OBJECTIVE: Indicator-dilution curves (IDCs) for the estimation of pulmonary transit times (PTTs) can be generated noninvasively using contrast echocardiography. Currently, these IDCs are analyzed by manual inspection, which is not feasible in a clinical setting, or fit to a statistical model to derive parameters of interest. However, IDCs generated from patients are frequently subject to significant low-frequency noise and recirculation artifacts that obscure the first-pass signal and render model fitting impractical or inaccurate. Thus, the objective of this paper was to develop alternative computational methods to determine PTT using noisy clinical data in which the signal decay is not adequately visible. METHODS: We report on a method that uses a model fit to the rise portion of the IDCs to determine the signal inflection point. Additionally, a signal truncation algorithm was developed that enables automated analysis of the IDCs. RESULTS: We compare PTTs derived from our inflection point method to those obtained by manual inspection in 25 patients (R(2) = 0.86) and to those obtained by mean transit time calculation following fitting to a local density random walk model (R(2) = 0.80) in a subset of this cohort. CONCLUSION: Combined with a signal truncation algorithm, the inflection point method provides robust, automated determination of PTT from noisy IDCs containing recirculation artifacts. SIGNIFICANCE: The inflection point method addresses the need for computational analysis of IDCs obtained from contrast echocardiograms that are not amenable to first-pass model fitting.
OBJECTIVE: Indicator-dilution curves (IDCs) for the estimation of pulmonary transit times (PTTs) can be generated noninvasively using contrast echocardiography. Currently, these IDCs are analyzed by manual inspection, which is not feasible in a clinical setting, or fit to a statistical model to derive parameters of interest. However, IDCs generated from patients are frequently subject to significant low-frequency noise and recirculation artifacts that obscure the first-pass signal and render model fitting impractical or inaccurate. Thus, the objective of this paper was to develop alternative computational methods to determine PTT using noisy clinical data in which the signal decay is not adequately visible. METHODS: We report on a method that uses a model fit to the rise portion of the IDCs to determine the signal inflection point. Additionally, a signal truncation algorithm was developed that enables automated analysis of the IDCs. RESULTS: We compare PTTs derived from our inflection point method to those obtained by manual inspection in 25 patients (R(2) = 0.86) and to those obtained by mean transit time calculation following fitting to a local density random walk model (R(2) = 0.80) in a subset of this cohort. CONCLUSION: Combined with a signal truncation algorithm, the inflection point method provides robust, automated determination of PTT from noisy IDCs containing recirculation artifacts. SIGNIFICANCE: The inflection point method addresses the need for computational analysis of IDCs obtained from contrast echocardiograms that are not amenable to first-pass model fitting.
Authors: Ken Monahan; Daniel Lenihan; Evan L Brittain; Linda Saliba; Robert N Piana; Leslie L Robison; Melissa M Hudson; Gregory T Armstrong Journal: Echocardiography Date: 2018-05-14 Impact factor: 1.724