Every Robotic Assisted Surgeries (RAS) requires some level of navigation. While in open surgery the target is viewed directly, minimally invasive RAS views come from inside the body cavity, with a restricted field-of-view (FOV). Also, the surgeon's hands are occupied with the tools, whereas the camera is controlled by an assistant, adding another complication to the procedure – requiring perfect collaboration between them. Another challenge arises from anatomical and physiological differences between patients which make it difficult to accurately position surgical tools and recognize target organs. In gastroscopies or colonoscopies, the singular wide-angle view is often difficult to interpret, and objective navigational aid can be beneficial.

The model is based on convolutional neural networks, a recent technique which have been proven to be very effective for various types of tasks. In particular, deep neural networks-based models outperform all previous approaches for image segmentation. However, 3D reconstruction from 2D images is still challenging for neural networks, due to the difficulty of representing a dimensional enlargement with standard differentiable layers. Reconstruction of bone surfaces in particular is extremely challenging, due to the transparent nature of the X-ray images. The main usage of the 3D model is for planning and accurate measurements that are needed to precise fit of the implant or for patient specific intraoperative guidance.

Segmentation is highly important both for examination and planning of knee replacement, hip replacement, shoulder surgery, lesion detection, osteotomy and many other orthopedic procedures. RSIP Vision's CTO Ilya Kovler explains how to improve the segmentation in orthopedics with deep learning. Deep learning is repeatedly being proven to be the most powerful framework for various tasks, and segmentation in orthopedics is no exception. Generic out-of-the-box solutions exist and can produce fair results, but carefully crafted and tailored solutions are needed to make the most out of a deep learning approach. Choosing the correct input, selecting the most suitable neural network architecture and incorporating task-specific prior knowledge into the model can all significantly improve the results.

The National Safety Council says at least nine people in the U.S. die and another 100 are injured every day in crashes caused by distracted driving. In-vehicle technologies such as dashboard touchscreens have contributed to this enormous safety threat. But consumers are fond of these technologies and they aren't going away. However, there is a pantheon of other distractions that occur behind the wheel that vary greatly in form and severity. The mostly illegal use of cell phones or texting while driving tops the list.

Artificial intelligence is a huge buzzword these days, especially in the automotive industry. AI has many applications and can mean different things, even within the automotive world. But the general concept can be imagined as a future where human beings have been completely removed from the entire driving equation. This could lead to a vehicular utopia with no more highway accidents, injuries, or deaths-- all the result of driverless cars. However, we are a long way from reaching this point.

Precise medical imaging and analysis could enable early detection of lung cancer, help determine its exact size and location, and significantly improve diagnosis and treatment. This is usually done in a process called segmentation, which uses computers to identify the boundaries of the lung from surrounding thoracic tissue on CT images. From this process, a detailed 3-D map of the airways may be generated that can help to plan and navigate a bronchoscopy procedure to obtain biopsy samples and to perform other clinical interventions. "Until now, this process was very difficult because you need the radiologist, or even the surgeon, to spend much time to understand how to get to the specific place [where the lesion is located]. And this is sometimes prone to error," said Ron Soferman, founder and CEO of RSIP Vision, in an interview with MD DI. "It's very critical [to know the precise location] because, if you miss the lesion, you will take a biopsy from some random part of the lung and it will give a negative result."

Robotic automation can be found in shipment and distribution centers where they perform picking, sorting and packaging activities. These tasks are a natural evolution of automated warehouses – to which robots move in order to store and retrieve items. Identification of objects (aka object recognition) is performed by a machine learning classifier along with OCR recognition of the attached text. Since the objects are not necessarily located and oriented in the optimal way for the robot's cameras – a CNN classifier is used to identify clues of textual and visual features. Mobile robots navigate in a space where other robots are operating, thus requiring real-time update of their environment perception.

The automatic extraction of features from images is then called to tag the scene in a manner which is close to the human perception of it. For example, image features might include landscape at dusk or broad day time, human or animal interactions, textured image and so on. To this purpose, computer vision algorithms need first to extract salient features of the image and then cluster them according to semantically meaningful groups. Region-growing techniques are utilize to subdivide the image into non-overlapping regions containing salient features. The edge map of the image (in all color channels) is used to determine stiff boundaries for region growing and peaks of the edge distance-transform are used as initial seeds for the process.