Intracoronary optical coherence tomography (OCT) is increasingly being used for real-time visualization of coronary arteries aiming to help in the identification of highrisk atherosclerotic plaques associated with geometrical and morphological features of the arterial wall. This paper presents a framework towards the automatic detection of the inner wall of the coronary artery (lumen-endothelium border) in intracoronary OCT image sequences by employing a multistep image processing method. The major focus of this work was to address difficult cases that are frequently met in intracoronary OCT, e.g. images with small/big branches, multiple branches, blood presence, calcifications, artifacts, etc. We present each step employed and the results obtained both in qualitative and quantitative terms. The proposed segmentation algorithm has been proven very efficient in the majority of the examined cases.