Abstract. An integrated circuit implementation of a fully parallel analog artificial neural network is presented. We include details of the architecture, some of the important design considerations, a description of the circuits and finally actual performance data. The electrically trainable artificial neural network (ETANN) chip incorporates 64 analog neurons and 10,240 analog synapses and utilizes a 1-gin CMOS NVM process. The network calculates the dot product between a 64-element analog input vector and a 64 × 64 nonvolatile (EEPROM based) analog synaptic weight array. These calculations occur at a rate in excess of 1.3 billion interconnections per second. All elements of the computation are stored and calculated in the analog domain and strictly in parallel. A 2:1 input and neuron multiplex mode permits rates in excess of 2 billion interconnections per second and a single-chip effective network size of 64 inputs by 128 outputs. The ETANN incorporates differential signal techniques throughout for improved noise rejection. Current summing is employed for the sum of products calculations. The chip integrates approximately 400 op amps, including variable gain stages of from 20 to 54 dB. Inevitable component to component variations due to the use of minimum dimension elements are found not to be significant for operation in an adaptive environment.
Line integral convolution (LIC) is used as a texture-based technique in computer graphics for flow field visualization. In diffusion tensor imaging (DTI), LIC bridges the gap between local approaches, for example directionally encoded fractional anisotropy mapping and techniques analyzing global relationships between brain regions, such as streamline tracking. In this paper an advancement of a previously published multikernel LIC approach for high angular resolution diffusion imaging visualization is proposed: a novel sampling scheme is developed to generate anisotropic glyph samples that can be used as an input pattern to the LIC algorithm. Multicylindrical glyph samples, derived from fiber orientation distribution (FOD) functions, are used, which provide a method for anisotropic packing along integrated fiber lines controlled by a uniform random algorithm. This allows two- and three-dimensional LIC maps to be generated, depicting fiber structures with excellent contrast, even in regions of crossing and branching fibers. Furthermore, a color-coding model for the fused visualization of slices from T1 datasets together with directionally encoded LIC maps is proposed. The methodology is evaluated by a simulation study with a synthetic dataset, representing crossing and bending fibers. In addition, results from in vivo studies with a healthy volunteer and a brain tumor patient are presented to demonstrate the method's practicality.
Recently, a fiber visualization method for high-angular resolution diffusion-weighted magnetic resonance imaging (MRI) data was proposed using a multiple-kernel line integral convolution (LIC) algorithm and an anisotropic spot pattern. This processing routine leads to high contrast color-coded LIC maps that are capable of visualizing local anisotropy information and regional fiber architecture. In this paper, we evaluate and validate this method by applying it to simulated datasets and to in vivo diffusion MRI data of children and adults with different disease conditions and healthy volunteers. Compared to routine clinical fiber visualization (color-coded fractional anisotropy, FA maps, and fiber tractography), it has the advantage of visualizing complex local fiber architecture in a fully automated way. The results indicate that this method is capable of reliably delineating normal fiber architecture and fibers infiltrated, displaced, or disrupted by lesions and is therefore a promising tool in the clinical context.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.