Feature Of The Week 1/20/13: CWRU Demonstrates Automatic Stent Detection in Intravascular OCT Images Using Bagged Decision Trees
Cardiovascular disease is the leading cause of death worldwide. Stent implantation by means of percutaneous coronary intervention is the most common coronary revascularization procedure. Intravascular Optical Coherence Tomography (iOCT) is the only imaging modality with the resolution and contrast necessary to enable accurate measurements of luminal architecture and neointima stent coverage. Manual analysis of intravascular OCT pullbacks is time consuming, limiting the size and number of studies that can be performed.
We developed a highly automated method for detecting stent struts and measuring tissue coverage. Candidate struts were first identified using image processing techniques. We trained a bagged decision trees classifier to classify candidate struts using physically meaningful features extracted from the images. With 12 best features identified by forward feature selection, recall and precision were 90~94% and 85~90%, respectively. Including struts deemed insufficiently bright for manual analysis, precision improved to 94%. Strut detection statistics approached variability of manual analysis. Periodic cubic spline was used to construct stent contour to enable tissue coverage area measurement. Differences between manual and automatic area measurements were 0.12±0.20mm2 and 0.11±0.20mm2 for stent and tissue areas, respectively. With proposed algorithms, analyst time per stent should significantly reduce from the 6-16 hours now required.
For more information see recent Article. Courtesy Hong Lu from Case Western Reserve University.