UltraAIGenomics: Artificial Intelligence-Based Cardiovascular Disease Risk Assessment by Fusion of Ultrasound-Based Radiomics and Genomics Features for Preventive, Personalized and Precision Medicine: A Narrative Review

Abstract

Cardiovascular disease (CVD) diagnosis and treatment are challenging since symptoms appear late in the disease's progression. Despite clinical risk scores, cardiac event prediction is inadequate, and many at-risk patients are not adequately categorised by conventional risk factors alone. Integrating genomic-based biomarkers (GBBM), specifically those found in plasma and/or serum samples, along with novel non-invasive radiomic-based biomarkers (RBBM) such as plaque area and plaque burden can improve the overall specificity of CVD risk. This review proposes two hypotheses: (i) RBBM and GBBM biomarkers have a strong correlation and can be used to detect the severity of CVD and stroke precisely, and (ii) introduces a proposed artificial intelligence (AI)-based preventive, precision, and personalized ( ${\mathrm{aiP}}^3$ ) CVD/Stroke risk model. The PRISMA search selected 246 studies for the CVD/Stroke risk. It showed that using the RBBM and GBBM biomarkers, deep learning (DL) modelscould be used for CVD/Stroke risk stratification in the ${\mathrm{aiP}}^3$ framework. Furthermore, we present a concise overview of platelet function, complete blood count (CBC), and diagnostic methods. As part of the AI paradigm, we discuss explainability, pruning, bias, and benchmarking against previous studies and their potential impacts. The review proposes the integration of RBBM and GBBM, an innovative solution streamlined in the DL paradigm for predicting CVD/Stroke risk in the ${\mathrm{aiP}}^3$ framework. The combination of RBBM and GBBM introduces a powerful CVD/Stroke risk assessment paradigm. ${\mathrm{aiP}}^3$ model signifies a promising advancement in CVD/Stroke risk assessment.

Keywords: artificial intelligence; bias; cardiovascular disease; deep learning; explainable AI; genomics; pruning; radiomics; stroke.

Fig. 1.

The overview of composite biomarkers using an AI model for the preventive, personalized, and precise (${\mathrm{aiP}}^{\mathrm{3}}$) solution leads to multiclass CVD risk assessment. CVD, cardiovascular disease; OBBM, office-based biomarkers; LBBM, laboratory-based biomarkers; RBBM, radiomics-based biomarkers; GBBM, genomics-based biomarkers; PBBM, proteomics-based biomarkers; EBBM, environment-based biomarkers.

Fig. 2.

PRISMA model for study selection. I, included; E, excluded; AI, artificial intelligence.

Fig. 3.

Cardiac multivariate biomarker assessments (OBBM, LBBM, RBBM, and GBBM) for the risk stratification of atherosclerosis disease. OBBM, office-based biomarkers; LBBM, laboratory-based biomarkers; RBBM, radiomics-based biomarkers; GBBM, genomics-based biomarkers; PBBM, proteomics-based biomarkers; EBBM, environment-based biomarkers; BMI, body mass index; eGFR, estimated glomerular filtration rate; ESR, erythrocyte sedimentation rate; cIMT, carotid intima-media thickness; IMTV, intima media thickness variability; MPH, maximum plaque height.

Fig. 4.

DL-based architecture for CVD risk assessment. OBBM, office-based biomarkers; LBBM, laboratory-based biomarkers; RBBM, radiomics-based biomarkers; GBBM, genomics-based biomarkers; PBBM, proteomics-based biomarkers; EBBM, environment-based biomarkers; LSTM, long short-term memory network; RNN, recurrent neural network; GRU, gated recurrent units; Bi, bidirectional; CVD, cardiovascular disease; DL, deep learning; GT, ground truth; AI, artificial intelligence; ROC, receiver operating characteristic.

Fig. 5.

CUSIP Measurement. (a) Carotid artery is a potential surrogate marker for the coronary artery. Also, the grayscale images are shown for carotid longitudinal B-mode US scans and coronary IVUS transverse scan (b) B-mode carotid longitudinal imaging system using linear ultrasound [43]. CUSIP, carotid image-based phenotypes; US, Ultrasound; IVUS, intra vascular ultrasound.

Fig. 6.

UNet model for segmentation of the atherosclerotic plaque wall [29]. GT is ground truth, and Conv is convolution. The UNet-based DL model can transmit features extracted from the encoder to the decoder phases and preserve the desired features during shape reconstruction at the decoder phase. In contrast to geometric curves based on level sets, UNet-based DL does not require the positioning of the first curves. Moreover, it needs the gold standard for training the UNet-based DL models [193, 194]. DL, deep learning; GT, ground truth; UNet, U-shaped network.

Fig. 7.

LSTM architecture for CVD risk stratification. LSTM is a long short-term memory network, and ReLU is a rectified linear unit. CVD, cardiovascular disease; LSTM, long short-term memory network.