Empirical data drift detection experiments on real-world medical imaging data.

While it is common to monitor deployed clinical artificial intelligence (AI) models for performance degradation, it is less common for the input data to be monitored for data drift - systemic changes to input distributions. However, when real-time evaluation may not be practical (eg., labeling costs) or when gold-labels are automatically generated, we argue that tracking data drift becomes a vital addition for AI deployments. In this work, we perform empirical experiments on real-world medical imaging to evaluate three data drift detection methods' ability to detect data drift caused (a) naturally (emergence of COVID-19 in X-rays) and (b) synthetically. We find that monitoring performance alone is not a good proxy for detecting data drift and that drift-detection heavily depends on sample size and patient features. Our work discusses the need and utility of data drift detection in various scenarios and highlights gaps in knowledge for the practical application of existing methods.

Calibration of an Electrical Analog Model of Liver Hemodynamics in Fontan Patients.

Fontan associated liver disease is a common complication in patients with Fontan circulation, who were born with a single functioning heart ventricle. The hepatic venous pressure gradient (HVPG) is used to assess liver health and is a surrogate measure of the pressure gradient across the entire liver (portal pressure gradient (PPG)). However, it is thought to be inaccurate in Fontan patients. The main objectives of this study were (1) to apply an existing detailed lumped parameter model (LPM) of the liver to Fontan patients using patient-specific clinical data and (2) to determine whether HVPG is a suitable measurement of PPGs in these patients. An existing LPM of the liver blood circulation was applied and tuned to simulate patient-specific liver hemodynamics. Geometries were collected from seven adult Fontan patients and used to evaluate model parameters. The model was solved and tuned using waveform measurements of flows, inlet and outlet pressures. The predicted ratio of portal to hepatic venous pressures is comparable to in vivo measurements. The results confirmed that HVPG is not suitable for Fontan patients, as it would underestimate the portal pressures gradient by a factor of 3 to 4. Our patient-specific liver model provides an estimate of the pressure drop across the liver, which differs from the clinically used metric HVPG. This work represents a first step toward models suitable to assess liver health in Fontan patients and improve its long-term management.

Cardiovascular and abdominal flow alterations in adults with morphologic evidence of liver disease post Fontan palliation.

BACKGROUND: Although morphologic abnormalities in the liver are commonly encountered post Fontan palliation, the relationships between hepatic morphology, vascular flows, and clinical status remain incompletely understood. We therefore aimed to explore flow characteristics in hepatic and intestinal vessels and to examine cardiovascular associations with liver disease. METHODS: This was a retrospective study of adults post Fontan palliation undergoing clinically indicated cardiovascular magnetic resonance imaging (MRI). Patients were included if MRI flow quantification was available for cardiac, hepatic and intestinal vessels; patients were excluded if phase-contrast flow imaging was insufficient for analysis. RESULTS: Thirty patients were studied (median age at MRI 28.5 years [range 19-47]). Eighteen subjects (60%) were classified as having morphologic liver disease according to validated criteria based on available MRI imaging. Abdominal and cardiovascular flows were quantified. Patients with morphologic liver disease had a 41% reduction in superior mesenteric artery (211 ± 124 versus 358 ± 181 mL/min/m2, p = .004), a 36% reduction in hepatic vein (496 ± 247 versus 778 ± 220 mL/min/m2, p = .01), a 31% reduction in portal vein (399 ± 133 versus 580 ± 159 mL/min/m2, p = .004), and an 18% reduction in Fontan pathway flows (1358 ± 429 versus 1651 ± 270 mL/min/m2, p = .04) compared with the remaining population. Adverse cardiovascular events were not associated with morphologic liver disease. CONCLUSION: Morphologic liver disease appears to be associated with flow alterations within the heart, liver and intestine post Fontan palliation. These novel observations suggest that a potential relationship exists between morphologic disease and vascular flows thereby providing further insights into the pathophysiology of liver disease in this high-risk population.