Search

Cyclin-dependent kinase-like 5 (Cdkl5) deficiency disorder (CDD) is a severe neurodevelopmental condition caused by mutations in the X-linked Cdkl5 gene. CDD is characterized by early-onset seizures in the first month of life, intellectual disability, motor and social impairment. No effective treatment is currently available and medical management is only symptomatic and supportive. Recently, mouse models of Cdkl5 disorder have demonstrated that mice lacking Cdkl5 exhibit autism-like phenotypes, hyperactivity, and dysregulations of the arousal system, suggesting the possibility to use these features as translational biomarkers. In this study, we tested Cdkl5 male and female mutant mice in an appetitive operant conditioning chamber to assess cognitive and motor abilities, and performed pupillometry to assess the integrity of the arousal system. Then, we evaluated the performance of artificial intelligence models to classify the genotype of the animals from the behavioral and physiological phenotype. The behavioral results show that CDD mice display impulsivity, together with low levels of cognitive flexibility and perseverative behaviors. We assessed arousal levels by simultaneously recording pupil size and locomotor activity. Pupillometry reveals in CDD mice a smaller pupil size and an impaired response to unexpected stimuli associated with hyperlocomotion, demonstrating a global defect in arousal modulation. Finally, machine learning reveals that both behavioral and pupillometry parameters can be considered good predictors of CDD. Since early diagnosis is essential to evaluate treatment outcomes and pupillary measures can be performed easily, we proposed the monitoring of pupil size as a promising biomarker for CDD. Cyclin-dependent kinase-like 5 (Cdkl5) deficiency disorder (CDD) is a severe neurodevelopmental condition caused by mutations in the X-linked Cdkl5 gene. CDD is characterized by early-onset seizures in the first month of life, intellectual disability, motor and social impairment. No effective treatment is currently available and medical management is only symptomatic and supportive. Recently, mouse models of Cdkl5 disorder have demonstrated that mice lacking Cdkl5 exhibit autism-like phenotypes, hyperactivity, and dysregulations of the arousal system, suggesting the possibility to use these features as translational biomarkers. In this study, we tested Cdkl5 male and female mutant mice in an appetitive operant conditioning chamber to assess cognitive and motor abilities, and performed pupillometry to assess the integrity of the arousal system. Then, we evaluated the performance of artificial intelligence models to classify the genotype of the animals from the behavioral and physiological phenotype. The behavioral results show that CDD mice display impulsivity, together with low levels of cognitive flexibility and perseverative behaviors. We assessed arousal levels by simultaneously recording pupil size and locomotor activity. Pupillometry reveals in CDD mice a smaller pupil size and an impaired response to unexpected stimuli associated with hyperlocomotion, demonstrating a global defect in arousal modulation. Finally, machine learning reveals that both behavioral and pupillometry parameters can be considered good predictors of CDD. Since early diagnosis is essential to evaluate treatment outcomes and pupillary measures can be performed easily, we proposed the monitoring of pupil size as a promising biomarker for CDD.

Exploiting well-labeled training sets has led deep learning models to astonishing results for counting biological structures in microscopy images. However, dealing with weak multi-rater annotations, i.e., when multiple human raters disagree due to non-trivial patterns, remains a relatively unexplored problem. More reliable labels can be obtained by aggregating and averaging the decisions given by several raters to the same data. Still, the scale of the counting task and the limited budget for labeling prohibit this. As a result, making the most with small quantities of multi-rater data is crucial. To this end, we propose a two-stage counting strategy in a weakly labeled data scenario. First, we detect and count the biological structures; then, in the second step, we refine the predictions, increasing the correlation between the scores assigned to the samples and the raters' agreement on the annotations. We assess our methodology on a novel dataset comprising fluorescence microscopy images of mice brains containing extracellular matrix aggregates named perineuronal nets. We demonstrate that we significantly enhance counting performance, improving confidence calibration by taking advantage of the redundant information characterizing the small sets of available multi-rater data.