Experimental Validation of Revised Criteria for Pulmonary Hypertension Diagnosis by Uncertainty Evaluation

Pulmonary hypertension (PH) is a complex pathology, which affects the pulmonary circulatory system. The European Society of Cardiology (ESC)/European Respiratory Society (ERS) Guidelines on diagnosis and treatment of pulmonary hypertension represents the standard used in the clinical practice for the definition and the treatment of the pathology. At the moment, medical screenings used for the diagnosis of PH are both invasive and noninvasive. The noninvasive diagnostic tools use continuous-wave Doppler echocardiography to measure the tricuspid and/or pulmonary regurgitation velocity. Such parameter represents the input variable of the Bernoulli equation so as to estimate specific pulmonary hemodynamic variables used in the diagnosis. In detail, the PH diagnosis is positive when such hemodynamic variables exceed fixed cutoff values. Nevertheless the poor accuracy of Doppler-derived pressure estimations affects the diagnosis result negatively. Thus, false-negative and false-positive diagnoses are possible. We deal with such issue to provide physicians with useful and fault-tolerant diagnostic procedures. The aim is to guarantee a more reliable interpretation of the echocardiographic variables. In view of this, the criteria at present used in clinical practice for detecting the PH have been revised to improve the accuracy of diagnosis. The proposed procedure is based on the accurate and reliable interpretation of hemodynamic indexes by evaluating the uncertainty impact. In detail, the procedure is based on the analysis of the standard uncertainty associated with the hemodynamic parameters estimation. An algorithm has been implemented so to perform the data processing and the revised criteria. The proposed criteria have been validated both by numerical simulations and by experimentation on real clinical cases. Comparisons with the criteria of the ESC/ERS Guidelines have been carried out. Results are reported in the manuscript. Reliability and sensitivity of the computing algorithm have been tested. Experimental results and numerical simulations have shown interesting outcomes. During the tests, the algorithm has properly diagnosed the pathology. It has shown good tolerance to false-positive and false-negative diagnoses with respect to the standard criteria.