Application of Machine Learning in Pulmonary Function Assessment Where Are We Now and Where Are We Going?

Analysis of pulmonary function tests (PFTs) is an area where machine learning (ML) may benefit clinicians, researchers, and the patients. PFT measures spirometry, lung volumes, and carbon monoxide diffusion capacity of the lung (DLCO). The results are usually interpreted by the clinicians using discrete numeric data according to published guidelines. PFT interpretations by clinicians, however, are known to have inter-rater variability and the inaccuracy can impact patient care. This variability may be caused by unfamiliarity of the guidelines, lack of training, inadequate understanding of lung physiology, or simply mental lapses. A rules-based automated interpretation system can recapitulate expert’s pattern recognition capability and decrease errors. ML can also be used to analyze continuous data or the graphics, including the flow-volume loop, the DLCO and the nitrogen washout curves. These analyses can discover novel physiological biomarkers. In the era of wearables and telehealth, particularly with the COVID-19 pandemic restricting PFTs to be done in the clinical laboratories, ML can also be used to combine mobile spirometry results with an individual’s clinical profile to deliver precision medicine. There are, however, hurdles in the development and commercialization of the ML-assisted PFT interpretation programs, including the need for high quality representative data, the existence of different formats for data acquisition and sharing in PFT software by different vendors, and the need for collaboration amongst clinicians, biomedical engineers, and information technologists. Hurdles notwithstanding, the new developments would represent significant advances that could be the future of PFT, the oldest test still in use in clinical medicine.

Washout and Awakening Times after Inhaled Sedation of Critically Ill Patients: Desflurane Versus Isoflurane

In recent years, inhaled sedation has been increasingly used in the intensive care unit (ICU). The aim of this prospective, controlled trial was to compare washout and awakening times after long term sedation with desflurane and isoflurane both administered with the Mirus™ system (TIM GmbH, Koblenz, Germany). Twenty-one consecutive critically ill patients were alternately allocated to the two study groups, obtaining inhaled sedation with either desflurane or isoflurane. After 24 h study sedation, anesthetic washout curves were recorded, and a standardized wake-up test was performed. The primary outcome measure was the time required to decrease the endtidal concentration to 50% (T50%). Secondary outcome measures were T80% and awakening times (all extremities moved, RASS −2). Decrement times (min) (desflurane versus isoflurane, median (1st quartile—3rd quartile)) (T50%: 0.3 (0.3–0.4) vs. 1.3 (0.4–2.3), log-rank test P = 0.002; P80%: 2.5 (2–5.9) vs. 12.1 (5.1–20.2), P = 0.022) and awakening times (to RASS −2: 7.5 (5.5–8.8) vs. 41.0 (24.5–43.0), P = 0.007; all extremities moved: 5.0 (4.0–8.5) vs. 13.0 (8.0–41.25), P = 0.037) were significantly shorter after desflurane compared to isoflurane. The use of desflurane with the Mirus™ system significantly shortens the washout times and leads to faster awakening after sedation of critically ill patients.

Relationships between the lung clearance index and conductive and acinar ventilation heterogeneity

The lung clearance index (LCI) derived from a multiple breath washout test has regained considerable popularity in recent years, alternatively being promoted as an early detection tool or a marker of small airways function. In this study, we systematically investigated the link between LCI and indexes of acinar and conductive airways ventilation heterogeneity (Sacin, Scond) to assess potential contributions from both lung zones. Relationships were examined in 55 normal subjects after provocation, where only Scond is known to be markedly increased, and in 55 asthma patients after bronchodilation, in whom both Scond and Sacin ranged between normal and abnormal. LCI was correlated to Scond in both groups ( R = 0.37–0.43; P < 0.01 for both); in the asthma group, LCI was also tightly correlated to Sacin ( R = 0.70; P < 0.001). Potential mechanisms operational at various levels of the bronchial tree were identified by considering washout curvilinearity in addition to LCI to distinguish specific ventilation and dead space effects (also illustrated by simple 2-compartment model simulations). Although the asthma data clearly demonstrate that LCI can reflect very peripheral ventilation heterogeneities, the normal provocation data also convincingly show that LCI increases may be the exclusive result of far more proximal ventilation heterogeneities. Because LCI potentially includes heterogeneities at all length scales, it is suggested that ventilation imaging in combination with LCI measurement at the mouth could identify the scale of relevant ventilation heterogeneities. In the meantime, interpretations of LCI results in the clinic based on washout curves collected at the mouth should be handled with caution.

Minimally invasive quantification of lymph flow in mice and rats by imaging depot clearance of near-infrared albumin

There is a lack of available methods to noninvasively quantify lymphatic function in small experimental animals, a necessity for studies on lymphatic system pathophysiology. We present a new method to quantify lymph flow in mice and rats, based on optically monitoring the depot clearance of near-infrared fluorescently labeled albumin and subsequent calculation of removal rate constants ( k). BSA was conjugated with Alexa680 NHS ester and remained stable in protein-rich solutions without free dye dissociation. To assess lymph flow, mice or rats were imaged every 30 or 60 min during a 3- to 6-h period following an intradermal injection of 0.5 or 1 μl Alexa680-albumin. Mice were awake between measurements, whereas rats were anesthetized throughout the experiment. The k, a parameter defined as equivalent to lymph flow, was calculated from the slopes of the resultant log-linear washout curves and averaged −0.40 ± 0.03 and −0.30 ± 0.02%/min for control C57BL/6 and C3H mice, respectively. Local administration of the vasoconstrictor endothelin-1 in mice led to a significant reduction in k, whereas overhydration in rats increased k, reflecting the coupling between capillary filtration and lymph flow. Furthermore, k was 50% of wild type in lymphedema Chy mice where dermal lymphatics are absent. We conclude that lymph flow can be determined as its rate constant k by optical imaging of depot clearance of submicroliter amounts of Alexa680-albumin. The method offers a minimally invasive, reproducible, and simple alternative to assess lymphatic function in mice and rats.

The nature of leukocyte shape changes in the pulmonary capillaries

The size discrepancy between leukocytes [white blood cells (WBCs)] and pulmonary capillaries requires WBCs to deform. We investigated the persistence of this deformation on cells leaving the capillary bed and the role played by the cytoskeleton. Isolated rabbit lungs were perfused in situ via the pulmonary artery with effluent fractions collected from the left ventricle. Washout curves from cell counts in each fraction confirmed that WBCs are preferentially retained over erythrocytes. WBC deformation present on exit from the circulation was compared with that present after recovery in paired fractions, fixed either immediately or 60 min later. These cells were compared with cells recovered from the capillary in perfused fixative or fixed in peripheral blood. Our results show that leukocyte deformation persisted after the cells exited the pulmonary circulation. This deformation was associated with minimal submembranous F-actin staining, and microtubule distribution and cell polarization were unchanged. We conclude that cytoskeletal changes that occur during WBC deformation in the pulmonary capillaries are minimal and differ from those known to occur in actively migrating cells during chemotaxis.

Analyzing 13NN lung washout curves in the presence of intraregional nonuniformities

The use of 13NN and positron imaging provides a powerful noninvasive means of assessing regional pulmonary function. Current techniques for analyzing lung washout curves, however, are subject to error due to intraregional nonuniformities, particularly in the presence of areas with very long time constants or gas trapping. This paper presents a simple “hybrid” method for analyzing 13NN-washout studies that addresses the problems of regions of air trapping and long time constants within a region of interest. This method assumes an exponential washout form for the slow regions, estimates their fractional volume, and subtracts their contribution from the overall washout curve. The modified Stewart-Hamilton method is then applied to the remaining washout curve to calculate its ventilation. The over-all specific ventilation of the region is calculated as the volume-weighted average of the specific ventilations of the two compartments. The performance of the hybrid method is compared with other currently used correction techniques by using a simulated two-compartment lung washout in which there is a variable amount of gas trapping. The analysis techniques are also applied to washout data with intraregional nonuniformities obtained during experimentally created unilateral bronchial obstruction. This new approach has the advantages of automatically correcting for washout truncation, estimating the relative size of the trapped or slow region while eliminating its bias on the regional ventilation, and yet retaining the unrestricted and robust nature of the Stewart-Hamilton method in the analysis of the well-ventilated regions.