Accurate intra-patient lung Computed Tomography (CT) registration is essential for tracking disease progression, evaluating treatment, and planning radiotherapy. Traditional intensity-based methods face challenges with large respiratory deformations and low soft-tissue contrast, which motivates our work on point-cloud approaches to better capture anatomical structures. This project adapts the Iterative Optimal Transport (IOT) framework from 2D multimodal image registration to 3D pulmonary vessel point clouds derived from paired expiratory and inspiratory CT scans. Two key modifications were introduced: a 3D Forstner corner detector for structure-aware point sampling and a nearest-neighbor distance evaluation protocol. Eight experimental configurations were tested on COPD subjects from the PVT1010 dataset, varying in the choice of sampler, polynomial degree, and optimal transport formulation. Forstner-scalar sampling with unbalanced optimal transport achieved a 59.9% reduction in the mean nearest-neighbor distance (from 10.8 mm to 4.3 mm) compared to pre-registration, significantly outperforming farthest-point sampling across all conditions (p < 0.001). Ongoing work aims to extend this framework from pairwise expiratory-inspiratory registration to full respiratory cycle registration across multiple time points (using a 5D CT dataset from 25 sequential time points). These extensions would support motion-based dose computation and longitudinal tracking of lung lesions across treatment cycles in clinical settings.

Introduction: We previously developed an in-line ventriculoperitoneal shunt (VPS) flow sensor to detect shunt failure earlier, and demonstrated feasibility under pulsatile-flow conditions ex vivo using normal saline. Clinically encountered CSF can vary which may alter signal behavior and affect flow estimation. In this study, we evaluated this sensor in an expanded ex vivo dataset of freshly harvested human CSF samples with heterogeneous laboratory profiles. Methods: Fresh CSF specimens were collected from patients with external ventricular drains (EVDs) (n = 7) or indwelling VPSs (n = 2) and characterized using standard CSF laboratory studies. The sensor was tested on our previously described benchtop pulsatile-flow platform across the physiologic CSF flow range. Log-log modeling was used to characterize the relationship between flow rate and time-to-peak. Results: Across specimens, the sensor preserved a flow-dependent timing response despite heterogeneity in CSF composition. While patient samples differed in absolute time-to-peak behavior relative to normal saline and to one another, the persistence of specimen-level power-law structure suggests that biologic variability affects calibration more than the fundamental detectability of flow. Conclusion: The sensor reliably measured ex vivo pulsatile flow across a broad range of human CSF samples. Future studies will validate performance in an in vivo large-animal model of hydrocephalus as a step toward continuous VPS flow surveillance and earlier detection of VPS failure

Antibody–drug conjugates (ADCs) have been a transformative development in cancer therapy; however, clinical responses remain variable and resistance frequently emerges. One well-characterized mechanism of resistance is drug efflux mediated by ATP-binding cassette (ABC) transporters. Notably, monomethyl auristatin E (MMAE), a widely used cytotoxic payload in both approved and investigational ADCs, is a known substrate of the efflux transporter ABCB1. The primary objective of this study was to generate ADC-resistant cancer cell lines and characterize the molecular changes associated with resistance. Two human cancer cell lines with acquired resistance to a CLDN6-targeting ADC—relevant to ovarian cancer—were successfully established. These resistant models were characterized using Western blotting, flow cytometry, and cell proliferation assays. The analyses revealed decreased expression of the target antigen CLDN6 and increased expression of the efflux transporter ABCB1, suggesting their potential roles in mediating resistance. Elucidating the mechanisms underlying ADC resistance may inform strategies to overcome resistance or determine its reversibility, ultimately contributing to the development of more effective therapeutic approaches. Future extensions will involve RNA sequencing (RNA-seq) to compare resistant cell lines with their parental counterparts, with the goal of identifying additional gene expression changes associated with resistance.

Placenta accreta spectrum (PAS) is a condition characterized by abnormally deep placental implantation into the uterine wall. There has been a drastic increase in PAS cases in recent years, and the condition is associated with significant maternal morbidity and mortality. Maternal outcomes are optimized when PAS patients deliver in tertiary hospitals with specialized resources, making early diagnosis critical. This study aims to identify maternal plasma biomarkers of placenta accreta spectrum using a proteomic approach and solidify these findings with a validation cohort. An outside facility sequenced a total of 85 maternal blood samples, 64 of which were in our discovery cohort and 21 that made up our validation cohort. The discovery cohort was used to identify candidate biomarkers, and the resulting data was superimposed onto our validation cohort to predict which samples were PAS and which were control. Analysis is in progress; current data identifies 42 proteins as significantly up or downregulated and predicts 14 PAS and 7 control samples within the validation group. The predicted classifications will be compared with the clinical diagnoses from delivery to assess the predictive strength of our candidate biomarkers. By identifying reliable biomarkers for PAS, this study hopes to establish a reliable detection method, enabling informed delivery planning and enhancing maternal outcomes.