Hyuntaek Lee

Enhancing fullchip ILT mask synthesis capability for IC manufacturability

It is well known in the industry that the technology nodes from 30nm and below will require model based SRAF / OPC for critical layers to meet production required process windows. Since the seminal paper by Saleh and Sayegh[1][2] thirty years ago, the idea of using inverse methods to solve mask layout problems has been receiving increasing attention as design sizes have been steadily shrinking. ILT in its present form represents an attempt to construct the inverse solution to a constrained problem where the constraints are all possible phenomena which can be simulated, including: DOF, sidelobes, MRC, MEEF, EL, shot-count, and other effects. Given current manufacturing constraints and process window requirements, inverse solutions must use all possible degrees of freedom to synthesize a mask. Various forms of inverse solutions differ greatly with respect to lithographic performance and mask complexity. Factors responsible for their differences include composition of the cost function that is minimized, constraints applied during optimization to ensure MRC compliance and limit complexity, and the data structure used to represent mask patterns. In this paper we describe the level set method to represent mask patterns, which allows the necessary degrees of freedom for required lithographic performance, and show how to derive Manhattan mask patterns from it, which can be manufactured with controllable complexity and limited shot-counts. We will demonstrate how full chip ILT masks can control e-beam write-time to the level comparable to traditional OPC masks, providing a solution with maximized lithographic performance and manageable cost of ownership that is vital to sub-30nm node IC manufacturing.

Automated measurement of fetal myocardial performance index in ultrasound Doppler waveforms

We introduce an automated method for myocardial performance index (MPI), also known as Tei index, which is one of the most substantial indicators in the early screening of heart defects. Since assessing fetal cardiac functions using MPI has become a routine and significant process, there have been explicit requirements to automate MPI measurements. Due to small heart sizes of fetuses, we focus on the automation of modified MPI (Mod-MPI) which uses a single Doppler gate. The proposed method detects four valve click signals in Doppler waveforms using four image features which are extracted by vertical projection of Doppler waveforms after several transformations. To evaluate performance, 88 of fetal examinations were collected from a commercial ultrasound machine, and two clinical experts measured the Mod-MPI both manually and automatically. Quantitative comparisons based on intra-class correlation coefficients (ICC) yield that intra-observer reproducibility is higher when performing the proposed method (ICC=0.951 and 0.932 for observer 1 and 2) comparing to those of manual measurements (ICC=0.868 and 0.857 for observer 1 and observer 2). Thus, our method (ICC=0.962) reveals superior inter-observer reproducibility than that of manual method (ICC=0.597). Although mean difference from observer 2 (−0.062) is over three times larger than that of observer 1 (−0.018) due to different experiences, both of mean differences are acceptable. In conclusion, the proposed MPI measurement method can improve intra- and inter-reproducibility while providing reliable results.

Streak artifact reduction for blind deconvolution of multi-beam image

In ultrasound imaging, blind deconvolution technique is used to enhance image resolution. However, it causes vertical streak artifact. In this paper, scanline grouping method for reducing streak artifact is proposed, based on the assumption that the vertical streak artifact is due to the difference of the scanline characteristics. The input scanlines with similar characteristics are grouped together, and then blind deconvolution is performed separately for each group. For performance evaluation, we used 75 in-vivo abdomen cases of 2 multi-beam and 36 in-vivo thyroid cases of RF interpolation images obtained with convex and linear probes. When evaluated by correlation between neighboring scanlines, the proposed method was able to reduce streak artifact, as correlation value was approximately 10% higher than the conventional method. In addition, processing time of the proposed method was approximately 5 times faster than the conventional method. The proposed scanline grouping method is expected to work effectively with other adaptive or iterative signal processing techniques, such as 2D or lateral direction filters.

Ultrasound semi-automated measurement of fetal nuchal translucency thickness based on principal direction estimation

The objective of the paper is to introduce a novel method for nuchal translucency (NT) boundary detection and thickness measurement, which is one of the most significant markers in the early screening of chromosomal defects, namely Down syndrome. To improve the reliability and reproducibility of NT measurements, several automated methods have been introduced. However, the performance of their methods degrades when NT borders are tilted due to varying fetal movements. Therefore, we propose a principal direction estimation based NT measurement method to provide reliable and consistent performance regardless of both fetal positions and NT directions. At first, Radon Transform and cost function are used to estimate the principal direction of NT borders. Then, on the estimated angle bin, i.e., the main direction of NT, gradient based features are employed to find initial NT lines which are beginning points of the active contour fitting method to find real NT borders. Finally, the maximum thickness is measured from distances between the upper and lower border of NT by searching along to the orthogonal lines of main NT direction. To evaluate the performance, 89 of in vivo fetal images were collected and the ground-truth database was measured by clinical experts. Quantitative results using intraclass correlation coefficients and difference analysis verify that the proposed method can improve the reliability and reproducibility in the measurement of maximum NT thickness.

Multi-scale synthetic filtering method for ultrasonic image enhancement

In this paper, a novel ultrasonic image enhancement method, i.e., multi-scale synthetic filtering (MSF) method is proposed for effectively suppressing speckle noise and enhancing structured region. The proposed MSF method is composed of the modified nonlinear coherence diffusion (NCD) model and the adaptive directional filter (ADF) both applied in multi-scale domain. Due to the difference in frequency characteristics of speckle and edge, Laplacian pyramid is adopted to analyze ultrasonic images. On each Laplacian sub-band, the modified NCD is performed to suppress speckle noise while preserving anatomical edges. Then, the ADF is applied to each synthesized images to sharpen and enhance detailed boundaries. The performance of the proposed MSF method is compared with the speckle reducing anisotropic diffusion (SRAD). For quantitative analysis, the contrast-to-noise ratio (CNR) values were computed on in vivo carotid artery ultrasound images. When compared with the SRAD, the average CNR gain of the proposed MSF method was 0.42 dB. The average processing time of the proposed method is 8.1 ms per one frame. The clinical evaluations and short running time indicate that the proposed MSF method can improve the diagnostic confidence in medical ultrasound imaging.