Multiview Diffusion Map Improves Prediction of Fluid Intelligence With Two Paradigms of fMRI Analysis

Latent Similarity Identifies Important Functional Connections for Phenotype Prediction

OBJECTIVE

Endophenotypes such as brain age and fluid intelligence are important biomarkers of disease status. However, brain imaging studies to identify these biomarkers often encounter limited numbers of subjects but high dimensional imaging features, hindering reproducibility. Therefore, we develop an interpretable, multivariate classification/regression algorithm, called Latent Similarity (LatSim), suitable for small sample size but high feature dimension datasets.

METHODS

LatSim combines metric learning with a kernel similarity function and softmax aggregation to identify task-related similarities between subjects. Inter-subject similarity is utilized to improve performance on three prediction tasks using multi-paradigm fMRI data. A greedy selection algorithm, made possible by LatSim's computational efficiency, is developed as an interpretability method.

RESULTS

LatSim achieved significantly higher predictive accuracy at small sample sizes on the Philadelphia Neurodevelopmental Cohort (PNC) dataset. Connections identified by LatSim gave superior discriminative power compared to those identified by other methods. We identified 4 functional brain networks enriched in connections for predicting brain age, sex, and intelligence.

CONCLUSION

We find that most information for a predictive task comes from only a few (1-5) connections. Additionally, we find that the default mode network is over-represented in the top connections of all predictive tasks.

SIGNIFICANCE

We propose a novel prediction algorithm for small sample, high feature dimension datasets and use it to identify connections in task fMRI data. Our work can lead to new insights in both algorithm design and neuroscience research.

The Value of Nuclear Magnetic Resonance in Liver Nodular Lesions

In order to analyze the value of contrast-enhanced ultrasound (CEUS) combined with functional magnetic resonance imaging (fMRI) in the early differential diagnosis of liver nodular lesions, the authors studied the value of MRI in liver nodular lesions. A total of 82 patients with liver nodular lesions admitted to the hospital were selected for retrospective analysis; all of them underwent CEUS and fMRI examinations, and taking a biopsy or postoperative pathological examination results as the gold standard, the diagnostic value of CEUS, fMRI single item, and the two combined examinations for liver nodular lesions was analyzed by four-table. The biopsy or postoperative pathological examination results showed that a total of 88 lesions were detected in 82 patients, including 51 patients with benign lesions, with 54 lesions, and 31 patients with malignant lesions, with 34 lesions. Taking biopsy or pathological examination results as the gold standard, the four-table analysis CEUS had a sensitivity of 79.63%, a specificity of 82.35%, an accuracy of 80.68%, and a Kappa value of 0.603 for diagnosing benign and malignant liver nodular lesions. The sensitivity of fMRI in diagnosing benign and malignant liver nodular lesions was 83.33%, the specificity was 85.29%, the accuracy was 84.09%, and the Kappa value was 0.672; the combined sensitivity of the two in the diagnosis of benign and malignant liver nodular lesions was 94.44%, the specificity was 91.18%, the accuracy was 93.18%, and the Kappa value was 0.856, both of which were superior to single detection, and the difference in accuracy was statistically significant (χ²  = 5.683, P < 0.05). CEUS and fMRI have a certain value in the differential diagnosis of liver nodular lesions; the combination of the two can improve the diagnostic sensitivity and accuracy, and has more clinical application value.