MIM-CyCIF: masked imaging modeling for enhancing cyclic immunofluorescence (CyCIF) with panel reduction and imputation.

Cyclic Immunofluorescence (CyCIF) can quantify multiple biomarkers, but panel capacity is limited by technical challenges. We propose a computational panel reduction approach that can impute the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns while concurrently increasing speed and panel content and decreasing cost. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer, illustrating applicability of our approach to diverse tissue types.

Ultra high content analyses of circulating and tumor associated hybrid cells reveal phenotypic heterogeneity.

Persistently high, worldwide mortality from cancer highlights the unresolved challenges of disease surveillance and detection that impact survival. Development of a non-invasive, blood-based biomarker would transform survival from cancer. We demonstrate the functionality of ultra-high content analyses of a newly identified population of tumor cells that are hybrids between neoplastic and immune cells in patient matched tumor and peripheral blood specimens. Using oligonucleotide conjugated antibodies (Ab-oligo) permitting cyclic immunofluorescence (cyCIF), we present analyses of phenotypes among tumor and peripheral blood hybrid cells. Interestingly, the majority of circulating hybrid cell (CHC) subpopulations were not identified in tumor-associated hybrids. These results highlight the efficacy of ultra-high content phenotypic analyses using Ab-oligo based cyCIF applied to both tumor and peripheral blood specimens. The combination of a multiplex phenotypic profiling platform that is gentle enough to analyze blood to detect and evaluate disseminated tumor cells represents a novel approach to exploring novel tumor biology and potential utility for developing the population as a blood-based biomarker in cancer.

Associations between Computationally Derived Parent Emotional Sentiment Scores and Child ADHD and ODD Over Time.

Family emotional climate is often assessed as expressed emotion (EE) using the five-minute speech sample (FMSS). Parent EE is related to child externalizing behavior, but the relationship with ADHD apart from externalizing is unclear. We report the largest ADHD-non-ADHD study of EE to date, introduce computational scoring of the FMSS to assay parent negative sentiment, and use this to evaluate reciprocal parent-child effects over time in ADHD while considering comorbid ODD. Parents of 810 children (nADHD = 509), aged 7-13 years old, completed the FMSS at three points. The FMSS was expert-coded for EE-Criticism at Time 1 and Time 2, negative sentiment was scored at all three time points. Sentiment and EE-Criticism were moderately correlated (r =.39, p <.001, 95% CI [0.32, 0.46]), and each was similarly correlated with baseline ADHD symptoms (r's range 0.31-0.33, p <.001) and ODD symptoms (r(ODD-EE) = 0.35, p <.001; r(ODD-sentiment = 0.28, p <.001). A longitudinal, cross-lagged panel model revealed that increases over time in parental negative sentiment scores led to increased ODD symptoms. Parent sex (namely fathers, but not mothers) showed an interaction effect of sentiment with ADHD. ADHD and ODD are independently and jointly associated with parental EE-Criticism and negative sentiment assessed by the FMSS cross-sectionally. A recursive effects model is supported for ODD, but for ADHD effects depend on which parent is assessed. For fathers, ADHD was related to negative sentiment in complex manners but for mothers, negative sentiment was related primarily to ODD.