Deep Learning for Classification and Selection of Cine CMR Images to Achieve Fully Automated Quality-Controlled CMR Analysis From Scanner to Report

Abstract

Introduction: Deep learning demonstrates great promise for automated analysis of CMR. However, existing limitations, such as insufficient quality control and selection of target acquisitions from the full CMR exam, are holding back the introduction of deep learning tools in the clinical environment. This study aimed to develop a framework for automated detection and quality-controlled selection of standard cine sequences images from clinical CMR exams, prior to analysis of cardiac function.Materials and Methods: Retrospective study of 3,827 subjects that underwent CMR imaging. We used a total of 119,285 CMR acquisitions, acquired with scanners of different magnetic field strengths and from different vendors (1.5T Siemens and 1.5T and 3.0T Phillips). We developed a framework to select one good acquisition for each conventional cine class. The framework consisted of a first pre-processing step to exclude still acquisitions; two sequential convolutional neural networks (CNN), the first (CNNclass) to classify acquisitions in standard cine views (2/3/4-chamber and short axis), the second (CNNQC) to classify acquisitions according to image quality and orientation; a final algorithm to select one good acquisition of each class. For each CNN component, 7 state-of-the-art architectures were trained for 200 epochs, with cross entropy loss and data augmentation. Data were divided into 80% for training, 10% for validation, and 10% for testing.Results: CNNclass selected cine CMR acquisitions with accuracy ranging from 0.989 to 0.998. Accuracy of CNNQC reached 0.861 for 2-chamber, 0.806 for 3-chamber, and 0.859 for 4-chamber. The complete framework was presented with 379 new full CMR studies, not used for CNN training/validation/testing, and selected one good 2-, 3-, and 4-chamber acquisition from each study with sensitivity to detect erroneous cases of 89.7, 93.2, and 93.9%, respectively.Conclusions: We developed an accurate quality-controlled framework for automated selection of cine acquisitions prior to image analysis. This framework is robust and generalizable as it was developed on multivendor data and could be used at the beginning of a pipeline for automated cine CMR analysis to obtain full automatization from scanner to report.

Artificial intelligence in cardiology: Hope for the future and power for the present

“…The complete framework also achieved accuracies of 89.7, 93.2, and 93.9% for 2-, 3-, and 4chamber acquisition from each study, respectively. This study demonstrates the future potential for high quality automated cine CMR analysis from the scanner to report (72).…”

Artificial intelligence in cardiac magnetic resonance fingerprinting

“…They are therefore ideal for cardiac MRF that involves complex acquisition strategies, scan-specific information, multi-dimensional image data dominated by noise, complicated reconstruction steps and computationally expensive calculations using conventional methods. Indeed, the ability of ML and DL has already been proven to be of great value in many domains of CMR imaging ( 30 , 31 ), from image reconstruction ( 32 – 34 ) to diagnosis of cardiomyopathies ( 35 , 36 ), reporting of cardiac function ( 37 , 38 ), segmentation of cardiac CINE imaging ( 39 , 40 ) and quantification of tissue parameters ( 41 ).…”

Clinical validation of a fully automated framework for ventricular segmentation and wall motion classification from cine CMR

“…The high amount of data required to train a deep learning model allows to achieve a high degree of accuracy and robustness of the predicted results. The automation of the imaging analysis allows optimizing the physicians' time, who can focus on critically reviewing all clinical information to reach a correct diagnosis [11]. Nevertheless, to the best of our knowledge, there is still no tool that automatically detects wall motion abnormalities of the ventricle.…”

“…Completely automated tools for ventricle segmentation and function quantification are becoming increasingly common [7,8]. Deep learning, a branch of artificial intelligence (AI), is an emergent methodology in the field of CMR [9,10,11]. This technique provides for the automation of repetitive tasks, significantly reducing the time required for clinical image analysis.…”