Identification of high-risk plaque features in intracranial atherosclerosis: initial experience using a radiomic approach

Abstract

ObjectivesTo evaluate a quantitative radiomic approach based on high-resolution magnetic resonance imaging (HR-MRI) to differentiate acute/sub-acute symptomatic basilar artery plaque from asymptomatic plaque.MethodsNinety-six patients with basilar artery stenosis underwent HR-MRI between January 2014 and December 2016. Patients were scanned with T1- and T2-weighted imaging, as well as T1 imaging following gadolinium-contrast injection (CE-T1). The stenosis value, plaque area/burden, lumen area, minimal luminal area (MLA), intraplaque haemorrhage (IPH), contrast enhancement ratio and 94 quantitative radiomic features were extracted and compared between acute/sub-acute and asymptomatic patients. Multi-variate logistic analysis and a random forest model were used to evaluate the diagnostic performance.ResultsIPH, MLA and enhancement ratio were independently associated with acute/subacute symptoms. Radiomic features in T1 and CE-T1 images were associated with acute/subacute symptoms, but the features from T2 images were not. The combined IPH, MLA and enhancement ratio had an area under the curve (AUC) of 0.833 for identifying acute/sub-acute symptomatic plaques, and the combined T1 and CE-T1 radiomic approach had a significantly higher AUC of 0.936 (p = 0.01). Combining all features achieved an AUC of 0.974 and accuracy of 90.5%.ConclusionsRadiomic analysis of plaque texture on HR-MRI accurately distinguished between acutely symptomatic and asymptomatic basilar plaques.Key Points • High-resolution magnetic resonance imaging can assess basilar artery atherosclerotic plaque. • Radiomic features in T1 and CE-T1 images are associated with acute symptoms. • Radiomic analysis can accurately distinguish between acute symptomatic and asymptomatic plaque. • The highest accuracy may be achieved by combining radiomic and conventional features. Electronic supplementary materialThe online version of this article (10.1007/s00330-018-5395-1) contains supplementary material, which is available to authorized users.

Evaluation of the distribution and progression of intraluminal thrombus in abdominal aortic aneurysms using high‐resolution MRI

“…Second, no gross or histologic validation of ILT macrostructure and biochemical composition was performed in this cohort; such validation could strengthen our hypotheses regarding the etiology of signal changes, and shed light on the underlying biochemomechanics that govern the complex and related evolution of aortic aneurysms and intraluminal thrombus. Third, although we provided both qualitative and quantitative descriptions of ILT signal and thrombus burden, future work including quantification of 3D morphology and texture could improve the characterization of ILT from the rich 3D datasets . Future work may also incorporate more advanced imaging methods including MRI and PET/CT markers of inflammation, providing a more comprehensive evaluation of both ILT and the underlying vessel wall .…”

“…Third, although we provided both qualitative and quantitative descriptions of ILT signal and thrombus burden, future work including quantification of 3D morphology and texture could improve the characterization of ILT from the rich 3D datasets. 12,24 Future work may also incorporate more advanced imaging methods including MRI and PET/CT markers of inflammation, providing a more comprehensive evaluation of both ILT and the underlying vessel wall. [25][26][27][28] Additionally, different ILT signal intensity may correlate with different mechanical characteristics, 9,29 which could better inform computational modeling of AAA mechanics.…”

Radiomics versus Visual and Histogram-based Assessment to Identify Atheromatous Lesions at Coronary CT Angiography: An ex Vivo Study

“…Radiomics has been shown to identify napkin-ring sign plaques with excellent diagnostic accuracy ( 15 ). Similarly, MRI may also be used to classify atherosclerotic lesions, and radiomic analysis of images from MRI has been shown to differentiate between acute or subacute symptomatic and asymptomatic plaques in the basilar artery ( 32 – 34 ). Furthermore, ML has proven to be a valuable tool in medical data analysis ( 35 , 36 ), identifying insights from big data databases by using alternative statistical techniques.…”

Preoperative Noninvasive Radiomics Approach Predicts Tumor Consistency in Patients With Acromegaly: Development and Multicenter Prospective Validation

“…Textural features have also been depicted as patterns or spatial distributions of voxel intensity, and they can be computed from the GLCM (46). The voxel intensity values in the volume of interest are required to determine the representative texture matrix (47), and such a step can reduce the image noise and normalize the intensities among all patients. Thus, it is possible to directly compare all of the computed textural features among different patients.…”