Arq Bras Cardiol: Imagem cardiovasc. 2023; 36(1): e376
Mitral Valve Area Quantification Using Digital Image Processing: Is That Feasible?
Echocardiography, since its initial descriptions in the 1950s, has evolved every decade with the incorporation of numerous new techniques, enabling structural cardiac analysis from multiple planes of spatial observation in real time. This occurred with the advent of nanotechnology, digital technology, application of modifiable algorithms based on the concept of machine learning and deep learning through the development of artificial intelligence (AI).– The use of AI is an ever-increasing reality in medicine as a whole, mainly in operator-dependent imaging areas, such as ultrasonography imaging.
In the concept of analyzing the confluence of a very large number of pieces of information (e.g. radiomics, metabolomics, proteomics) for the formation of data clusters with the aim of integrated patient observation, the formation of precision medicine, AI is used in a rational and organized way. The use of AI allows faster, more balanced and homogeneous observation of inputs, in the sense of minimizing inter and intraobserver errors.– In the specific echocardiographic evaluation, the use of AI presents a very wide range of applications, having been used for the acquisition of echocardiographic information for the quantification of left ventricular volumes and mass, analysis of biventricular systolic function, segmentation and qualification of the cusps, annulus and mitral valve apparatus, analysis of hypertrophic phenocopy (athletes, patients with hypertrophic cardiomyopathy), to predict the outcome of percutaneous hemodynamic procedures (e.g.: mitraclip). We are currently working on a project in association with the University of Chicago for AI analysis of patients affected by transthyretin cardiac amyloidosis.
[…]
484
