An automated method for BRISQUE quantification of image quality in echocardiography

Abstract Background Echocardiography (echo) remains the most widely used imaging modality for assessment, monitoring, and prognostication of heart. Despite its prevalence, standardisation efforts echo chamber quantification are ongoing, with challenges owing to subjectivity during acquisition analysis. Furthermore, confidence in derived functional indices is often dependent on quality acquired images. However, few studies have investigated accuracy measurements compared a reference such as cardiac magnetic resonance (CMR) imaging, when stratified by image quality. Purpose To develop an objective automated method quantify quality, subsequently investigate relationship between patient demographics, well magnitude bias left ventricular (LV) CMR. Methods Transthoracic apical 2D (2DE) 3D (3DE) data from 128 participants (72 healthy controls 56 patients heart disease) were train BRISQUE (Blind/Referenceless Image Spatial Quality Evaluator) algorithm [1]. Briefly, feature extraction was performed fitting pixel luminances generalised Gaussian distribution (GGD), followed support vector regression correlate features (i.e., shape, variance, mean parameters GGD) scores (Fig. 1). Independent models trained 580 2DE images (consisting 2-, 3-, 4-chamber views) targeted LV 3DE acquisitions at end-diastole, each assigned subjective perceived score 1 (poor) 9 (excellent) single observer. including end-diastolic volume (EDV), end-systolic (ESV), ejection fraction (EF), global longitudinal strain (GLS), assessed according standard guidelines. Resultant plotted against demographics (age, height, weight) measurement comparison CMR (acquired within hour echo). Results Several linear relationships (where P-value slope <0.05) observed indices, scores. Increasing weight (and height 3DE) found be associated poorer There no apparent age. Of interest, EF exhibited both 2), whereby higher tended overestimate EF, while lower underestimated EF. For 3DE, dependency also ESV GLS biases. Conclusions can objectively produce which those expert observer, potential utility standardised Using this method, it may possible predict characteristics adversely impact biases certain indices. Funding Acknowledgement Type funding sources: Public grant(s) – National budget only. Main source(s): Health Research Council (HRC) New Zealand; Heart Foundation (NHF) Zealand