires scribble guidance in the difficult areas. A segmentation network is initially trained with a small amount of fully annotated data and then used to produce pseudo labels for more training data. Human supervisors draw scribbles in the areas of incorrect pseudo labels (i.e., difficult areas), and the scribbles are converted into pseudo label maps using a probability-modulated geodesic transform. To reduce the influence of the potential errors in the pseudo labels, a confidence map of the pseudo labels is generated by jointly considering the pixel-to-scribble geodesic distance and the network output probability. The pseudo labels and confidence maps are iteratively optimized with the update of the network, and the network training is promoted by the pseudo labels and the confidence maps in turn.

RESULTS

Cross-validation based on two data sets (brain tumor MRI and liver tumor CT) showed that our method significantly reduces the annotation time while maintains the segmentation accuracy of difficult areas (e.g., tumors). Using 90 scribble-annotated training images (annotated time: ~ 9 h), our method achieved the same performance as using 45 fully annotated images (annotation time: > 100 h) but required much shorter annotation time.

CONCLUSION

Compared to the conventional full annotation approaches, the proposed method significantly saves the annotation efforts by focusing the human supervisions on the most difficult regions. It provides an annotation-efficient way for training medical image segmentation networks in complex clinical scenario.

What Causes the Discrepancy in SARS-CoV-2 Vaccine Between Parental Hesitancy for Themselves and for Their Children During Lockdown Period?

BACKGROUND

Parents are usually the decision-makers for vaccinations of children. Therefore, it is important to understand parental beliefs and attitudes toward severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccine for themselves and their children when it was approved for children age 3-17.

METHOD

A cross-sectional survey based on an anonymous online questionnaire for parents was conducted in seven provinces of China, and demographic information, vaccination history, parental decision motives, and health belief model toward themselves and their children were collected, respectively.

RESULTS

The overall parental hesitancy rate toward themselves was 20.30%, and that toward their children was 7.80%. More parental concerns on disease severity (odd ratio [OR] = 1.11, 95% confidence interval [CI]: 1.01-1.61) and susceptibility (OR = 1.29, 95% CI: 1.01-1.63) of children could be the causes of discrepancy in hesitancy for themselves and for their children. Parents who hesitated to vaccinate themselves might also be hesitated to vaccinate their children (β = 0.077, P < 0.001).

CONCLUSION

Threat perception may lead to inconsistencies in parental vaccination decisions toward themselves and toward their children. Correcting misinformation and strengthening education about COVID-19 are of great significance in addressing vaccine hesitancy among parents and children.

Cytotype distribution and chloroplast phylogeography of the Actinidia chinensis complex

BACKGROUND

Plant phylogeographic studies of species in subtropical China have mainly focused on rare and endangered species, whereas few studies have been conducted on taxa with relatively wide distribution, especially polyploid species. We investigated the cytotype and haplotype distribution pattern of the Actinidia chinensis complex, a widespread geographically woody liana with variable ploidy in subtropical China comprising two varieties, with three chloroplast fragments DNA (ndhF-rpl132, rps16-trnQ and trnE-trnT). Macroevolutionary, microevolutionary and niche modeling tools were also combined to disentangle the origin and the demographic history of the species or cytotypes.

RESULTS

The ploidy levels of 3338 individuals from 128 populations sampled throughout the species distribution range were estimated with flow cytometry. The widespread cytotypes were diploids followed by tetraploids and hexaploids, whereas triploids and octoploids occurred in a few populations. Thirty-one chloroplast haplotypes were detected. The genetic diversity and genetic structure were found to be high between varieties (or ploidy races) chinensis and deliciosa. Our results revealed that these two varieties inhabit significantly different climatic niche spaces. Ecological niche models (ENMs) indicate that all varieties' ranges contracted during the Last Inter Glacial (LIG), and expanded eastward or northward during the Last Glacial Maximum (LGM).

CONCLUSIONS

Pliocene and Plio-Pleistocene climatic fluctuations and vicariance appear to have played key roles in shaping current population structure and historical demography in the A. chinensis complex. The polyploidization process also appears to have played an important role in the historical demography of the complex through improving their adaptability to environmental changes.

Genome-wide association and differential expression analysis of salt tolerance in Gossypium hirsutum L at the germination stage

BACKGROUND

Salinity is a major abiotic stress seriously hindering crop yield. Development and utilization of tolerant varieties is the most economical way to address soil salinity. Upland cotton is a major fiber crop and pioneer plant on saline soil and thus its genetic architecture underlying salt tolerance should be extensively explored.

RESULTS

In this study, genome-wide association analysis and RNA sequencing were employed to detect salt-tolerant qualitative-trait loci (QTLs) and candidate genes in 196 upland cotton genotypes at the germination stage. Using comprehensive evaluation values of salt tolerance in four environments, we identified 33 significant single-nucleotide polymorphisms (SNPs), including 17 and 7 SNPs under at least two and four environments, respectively. The 17 stable SNPs were located within or near 98 candidate genes in 13 QTLs, including 35 genes that were functionally annotated to be involved in salt stress responses. RNA-seq analysis indicated that among the 98 candidate genes, 13 were stably differentially expressed. Furthermore, 12 of the 13 candidate genes were verified by qRT-PCR. RNA-seq analysis detected 6640, 3878, and 6462 differentially expressed genes at three sampling time points, of which 869 were shared.

CONCLUSIONS

These results, including the elite cotton accessions with accurate salt tolerance evaluation, the significant SNP markers, the candidate genes, and the salt-tolerant pathways, could improve our understanding of the molecular regulatory mechanisms under salt stress tolerance and genetic manipulation for cotton improvement.