Computer Vision News - December 2019

Projects by RSIP Vision 40 RSIP Vision boasts extensive experience across a range of medical segmentation projects: CT, MRI, fluorescent, ultrasound, OCT, laparoscopy images, endoscopy and more. All this work is based on deep learning technology , which offers much more robust solutions. Let ’s have a look at three projects in particular. The first project deals with airway segmentation for the purpose of bronchoscopy surgical intervention . This segmentation helps during pre-operation planning. In addition, it assists the surgeon in planning the optimal path to reach the lesion, for instance, to extract a tissue for biopsy or some other form of treatment. Since the catheter does not carry a camera, the surgeonmust findanother method to guide the navigation. This is generally performed with the help of an electro-magnetic sensor placed at the tip of the catheter. In such cases, the surgeon only knows information gathered through fluoroscopic imaging or by the position of the catheter in the electromagnetic field. The previously designed segmentation places a strong constraint on the position of the catheter in the CT used in pre-op planning. The second project involves joints segmentation , a crucial process in developing a solution specific to the patient ’s body. This segmentation produces a patient specific implant based on the most accurate size and angles. In some cases, a 3D print of the specific implant integrates with the precise anatomy of the patient. The third project focuses on lung nodules . RSIP Vision developed a system called AUTORECIST , an automated system capable of calculating the volume of nodules and their evolution during and after treatment. The radiologist measures the volume of nodules at the time of the diagnosis. This measure is then compared over time to assess the progress of the nodule in reaction tothe treatment and to monitor whether new nodules have sprouted. Segmentation provides very precious information: the shape and the size of the nodule, whichmight reveal whether specific treatment is successful or the need for an alternative treatment. By Ilya Kolver