arXiv subjects

Yanjun Hu

Publications and source records attributed to Yanjun Hu.

5 recordsLinked to original sources

arXiv2020

A Natural Language Processing Pipeline of Chinese Free-text Radiology Reports for Liver Cancer Diagnosis

Despite the rapid development of natural language processing (NLP) implementation in electronic medical records (EMRs), Chinese EMRs processing remains challenging due to the limited corpus and specific grammatical characteristics, especially for radiology reports. In this study, we designed an NLP pipeline for the direct extraction of clinically relevant features from Chinese radiology reports, which is the first key step in computer-aided radiologic diagnosis. The pipeline was comprised of named entity recognition, synonyms normalization, and relationship extraction to finally derive the radiological features composed of one or more terms. In named entity recognition, we incorporated lexicon into deep learning model bidirectional long short-term memory-conditional random field (BiLSTM-CRF), and the model finally achieved an F1 score of 93.00%. With the extracted radiological features, least absolute shrinkage and selection operator and machine learning methods (support vector machine, random forest, decision tree, and logistic regression) were used to build the classifiers for liver cancer prediction. For liver cancer diagnosis, random forest had the highest predictive performance in liver cancer diagnosis (F1 score 86.97%, precision 87.71%, and recall 86.25%). This work was a comprehensive NLP study focusing on Chinese radiology reports and the application of NLP in cancer risk prediction. The proposed NLP pipeline for the radiological feature extraction could be easily implemented in other kinds of Chinese clinical texts and other disease predictive tasks.

cs.CL↗

arXiv2018

Channel Estimation with Systematic Polar Codes

Study of polar codes in fading channels is of great importance when applying polar codes in wireless communications. Channel estimation is a fundamental step for communication to be possible in fading channels. For both systematic and non-systematic polar codes, construction of them is based on an information set and the known frozen bits. Efficient implementation of systematic and non-systematic polar codes exists. When it comes to channel estimation or channel tracking, additional pilot symbols are inserted in the codeword traditionally. In this paper, to improve the performance of polar codes in the finite domain, pilot symbols are selected from the coded symbols themselves. In order to keep the existing efficient structure of polar code encoding, pilot selection is critical since not all selections can reuse the existing structure. In this paper, two pilot selections denoted as Uneven Pilot Selection (UEPS) and Even Pilot Selection (EPS) are proposed, which do not change the efficient polar encoding structure. The proposed UEPS and EPS is proven to satisfy the efficient construction condition. The performance of EPS is shown in this paper to outperform both the UEPS and the traditional pilot insertion scheme. Simulation results are provided which verify the performance of the proposed pilot selection schemes.

cs.IT↗

arXiv2016

On the Polar Code Encoding in Fading Channels

Besides the determined construction of polar codes in BEC channels, different construction techniques have been proposed for AWGN channels. The current state-of-the-art algorithm starts with a design-SNR (or an operating SNR) and then processing is carried out to approximate each individual bit channel. However, as found in this paper, for fading channels, an operating SNR can not be directly used in approximating the bit channels. To achieve a better BER performance, the input SNR for the polar code construction in fadding channels is derived. A selection of the design-SNR for both the AWGN and the fading channels from an information theoretical point of view is studied. Also presented in this paper is the study of sacrificing a small data rate to gain orders of magnitude increase in the BER performance.

cs.IT↗

arXiv2016

A Novel Interleaving Scheme for Polar Codes

It's known that the bit errors of polar codes with successive cancellation (SC) decoding are coupled. We call the coupled information bits the correlated bits. In this paper, concatenation schemes are studied for polar codes (as inner codes) and LDPC codes (as outer codes). In a conventional concatenation scheme, to achieve a better BER performance, one can divide all $N_l$ bits in a LDPC block into $N_l$ polar blocks to completely de-correlate the possible coupled errors. In this paper, we propose a novel interleaving scheme between a LDPC code and a polar code which breaks the correlation of the errors among the correlated bits. This interleaving scheme still keeps the simple SC decoding of polar codes while achieves a comparable BER performance at a much smaller delay compared with a $N_l$-block delay scheme.

cs.IT↗

arXiv2015

On the Error Performance of Systematic Polar Codes

Systematic polar codes are shown to outperform non-systematic polar codes in terms of the bit-error-rate (BER) performance. However theoretically the mechanism behind the better performance of systematic polar codes is not yet clear. In this paper, we set the theoretical framework to analyze the performance of systematic polar codes. The exact evaluation of the BER of systematic polar codes conditioned on the BER of non-systematic polar codes involves in $2^{NR}$ terms where $N$ is the code block length and $R$ is the code rate, resulting in a prohibitive number of computations for large block lengths. By analyzing the polar code construction and the successive-cancellation (SC) decoding process, we use a statistical model to quantify the advantage of systematic polar codes over non-systematic polar codes, so called the systematic gain in this paper. A composite model is proposed to approximate the dominant error cases in the SC decoding process. This composite model divides the errors into independent regions and coupled regions, controlled by a coupling coefficient. Based on this model, the systematic gain can be conveniently calculated. Numerical simulations are provided in the paper showing very close approximations of the proposed model in quantifying the systematic gain.

cs.IT↗