Deep learning-based carotid plaque vulnerability classification with multicentre contrast-enhanced ultrasound video: a comparative diagnostic study

Automated diagnosis of atherosclerosis using multi-layer ensemble models and bio-inspired optimization in intravascular ultrasound imaging

Atherosclerosis causes heart disease by forming plaques in arterial walls. IVUS imaging provides a high-resolution cross-sectional view of coronary arteries and plaque morphology. Healthcare professionals diagnose and quantify atherosclerosis physically or using VH-IVUS software. Since manual or VH-IVUS software-based diagnosis is time-consuming, automated plaque characterization tools are essential for accurate atherosclerosis detection and classification. Recently, deep learning (DL) and computer vision (CV) approaches are promising tools for automatically classifying plaques on IVUS images. With this motivation, this manuscript proposes an automated atherosclerotic plaque classification method using a hybrid Ant Lion Optimizer with Deep Learning (AAPC-HALODL) technique on IVUS images. The AAPC-HALODL technique uses the faster regional convolutional neural network (Faster RCNN)-based segmentation approach to identify diseased regions in the IVUS images. Next, the ShuffleNet-v2 model generates a useful set of feature vectors from the segmented IVUS images, and its hyperparameters can be optimally selected by using the HALO technique. Finally, an average ensemble classification process comprising a stacked autoencoder (SAE) and deep extreme learning machine (DELM) model can be utilized. The MICCAI Challenge 2011 dataset was used for AAPC-HALODL simulation analysis. A detailed comparative study showed that the AAPC-HALODL approach outperformed other DL models with a maximum accuracy of 98.33%, precision of 97.87%, sensitivity of 98.33%, and F score of 98.10%.

Deep learning-based fully automatic screening of carotid artery plaques in computed tomography angiography: a multicenter study

AIM

To develop and validate a deep learning (DL) algorithm for the automated detection and classification of carotid artery plaques (CAPs) on computed tomography angiography (CTA) images.

MATERIALS AND METHODS

This retrospective study enrolled 400 patients (300 in the Center Ⅰ and 100 in Ⅱ). Three radiologists co-labeled CAPs, and their revised calcification status (noncalcified, mixed, and calcified) was regarded as ground truth. Center Ⅰ patients were randomly divided into training and internal validation datasets, while Center Ⅱ patients served as the external validation dataset. Carotid artery regions were segmented using a modified 3D-UNet network, followed by CAPs detection and classification using a ResUNet-based architecture in a two-step DL system. The DL model's detection and classification performance were evaluated on the validation dataset using precision-recall curve, free-response receiver operating characteristic (fROC) curve, Cohen's kappa, and ROC curve analysis.

RESULTS

The DL model had achieved 83.4% sensitivity at 3.0 false positives (FPs)/CTA scan in internal validation and 78.9% in external validation. F1-scores were 0.764 and 0.769 at the optimal threshold, and area under fROC curves were 0.756 and 0.738, respectively, indicating good overall accuracy for CAP detection. The DL model also showed good performance for the ternary classification of CAPs, with Cohen's kappa achieved 0.728 and 0.703 in both validation datasets.

CONCLUSION

This study demonstrated the feasibility of using a fully automated DL-based algorithm for the detection and ternary classification of CAPs, which could be helpful for the workloads of radiologists.

Contrast-enhanced ultrasound reveals free-floating thrombus in carotid artery: The cause of stroke is surprisingly plaque rupture

BACKGROUND

Acute stroke poses a serious threat to people's health. The occurrence of a thrombus following the rupture of vulnerable plaques in the carotid artery is a significant contributor to the development of stroke. In previous case reports, it has been challenging to visualize tiny ulcerations within carotid artery plaques using computed tomography angiography (CTA) and digital subtraction angiography (DSA), even when the rupture of the plaque leads to the formation of a free-floating thrombus (FFT). However, in this particular case, contrast-enhanced ultrasound (CEUS) was able to overcome this limitation and provide a more precise assessment, confirming that the FFT formation was indeed a result of plaque rupture rather than any other potential causes. Cases that utilize CEUS to visualize the formation of ulcers and FFT resulting from plaque rupture are even more rare. As such, we present this case to shed light on this infrequent phenomenon.

CASE SUMMARY

In this case study, we present a 65-year-old male patient who was admitted to the hospital due to headache and abnormal mental behavior for one day. During the routine cervical artery ultrasound examination upon admission, we detected the presence of plaque in the right internal carotid artery of the patient, resulting in luminal stenosis. Additionally, we observed suspected hypoechoic material at the distal end of the plaque. After undergoing CEUS examination, it was definitively determined that an ulcer had formed and a FFT had developed due to the rupture of carotid artery plaque. Subsequent CTA and DSA examinations further confirmed the presence of the FFT. The magnetic resonance imaging (MRI) reveals an acute lacunar infarction in the head of the right caput nuclei caudate, which strengthens the potential link between the patient's neurological and psychiatric symptoms observed during admission. The patient received prompt antiplatelet therapy and underwent cervical artery stenting surgery with the assistance of a distal embolic protection device. Following the procedure, the patient was discharged on the fourth day and experienced a complete recovery.

CONCLUSION

CEUS is a valuable tool for visualizing FFT resulting from the rupture of vulnerable plaques in the carotid artery.

Radiomics in Carotid Plaque: A Systematic Review and Radiomics Quality Score Assessment

Imaging modalities provide information on plaque morphology and vulnerability; however, they are operator dependent and miss a great deal of microscopic information. Recently, many radiomics models for carotid plaque that identify unstable plaques and predict cardiovascular outcomes have been proposed. This systematic review was aimed at assessing whether radiomics is a reliable and reproducible method for the clinical prediction of carotid plaque. A systematic search was conducted to identify studies published in PubMed and Cochrane library from January 1, 2001, to September 30, 2022. Both retrospective and prospective studies that developed and/or validated machine learning models based on radiomics data to classify or predict carotid plaques were included. The general characteristics of each included study were selected, and the methodological quality of radiomics reports and risk of bias were evaluated using the radiomics quality score (RQS) tool and Quality Assessment of Diagnostic Accuracy Studies-2, respectively. Two investigators independently reviewed each study, and the consensus data were used for analysis. A total of 2429 patients from 16 studies were included. The mean area under the curve of radiomics models for diagnostic or predictive performance of the included studies was 0.88 ± 0.02, with a range of 0.741-0.989. The mean RQS was 9.25 (standard deviation: 6.04), representing 25.7% of the possible maximum value of 36, whereas the lowest point was -2, and the highest score was 22. Radiomics models have revealed additional information on patients with carotid plaque, but with respect to methodological quality, radiomics reports are still in their infancy, and many hurdles need to be overcome.

Artificial intelligence-based predictive models in vascular diseases

Cardiovascular disease represents a source of major health problems worldwide, and although medical and technical advances have been achieved, they are still associated with high morbidity and mortality rates. Personalized medicine would benefit from novel tools to better predict individual prognosis and outcomes after intervention. Artificial intelligence (AI) has brought new insights to cardiovascular medicine, especially with the use of machine learning techniques that allow the identification of hidden patterns and complex associations in health data without any a priori assumptions. This review provides an overview on the use of artificial intelligence-based prediction models in vascular diseases, specifically focusing on aortic aneurysm, lower extremity arterial disease, and carotid stenosis. Potential benefits include the development of precision medicine in patients with vascular diseases. In addition, the main challenges that remain to be overcome to integrate artificial intelligence-based predictive models in clinical practice are discussed.

Predicting the Efficacy of Neoadjuvant Chemotherapy for Pancreatic Cancer Using Deep Learning of Contrast-Enhanced Ultrasound Videos

Contrast-enhanced ultrasound (CEUS) is a promising imaging modality in predicting the efficacy of neoadjuvant chemotherapy for pancreatic cancer, a tumor with high mortality. In this study, we proposed a deep-learning-based strategy for analyzing CEUS videos to predict the prognosis of pancreatic cancer neoadjuvant chemotherapy. Pre-trained convolutional neural network (CNN) models were used for binary classification of the chemotherapy as effective or ineffective, with CEUS videos collected before chemotherapy as the model input, and with the efficacy after chemotherapy as the reference standard. We proposed two deep learning models. The first CNN model used videos of ultrasound (US) and CEUS (US+CEUS), while the second CNN model only used videos of selected regions of interest (ROIs) within CEUS (CEUS-ROI). A total of 38 patients with strict restriction of clinical factors were enrolled, with 76 original CEUS videos collected. After data augmentation, 760 and 720 videos were included for the two CNN models, respectively. Seventy-six-fold and 72-fold cross-validations were performed to validate the classification performance of the two CNN models. The areas under the curve were 0.892 and 0.908 for the two models. The accuracy, recall, precision and F1 score were 0.829, 0.759, 0.786, and 0.772 for the first model. Those were 0.864, 0.930, 0.866, and 0.897 for the second model. A total of 38.2% and 40.3% of the original videos could be clearly distinguished by the deep learning models when the naked eye made an inaccurate classification. This study is the first to demonstrate the feasibility and potential of deep learning models based on pre-chemotherapy CEUS videos in predicting the efficacy of neoadjuvant chemotherapy for pancreas cancer.

Value of deep learning models based on ultrasonic dynamic videos for distinguishing thyroid nodules

Objective

This study was designed to distinguish benign and malignant thyroid nodules by using deep learning(DL) models based on ultrasound dynamic videos.

Methods

Ultrasound dynamic videos of 1018 thyroid nodules were retrospectively collected from 657 patients in Zhejiang Cancer Hospital from January 2020 to December 2020 for the tests with 5 DL models.

Results

In the internal test set, the area under the receiver operating characteristic curve (AUROC) was 0.929(95% CI: 0.888,0.970) for the best-performing model LSTM Two radiologists interpreted the dynamic video with AUROC values of 0.760 (95% CI: 0.653, 0.867) and 0.815 (95% CI: 0.778, 0.853). In the external test set, the best-performing DL model had AUROC values of 0.896(95% CI: 0.847,0.945), and two ultrasound radiologist had AUROC values of 0.754 (95% CI: 0.649,0.850) and 0.833 (95% CI: 0.797,0.869).

Conclusion

This study demonstrates that the DL model based on ultrasound dynamic videos performs better than the ultrasound radiologists in distinguishing thyroid nodules.

Deep learning based on carotid transverse B-mode scan videos for the diagnosis of carotid plaque: a prospective multicenter study

OBJECTIVES

Accurate detection of carotid plaque using ultrasound (US) is essential for preventing stroke. However, the diagnostic performance of junior radiologists (with approximately 1 year of experience in carotid US evaluation) is relatively poor. We thus aim to develop a deep learning (DL) model based on US videos to improve junior radiologists' performance in plaque detection.

METHODS

This multicenter prospective study was conducted at five hospitals. CaroNet-Dynamic automatically detected carotid plaque from carotid transverse US videos allowing clinical detection. Model performance was evaluated using expert annotations (with more than 10 years of experience in carotid US evaluation) as the ground truth. Model robustness was investigated on different plaque characteristics and US scanning systems. Furthermore, its clinical applicability was evaluated by comparing the junior radiologists' diagnoses with and without DL-model assistance.

RESULTS

A total of 1647 videos from 825 patients were evaluated. The DL model yielded high performance with sensitivities of 87.03% and 94.17%, specificities of 82.07% and 74.04%, and areas under the receiver operating characteristic curve of 0.845 and 0.841 on the internal and multicenter external test sets, respectively. Moreover, no significant difference in performance was noted among different plaque characteristics and scanning systems. Using the DL model, the performance of the junior radiologists improved significantly, especially in terms of sensitivity (largest increase from 46.3 to 94.44%).

CONCLUSIONS

The DL model based on US videos corresponding to real examinations showed robust performance for plaque detection and significantly improved the diagnostic performance of junior radiologists.

KEY POINTS

• The deep learning model based on US videos conforming to real examinations showed robust performance for plaque detection. • Computer-aided diagnosis can significantly improve the diagnostic performance of junior radiologists in clinical practice.

Artificial intelligence for the evaluation of peripheral artery disease using arterial Doppler waveforms to predict abnormal ankle-brachial index

BACKGROUND

Patients with peripheral artery disease (PAD) are at increased risk for major adverse limb and cardiac events including mortality. Developing screening tools capable of accurate PAD identification is a necessary first step for strategies of adverse outcome prevention. This study aimed to determine whether machine analysis of a resting Doppler waveform using deep neural networks can accurately identify patients with PAD.

METHODS

Consecutive patients (4/1/2015 - 12/31/2020) undergoing rest and postexercise ankle-brachial index (ABI) testing were included. Patients were randomly allocated to training, validation, and testing subsets (70%/15%/15%). Deep neural networks were trained on resting posterior tibial arterial Doppler waveforms to predict normal (> 0.9) or PAD (⩽ 0.9) using rest and postexercise ABI. A separate dataset of 151 patients who underwent testing during a period after the model had been created and validated (1/1/2021 - 3/31/2021) was used for secondary validation. Area under the receiver operating characteristic curves (AUC) were constructed to evaluate test performance.

RESULTS

Among 11,748 total patients, 3432 patients met study criteria: 1941 with PAD (mean age 69 ± 12 years) and 1491 without PAD (64 ± 14 years). The predictive model with highest performance identified PAD with an AUC 0.94 (CI = 0.92-0.96), sensitivity 0.83, specificity 0.88, accuracy 0.85, and positive predictive value (PPV) 0.90. Results were similar for the validation dataset: AUC 0.94 (CI = 0.91-0.98), sensitivity 0.91, specificity 0.85, accuracy 0.89, and PPV 0.89 (postexercise ABI comparison).

CONCLUSION

An artificial intelligence-enabled analysis of a resting Doppler arterial waveform permits identification of PAD at a clinically relevant performance level.

The Value of Contrast-Enhanced Ultrasound in the Evaluation of Carotid Web

Objectives

The purpose of this study was to investigate whether contrast-enhanced ultrasound (CEUS) is more advantageous than conventional ultrasound in the diagnosis of carotid web (CaW) and to compare the clinical characteristics of patients in different age groups.

Methods

Seventeen patients admitted to the hospital from October 2019 to December 2021 were included in our study. Patients were initially diagnosed with CaW using digital subtraction angiography (DSA), and conventional ultrasound and CEUS were completed. Baseline patient data were analyzed and compared between the <60 years old CaW group and the ≥60 years old CaW group to explore the differences between the two groups. Then, comparing the accuracy of conventional ultrasound and CEUS.

Results

A total of 17 CaW patients participated in this study, including 4 female patients (23.5%) and 13 male patients (76.5%), with an average age of 59.41 (±10.86) years. There were 9 patients (52.9%) with left CaW and 8 patients (47.1%) with right CaW. Acute ischemic stroke (AIS) occurred in 14 patients (82.4%). Thrombosis occurred in five of 17 patients (29.4%). There was a significant statistical difference about the thrombosis between the <60 years old CaW group and the ≥60 years old CaW group [<60 years group: 0 (0%), ≥60 years group: 5 (62.5%), P = 0.005]. Seven patients (41.2%) received medical management, nine patients (52.9%) had carotid artery stenting (CAS), and one patient (5.9%) had carotid endarterectomy (CEA). None of the patients had recurrent stroke during the follow-up period. The diagnostic rate of CaW and thrombus by CEUS was higher than that by conventional ultrasound, and there was a significant statistical difference in the diagnosis of thrombus between CEUS and conventional ultrasound (χ2 = 4.286, P = 0.038).

Conclusions

CEUS may have a higher diagnostic accuracy for CaW with thrombosis, and it has a higher clinical application prospect.