Over the last few years, major breakthroughs were achieved in the application of deep learning in many computer vision tasks, such as image classification and segmentation. The automatic liver segmentation from CT images has become an important area in clinical research, including radiotherapy, liver volume measurement, and liver transplant surgery. This paper proposes a novel convolutional neural network for liver segmentation (CNN-LivSeg) algorithm that involves three convolutional (each convolutional layer followed by max-pooling layer) and two fully connected layers with a final 2- way softmax is used for liver discrimination. The weight initialization is based on a random Gaussian, which performed a distance preserving-embedding of the data. To avoid using the fully 3D CNN network which is computationally expensive and time-consuming, 2D patches were extracted and processed for segmentation. Experiments were performed on MICCAI-SLiver07 as a benchmark dataset. The mean ratios of Dice similarity coefficient, Jaccard similarity index, accuracy, specificity, and sensitivity were 0.9541, 0.9122, 0.9725, 0.9904, and 0.9652, respectively, thereby suggesting that the proposed method performed well on the test images.
Vascular anastomosis, connecting two vessel ends together, is the foundation of plastic and reconstructive surgery. Every living tissue transplantation requires vascular anastomosis. While optical coherence tomography (OCT) imaging can provide objective information for intraoperative evaluation, currently it suffers from limited imaging penetration depth for relative large vessels, which will result in information loss at the bottom region of vessel. In this work, we designed a multi-view scanning scheme for the vessel imaging to increase the FOV effectively. To push the clinical translation, we used MEMS mirror to steer the beam and made a miniature handheld probe for testing. Preliminary results on IR sensing card, multilayer scotch tape, and plastic tube imaging showed the performance of our probe.
A designed star sensor must be extensively tested before launching. Testing star sensor requires complicated process
with much time and resources input. Even observing sky on the ground is a challenging and time-consuming job,
requiring complicated and expensive equipments, suitable time and location, and prone to be interfered by weather. And
moreover, not all stars distributed on the sky can be observed by this testing method. Semi-physical simulation in
laboratory reduces the testing cost and helps to debug, analyze and evaluate the star sensor system while developing the
model. The test system is composed of optical platform, star field simulator, star field simulator computer, star sensor
and the central data processing computer. The test system simulates the starlight with high accuracy and good
parallelism, and creates static or dynamic image in FOV (Field of View). The conditions of the test are close to
observing real sky. With this system, the test of a micro star tracker designed by Beijing University of Aeronautics and
Astronautics has been performed successfully. Some indices including full-sky autonomous star identification time,
attitude update frequency and attitude precision etc. meet design requirement of the star sensor. Error source of the
testing system is also analyzed. It is concluded that the testing system is cost-saving, efficient, and contributes to
optimizing the embed arithmetic, shortening the development cycle and improving engineering design processes.
In this paper, we have developed a model-based approach to match two X-ray angiograms from different views. Under
the guidance of the prior knowledge of anatomic structure of human coronary vessels, this method can build a node
attribute table and assign unique anatomic labels to coronary arteries in X-ray angiograms automatically by the
father-son relationship of the nodes, which is essential in reconstruction of vessels.
We constructed a computer controlled high stability wide tunable parametric amplifier, and reported on a (beta) - borium borate (BBO) optical parametric amplifier of injection-seeded with a narrow linewidth pulsed Ti:sapphire laser. The pump source of Ti:sapphire laser is the residual second harmonic (532 nm) in frequency tripling (355 nm) of the Nd:YAG laser. We obtained the linewidth interferograph of Ti:sapphire laser is less than 0.003 nm, and tested six times curves of the output energy as a function of tunable wavelengths in injection seeding more than that of no injection. The narrow linewidth (<0.1 nm) continuous tunable range of 570 - 670 nm is achieved.