PENGEMBANGAN SKEMA PENGOLAHAN CITRA USG MODUS-B UNTUK PENINGKATAN VISIBILITAS JARUM SPINAL TERHADAP VARIABEL FREKUENSI TRANSDUSER

• Main Author: Emir Muchamad, Ghiffari
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/57382
• Institution: Institut Teknologi Bandung

Minimally invasive procedures are surgical procedures that involve inserting an intermediary device through small incisions to minimize injury and speed up the patient's recovery process, such as the use of needles in anesthetic procedures. The use of ultrasonography (USG) as a feedback tool is expected to increase the success of the procedure. However, in practice, anesthesia procedures with ultrasound aids have problems, one of which is the inconsistency of needle visibility. In a previous study, an image processing scheme was tested on a water phantom with one type of needle in various depths and angles of the needle. The previous processing scheme includes the selection of ROI around the needle tip on the ultrasound image. In the detected area, needle angle estimation for image rotation is performed. Then on the rotated image, the needle position is estimated using a linear derivative. Needle position data is done by removing outlier with a combined method of moving median and moving median absolute deviation. Data without outliers is interpolated by polynomial so that the needle is detected on the ultrasound image. In this final project, the increased visibility of spinal needles carried out this time includes the use of silk tofu phantom medium and two stages of image processing. The schematic of the image processing was carried out in two stages and the evaluation of the image processing performance was carried out on a spinal needle insertion simulator on a phantom made of silken tofu. The selection of silk tofu phantom test medium was carried out because the medium has homogeneous properties and its density is greater than that of water phantom. In B-mode ultrasound imaging on silken tofu phantom, there are artifact problems that also occur in the case of needle sticking in soft tissue, which can also be simulated, namely acoustic propagation velocity artifacts and reflection artifacts. With the image processing scheme developed in previous studies, the condition of the artifact causes false detection. Therefore, in this study, an additional stage was added, namely pre-processing, in which artifacts that appear in the boundary plane will be removed. Meanwhile, other artifacts that arise due to the interaction of waves with the test medium and needles will be a challenge for the scheme in post-processing. The needle detection algorithm used is an algorithm that has been carried out in previous studies and modified. The modification made to this scheme is to remove the limitation of needle detection only in the area with the highest intensity which was previously assumed to be the tip of the needle. This is done on the grounds that there are several cases of imagery where the area of the highest intensity is not located at the tip of the needle, but at the stem of the needle. The designed scheme was tested on two types of spinal needles with a size of 25 G and varied with three needle insertion angles and four transducer frequencies, namely 5MHz, 6.5MHz, 7.5MHz, and 10MHz. The results of the detection of both types of needles with frequency variations have the highest success rate at a frequency of 5MHz. The results of the detection of both types of needles with variations in angles have the highest success rate on images with an angle of 90°. Meanwhile, from all experiments of the two types of needles, the SN-I needle type had a success rate of 50%, while the SN-II was 33.3%. The low success in SN-II is due to the high contrast of the speckle noise and artifacts formed around the needle compared to the contrast of the needle. This shows that different types of needles can affect the level of visibility of the needles.