Abstract:

Image segmentation plays an increasingly important role in image processing. It allows for various applications including the analysis of an image for automatic image understanding and the integration of complementary data. During vascular surgeries, the blood flow in the vessels has to be checked constantly, which could be facilitated by a segmentation of the affected vessels. The segmentation of medical images is still done manually, which depends on the surgeon’s experience and is time-consuming. As a result, there is a growing need for automatic image segmentation methods. We propose an unsupervised method to detect the regions of no interest (RONI) in intraoperative images with low depth-of-field (DOF). The proposed method is divided into three steps. First, a color segmentation using a clustering algorithm is performed. In a second step, we assume that the regions of interest (ROI) are in focus whereas the RONI are unfocused. This allows us to segment the image using an edge-based focus measure. Finally, we combine the focused edges with the color RONI to determine the final segmentation result. When tested on different intraoperative images of aneurysm clipping surgeries, the algorithm is able to segment most of the RONI not belonging to the pulsating vessel of interest. Surgical instruments like the metallic clips can also be excluded. Although the image data for the validation of the proposed method is limited to one intraoperative video, a proof of concept is demonstrated.

Abstract:

Imaging systems are increasingly used in medical practice to improve the quality and accuracy of diagnosis and therapy. Although the data acquired reveals useful information, further processing is often needed to provide a comprehensive and valuable representation. As most processing should only be applied to the regions of interest (ROI) in the image, a previous segmentation step is necessary. For algorithms containing spatial low-pass filtering, a prior segmentation prevents the loss of information at the borders of different regions. Nowadays, the segmentation of medical images is mainly performed manually, which is extremely time-consuming. Additionally, the segmentation outcome highly depends on the experience of the clinician. For this reason, there exists an increasing demand for unsupervised image segmentation. In this work, a method to automatically detect the regions of no interest (RONI) in medical sceneries is proposed. The algorithm is aimed at images of surgical procedures on intracranial aneurysms, in which a low depth-of-field (DOF) is induced by the deep access channel and the high magnification of the surgical microscope. The developed approach consists of three parts. First, a color segmentation based on a clustering algorithm is applied. It identifies the RONI with colors different from the vessel of interest. Second, a focus segmentation is used to detect all RONI that are out of focus. This is realized by determining the focused edges in the image and enclosing the ROI by linking those edges. In the last part, the results from both previously applied methods are combined to obtain the final segmentation result. A proof of concept for the proposed method combining color and focus segmentation is presented in this work. The validation on six intraoperative images of aneurysm clipping surgeries shows a high sensitivity and specificity of the algorithm. In comparison to a separate color or focus segmentation, the combination of those methods is demonstrated to increase the accuracy of the segmentation results. An extensive validation of the method developed in this thesis could not be performed due to a lack of sufficient image data.