Methodology for fully automated segmentation and plaque characterization in intracoronary optical coherence tomography images

Abstract

Optical coherence tomography (OCT) is a light-based intracoronary imaging modality that provides high-resolution cross-sectional images of the luminal and plaque morphology. Currently, the segmentation of OCT images and identification of the composition of plaque are mainly performed manually by expert observers. However, this process is laborious and time consuming and its accuracy relies on the expertise of the observer. To address these limitations, we present a methodology that is able to process the OCT data in a fully automated fashion. The proposed methodology is able to detect the lumen borders in the OCT frames, identify the plaque region, and detect four tissue types: calcium (CA), lipid tissue (LT), fibrous tissue (FT), and mixed tissue (MT). The efficiency of the developed methodology was evaluated using annotations from 27 OCT pullbacks acquired from 22 patients. High Pearson's correlation coefficients were obtained between the output of the developed methodology and the manual annotations (from 0.96 to 0.99), while no significant bias with good limits of agreement was shown in the Bland-Altman analysis. The overlapping areas ratio between experts' annotations and methodology in detecting CA, LT, FT, and MT was 0.81, 0.71, 0.87, and 0.81, respectively.

Spatial:Frequency Approach to Fibrous Tissue Classification in Intracoronary Optical Images

“…Athanasiou et al [10] developed an automated method based on segmentation and classification using K-means and achieved 87% and 0.09mm 2 for sensitivity and MADA respectively, whereas we obtained 80% and 1.1mm 2 for these measures. Despite not having a better result comparing with [10], we believe these results can be improved with the incorporation of the attenuation coefficients into the methodology proposed in [11].…”

Parameter estimation of atherosclerotic tissue optical properties from three-dimensional intravascular optical coherence tomography

“…37,38 Ughi et al 38 used μ t estimates from a layer model applied to single A-lines and two-dimensional (2-D) texture and geometric measures as features for classification. Athanasiou et al 37 used 2-D texture and intensity features alone and assumed "islands" of calcified tissue surrounded by other tissue types, which is not necessarily true in many image frames. These reports are encouraging and show that improved 3-D estimation of optical properties could improve the classification.…”

Wavelet Analysis for Multiresolution Tissue Characterization in Intracoronary Optical Images

“…An advantage of the proposed wavelet method over works [6,10] is that it does not require a training stage, and we believe our results can be improved with a parameters tuning for scalogram binarization. Moreover, the multiresolution aspect of wavelets can be also used for lipid and calcium plaque characterization.…”

“…Ughi et al [6] developed an automated tissue characterization based on texture analysis, attenuation coefficient and pixel classification using Random Forest, reaching an accuracy of 89.5% for fibrotic tissue, which is lower than our method (99.6%). Athanasiou et al [10] developed an automated method based on segmentation and classification using K-means and achieved 87% and 0.09mm 2 for sensitivity and MADA respectively, better than our solution.…”