• 제목/요약/키워드: Noise Classification

검색결과 674건 처리시간 0.031초

안정 상태에서의 정량 뇌파를 이용한 기계학습 기반의 경도인지장애 환자의 감별 진단 모델 개발 및 검증 (Development and Validation of a Machine Learning-based Differential Diagnosis Model for Patients with Mild Cognitive Impairment using Resting-State Quantitative EEG)

  • 문기욱;임승의;김진욱;하상원;이기원
    • 대한의용생체공학회:의공학회지
    • /
    • 제43권4호
    • /
    • pp.185-192
    • /
    • 2022
  • Early detection of mild cognitive impairment can help prevent the progression of dementia. The purpose of this study was to design and validate a machine learning model that automatically differential diagnosed patients with mild cognitive impairment and identified cognitive decline characteristics compared to a control group with normal cognition using resting-state quantitative electroencephalogram (qEEG) with eyes closed. In the first step, a rectified signal was obtained through a preprocessing process that receives a quantitative EEG signal as an input and removes noise through a filter and independent component analysis (ICA). Frequency analysis and non-linear features were extracted from the rectified signal, and the 3067 extracted features were used as input of a linear support vector machine (SVM), a representative algorithm among machine learning algorithms, and classified into mild cognitive impairment patients and normal cognitive adults. As a result of classification analysis of 58 normal cognitive group and 80 patients in mild cognitive impairment, the accuracy of SVM was 86.2%. In patients with mild cognitive impairment, alpha band power was decreased in the frontal lobe, and high beta band power was increased in the frontal lobe compared to the normal cognitive group. Also, the gamma band power of the occipital-parietal lobe was decreased in mild cognitive impairment. These results represented that quantitative EEG can be used as a meaningful biomarker to discriminate cognitive decline.

채널 임펄스 응답 패턴을 이용한 음원 깊이 추정 모호성 제거 기법 (Method for eliminating source depth ambiguity using channel impulse response patterns)

  • 조성일
    • 한국음향학회지
    • /
    • 제41권2호
    • /
    • pp.210-217
    • /
    • 2022
  • 수동 소나 시스템에서 음원 깊이 추정 분야는표적 식별, 추적 등 다양한 전술에 활용할 수 있기 때문에 많은 연구가 지난 수십년간 진행되어 왔다. 본 논문은 기존 논문[조성일 외, 한국음향학회지 제38권 제1호, 120-127(2019)]의 문제점이었던 음원 깊이가두 곳에서 추정되는 모호성을 해결하는데 목적이 있다. 해수면과 해저면에 반사되어 나타나는채널 임펄스 응답의 위상천이 패턴을 이용하여 모호성을 제거하며, 제거 후 하나의 깊이에서 채널 임펄스 응답의 교차점을 통해 음원의깊이를 추정한다. 음원에 대한 정보가 없고, 연속적인 신호 혹은 소음에서 채널 임펄스 응답을 추정하기 위해 음선 기반 블라인드 디컨벌루션 기법이 사용되며, 제안된 알고리즘은 시뮬레이션를 통하여 검증하였다.

Spam Image Detection Model based on Deep Learning for Improving Spam Filter

  • Seong-Guk Nam;Dong-Gun Lee;Yeong-Seok Seo
    • Journal of Information Processing Systems
    • /
    • 제19권3호
    • /
    • pp.289-301
    • /
    • 2023
  • Due to the development and dissemination of modern technology, anyone can easily communicate using services such as social network service (SNS) through a personal computer (PC) or smartphone. The development of these technologies has caused many beneficial effects. At the same time, bad effects also occurred, one of which was the spam problem. Spam refers to unwanted or rejected information received by unspecified users. The continuous exposure of such information to service users creates inconvenience in the user's use of the service, and if filtering is not performed correctly, the quality of service deteriorates. Recently, spammers are creating more malicious spam by distorting the image of spam text so that optical character recognition (OCR)-based spam filters cannot easily detect it. Fortunately, the level of transformation of image spam circulated on social media is not serious yet. However, in the mail system, spammers (the person who sends spam) showed various modifications to the spam image for neutralizing OCR, and therefore, the same situation can happen with spam images on social media. Spammers have been shown to interfere with OCR reading through geometric transformations such as image distortion, noise addition, and blurring. Various techniques have been studied to filter image spam, but at the same time, methods of interfering with image spam identification using obfuscated images are also continuously developing. In this paper, we propose a deep learning-based spam image detection model to improve the existing OCR-based spam image detection performance and compensate for vulnerabilities. The proposed model extracts text features and image features from the image using four sub-models. First, the OCR-based text model extracts the text-related features, whether the image contains spam words, and the word embedding vector from the input image. Then, the convolution neural network-based image model extracts image obfuscation and image feature vectors from the input image. The extracted feature is determined whether it is a spam image by the final spam image classifier. As a result of evaluating the F1-score of the proposed model, the performance was about 14 points higher than the OCR-based spam image detection performance.

오디오 정보를 이용한 골프 동영상 자동 색인 알고리즘 (Automatic Indexing Algorithm of Golf Video Using Audio Information)

  • 김형국
    • 한국음향학회지
    • /
    • 제28권5호
    • /
    • pp.441-446
    • /
    • 2009
  • 본 논문에서는 오디오 정보 분석을 이용하여 골프 통영상을 자동 색인하는 알고리즘을 제안한다. 제안하는 알고리즘에서는 입력되는 골프 동영상을 비디오 신호와 오디오 신호로 분리한 후에, 연속적인 오디오 스트림을 Adaboost Cascade 분류방식을 통하여 스튜디오 환경에서의 아나운서의 음성구간, 선수이름이 TV 화면에 소개 될 때 수반되는 음악구간, 선수들의 플레이에 따라 반응하는 관중들의 박수 및 환호성 소리구간, 필드에서의 레포터의 음성구간, 바다나 바람 등의 필드환경 잡음 사운드구간 등의 5가지 구간으로 분류한다. 그리고 드라이브 샷, 아이런 샷과 퍼팅 샷 시에 발생하는 스윙 사운드는 onset 검출과 변조스펙트럼 검증 방법을 통해 검출되며, 관객의 박수 소리 구간과 결합하여 액션 및 하이라이트를 효율적으로 색인할 수 있게 한다. 제안된 알고리즘은 오디오 신호의 간단한 연산을 통해 의미를 지니고 있는 기본구조들을 검출하기 때문에 골프 동영상에서 사용자가 원하는 부분을 빠르게 브라우징하는 임베이디드 시스템에 적용가능하다.

우도비 특징 벡터를 이용한 SVM 기반의 음성 검출기 (Voice Activity Detection Based on SVM Classifier Using Likelihood Ratio Feature Vector)

  • 조규행;강상기;장준혁
    • 한국음향학회지
    • /
    • 제26권8호
    • /
    • pp.397-402
    • /
    • 2007
  • 본 논문에서는 기존의 통계적 모델 기반의 음성 검출기의 성능 향상을 위해 이진 분류에 우수한 support vector machine(SVM)을 도입한다. 기존의 통계적 모델 기반 음성 검출기의 경우 음성의 존재와 부재에 대한 가설로부터 각각의 통계적 모델을 세워 입력 데이타에 의해 결정된 각 주파수 채널별 우도비(likelihood ratio)를 단순히 기하 평균을 취하여 문턱값과 비교, 음성 검출 여부를 판단한다. 제안된 음성 검출기는 기존의 기하 평균을 이용한 결정식을 대신하여 분류 오류 확률이 최소화되도록 각 주파수 채널별 우도비를 SVM의 특징 벡터로 적용한다. 제안된 SVM 기반의 통계적 모델 음성 검출기는 기존의 LRT를 이용한 음성 검출기 및 SVM 기반의 음성 검출기들과 비교하여 다양한 잡음 환경에서 우수한 성능을 나타낸다.

耳鳴에 관한 임상적 연구 (A Clinical Study of Tinnitus)

  • 최인화
    • 한방안이비인후피부과학회지
    • /
    • 제14권2호
    • /
    • pp.134-145
    • /
    • 2001
  • Introduction: Noises in the ear, whether real or imagined, are called tinnitus. Subjective causes of tinnitus(which is heard only by the patient) are extremely common and the majority of them are treated conservatively. For certain individuals their tinnitus is a major handicap; for others a trivial concern. The most common from of subjective tinnitus is a rushing, hissing or buzzing noise; it is frequently associated with sensorineural heanng loss. The patient may be unaware of the hearing loss, especially if it is a high frequency deficit of moderate severity. The character of the tinnitus may give a clue to the etiology. But the patient often has difficulty in explaining his/her tinnitus in absolute terms, as they have no other tinnitus with which to compare it but their own Tinnitus, like pain, is a subjective state and trying to objectively assess the severity is problematic. Audiological techniques to match subjective loudness to machine-produced noise may offer some help, in that sound intensity matches can bear little correspondence to subjective complaint. In spite of many studies, most patients presently seen complaining of tinnitus are told by their doctors that there is no treatment and that they will have to learn to live with this symptom. Objectives: To perform a clinical analysis of tinnitus and estimate the efficacy of Oriental Medical treatment according to the Byeonjeung(辨證). Subject: We studied 34 patients with complaints of tinnitus who had visited Pundang Cha Oriental Medicine Hospital Department of Otorhinolaryngology from March 1998 to February 2000. All of them had been treated 2 or 3 times a week with acupuncture treatment and had taken herbs according to the Byeonjeung(辨證) method. It was therefore possible for me to know whether their symptoms improved or not. Parameters Observed and Method: We treated them with acupuncture & herb-medication. Sometimes we gave them moxibustion or negative therapy with bloodletting at the acupuncture points(耳門, 聽宮, 聽會). Parameters Observed 1) Distribution of age & sex 2) Chief complaints 3) The sites of tinnitus 4) The quality of tinnitu 5) The duration of disease 6) The problem induced tinnitus 7) Factors increasing disease severity 8) The classification of the Byeonjeung(辨證) 9) The efficacy of treatments Results: 1. Age and sex distribution: The most common occurrence was found in males in their twenties: 6 males($17.7\%$), and in females in their thirties and over sixty: 8 females($23.5\%$). Total patient numbers for men and women were 20 men($58.8\%$), 14 women ($41.2\%$). 2. The most frequent major complaints were hearing disturbances related to tinnitus; and dizziness with tinnitus; each comprising 10 cases($29.4\%$). There were also 7 patients($20.6\%$) with only tinnitus. 3. Tinnitus sites: 13($38.2\%$) said that they felt tinnitus in both ears, equally. In the right ear, 9($26.5\%$), in the left, 6($17.7\%$). 4. The most frequent descriptive symptoms of tinnitus were: humming, hissing, buzzing etc. 5. The duration of disease. 14cases($41.2\%$) had a duration of less than 1 year. 6. 15cases($44.1\%$) complained that it was hard to watch TV or make a phone call because of tinnitus. 10 cases($29.4\%$) complained about depression. 7. Factors increasing severity of tinnitus: ⅰ) fatigue: 18cases($52.9\%$) ⅱ) stress/ tension: 10 cases($29.4\%$) ⅲ) alcohol and tobacco: 5cases($l4.7\%$) 8. Classification through Byeonjeung : ⅰ) 19 cases($55.9\%$) were classified as showing Deficiency syndrome. ⅱ) 15 cases($44.l\%$) were classified as showing Excess syndrome. The deficiency of Qi was 7($20.6\%$), deficiency of Xue, 8($23.5\%$) and insufficiency of the Kidney Yin & Yang, 4($11.8\%$). The flare of Liver fire was 8($23.5\%$) and phlegm-fire, 7($20.6\%$), 9. The efficacy of treatments showed: an improvement in 17cases($50.0\%$); no real improvement or changes in 13 cases($38.2\%$); and some worsening in 4 cases($11.8\%$). In the group with deficiency in Qi, 4($57.1\%$) improved, 1($14.3\%$) showed no change and 2($28.6\%$) were aggravated. In the cases of deficiency in Xue, 6($75.0\%$) improved, 2($25.0\%$) showed no change. In the cases of insufficiency of Kidney Yin & Yang, 3($75.0\%$) showed no change and 1($25.0\%$) were aggravated. In the group of flare of Liver fire, 4($50.0\%$) improved, 3($37.5\%$) no change and 1($12.5\%$) were aggravated. In the cases of phlegm-fire, 3($42.9\%$) improved, 4($57.1\%$) showed no change. Conclusion: We would recommend that any further studies of tinnitus utilize trial treatments of longer than 2 months duration, as any positive effects observed in our study showed that improvement occurred fairly slowly. And we suggest that this study could be utilized as a reference for clinical Oriental Medical treatment of tinnitus. If we try to apply music or sound therapy treatment properly combined with ours, we expect it to provide psycological stability in addition to inducing masking effects, even though it may not directly decrease or completely remove tinnitus.

  • PDF

소음 특수건강진단 자료를 이용한 순음청력검사 평가 (Evaluation of Puretone Threshold Using Periodic Health Examination Data on Noise-exposed Workers in Korea)

  • 김양호;최정근;박정선;문영한;김규상
    • Journal of Preventive Medicine and Public Health
    • /
    • 제32권1호
    • /
    • pp.30-39
    • /
    • 1999
  • 이 연구는 특수건강진단기관의 소음성 난청 진단결과의 유소견자$(D_1)$와 요관찰자(C)를 하나의 평가 지표로 설정하여, 첫째 소음 특수건강진단 결과 소음성 난청의 실태 파악, 둘째 소음성 난청 요관찰 자의 청력장애 평가, 셋째 정력장애 정도 에 따른 각 주파수 영역별 기도순음청력 검사 결과를 통해 청력손실의 정도를 파악하고 이의 판정기준에 따른 진단의 적정성을 검토하고자 하였다. 1. 1994년 l월부터 12월까지의 73개 특수건강진단기관의 특수건강진단 실시 사업장은 27,347개이며 이중 소음 특수 건강진단 설시 사업장은 16,388개(59.9%)이었으며, 전체 특수건강진단 수진 근로자는 731,029명이며 이중 소음 특수 건강진단 수진 근로자는 343,457명 (47.0%)이었다. 소음성 난청 요관찰자는 38,058명, 소음성 난청 유소견자는 1,358 명으로 소음성 난청 요관찰률은 11.1%, 유소견율은 0.44%이었다. 지역에 따라 소음성 난청 요관찰률의 차이를 보여주며 판정기준의 적용에 따른 기도순음 청력평균손실치가 일부 적정하게 판단되지 못하였음을 보여 주었다. 2. ISO 기준의 3분법에 의한 청력 평가시 97%가 경도난청 이하였으며, 회화음 역에서의 4분법에 비해 거의 비슷하였으나 약간 정상역이 많았고, 고음역을 포함하여 평가하는 4분법과 6분법의 적용시 정상자의 경도난청으로의 가능성이 높다고 볼 수 있어 청력평가시 평가방법의 적용에 따라 내재적인 판별능의 차이를 보여준다고 볼 수 있다. 3. 우측귀의 청력역치를 ISO 기준에 의해 평가한 후 양귀의 청력역치의 분포 및 차이를 보면, 우측귀의 평균역치(표준편차)가 20.54(9.56) dB, 좌측귀의 평균역치가 20.54(9.57) dB로 좌측귀의 평균역치가 우측보다 높았다. 양귀의 청력이 75.4%에서 정상역이었으며, 21,562명 (90.6%)의 양귀 청력역치 차이의 범위가 10dB이내였다. 4. 소음성 난청 요관찰자의 회화음역에 속하는 500, 1,000 및 2,000 Hz에서의 기도청력역치를 산술평균으로 하여 구하는 3분법의 청력손실도(표준편차)를 주파수 별로 보면, 우측귀에서 500 Hz 21.08(10.23), 1,000 Hz 18.44(10.01), 2,000 Hz 22.09(13.46), 4,000 Hz 52.36(16.38) dB이었다. 평균청력손실도를 10 dB 간격으로 구분한 후 각각의 주파수별 청력역치를 살펴보면, 정상역인 20 dB미만에서 고음역인 4,000 Hz에서 회화음역인 500, 1,000 및 2,000 Hz에서 보다 평균 30-40 dB 이상의 역치를 보이는 $C_5-dip$ 현상을 특징적으로 보였다. 평균정력손질이 증가함에 따라 4,000 Hz에서의 역치 증가 현상이 점차적으로 감소하다 평균청력손실이 50 dB 이상에서는 10dB 내외의 차이만을 나타내었다. 이상과 같이 소음성 난청 요관찰자에 대한 분석에서 소음성 난청의 평가방법 에 따른 실태와 의미, 소음에 의한 조기청력손실의 특정과 소음성 난청의 판정기준에 따른 진단의 적정성을 확인할 수 있었으며, 소음성 난청 요관찰자에 대한 관리의 필요성을 제언할 수 있겠다.

  • PDF

Spine Computed Tomography to Magnetic Resonance Image Synthesis Using Generative Adversarial Networks : A Preliminary Study

  • Lee, Jung Hwan;Han, In Ho;Kim, Dong Hwan;Yu, Seunghan;Lee, In Sook;Song, You Seon;Joo, Seongsu;Jin, Cheng-Bin;Kim, Hakil
    • Journal of Korean Neurosurgical Society
    • /
    • 제63권3호
    • /
    • pp.386-396
    • /
    • 2020
  • Objective : To generate synthetic spine magnetic resonance (MR) images from spine computed tomography (CT) using generative adversarial networks (GANs), as well as to determine the similarities between synthesized and real MR images. Methods : GANs were trained to transform spine CT image slices into spine magnetic resonance T2 weighted (MRT2) axial image slices by combining adversarial loss and voxel-wise loss. Experiments were performed using 280 pairs of lumbar spine CT scans and MRT2 images. The MRT2 images were then synthesized from 15 other spine CT scans. To evaluate whether the synthetic MR images were realistic, two radiologists, two spine surgeons, and two residents blindly classified the real and synthetic MRT2 images. Two experienced radiologists then evaluated the similarities between subdivisions of the real and synthetic MRT2 images. Quantitative analysis of the synthetic MRT2 images was performed using the mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results : The mean overall similarity of the synthetic MRT2 images evaluated by radiologists was 80.2%. In the blind classification of the real MRT2 images, the failure rate ranged from 0% to 40%. The MAE value of each image ranged from 13.75 to 34.24 pixels (mean, 21.19 pixels), and the PSNR of each image ranged from 61.96 to 68.16 dB (mean, 64.92 dB). Conclusion : This was the first study to apply GANs to synthesize spine MR images from CT images. Despite the small dataset of 280 pairs, the synthetic MR images were relatively well implemented. Synthesis of medical images using GANs is a new paradigm of artificial intelligence application in medical imaging. We expect that synthesis of MR images from spine CT images using GANs will improve the diagnostic usefulness of CT. To better inform the clinical applications of this technique, further studies are needed involving a large dataset, a variety of pathologies, and other MR sequence of the lumbar spine.

PFCM 클러스터링 기법의 개선 (Improvement of the PFCM(Possibilistic Fuzzy C-Means) Clustering Method)

  • 허경용;최세운;우영운
    • 한국정보통신학회논문지
    • /
    • 제13권1호
    • /
    • pp.177-185
    • /
    • 2009
  • 클러스터링은 주어진 데이터 포인트들을 주어진 개수의 그룹으로 나누는 비지도 학습의 한 방법이다. 클러스터링의 방법 중 하나로 널리 알려진 퍼지 클러스터링은 하나의 포인트가 모든 클러스터에 서로 다른 정도로 소속될 수 있도록 함으로써 하나의 클러스터에만 속할 수 있도록 하는 K-means와 같은 방법에 비해 자연스러운 클러스터 형태의 유추가 가능하고, 잡음에 강한 장점이 있다. 이 논문에서는 기존의 퍼지 클러스터링 방법 중 소속도(membership)와 전형성(typicality)을 동시에 계산해 낼 수 있는 Possibilistic Fuzzy C-Means(PFCM) 방법에 Gath-Geva(CG)의 방법을 적용하여 PFCM을 개선한다. 제안한 방법은 PFCM 장점을 그대로 가지면서도, GG의 거리 척도에 의해 클러스터들 사이의 경계를 강조함으로써 분류 목적에 적합한 소속도를 계산할 수 있으며 전형성은 가우스 형태의 분포에서 생성된 포인트들의 분포 함수를 정확하게 모사함으로써 확률 밀도 추정의 방법으로도 사용될 수 있다. 또한 GG 방법은 Gustafson-Kessel 방법과 달리 클러스터에 포함된 포인트의 개수가 확연히 차이나는 경우에도 정확한 결과를 얻을 수 있다. 이러한 사실들은 실험 결과를 통해 확인할 수 있다.

Importance of Postoperative Stimulated Thyroglobulin Level at the Time of 131I Ablation Therapy for Differentiated Thyroid Cancer

  • Hasbek, Zekiye;Turgut, Bulent;Kilicli, Fatih;Altuntas, Emine Elif;Yucel, Birsen
    • Asian Pacific Journal of Cancer Prevention
    • /
    • 제15권6호
    • /
    • pp.2523-2527
    • /
    • 2014
  • Background: Serum thyroglobulin detection plays an essential role during the follow-up of thyroid cancer patients treated with total/near total thyroidectomy and radioiodine ablation. The aim of this retrospective study was to evaluate the relationship between stimulated serum thyroglobulin (Tg) level at the time of high dose $^{131}I$ ablation and risk of recurrence, using a three-level classification in patients with differentiated thyroid cancer (DTC) according to the ATA guidelines. Also we investigated the relationship between postoperative stimulated Tg at the time of ablation and DxWBS results at 8-10 months thereafter. Materials and Methods: Patients with radioiodine accumulation were regarded as scan positive (scan+). If there was no relevant pathological radioiodine accumulation or minimal local accumulation in the thyroid bed region, this were regarded as scan negative (scan-) at the time of DxWBS. We classified patients in 3 groups as low, intermediate and high risk group for assessment of risk of recurrence according to the revised ATA guidelines. Also, we divided patients into 3 groups based on the stimulated serum Tg levels at the time of $^{131}I$ ablation therapy. Groups 1-3 consisted of patients who had Tg levels of ${\leq}2ng/ml$, 2-10 ng/ml, and ${\geq}10ng/ml$, respectively. Results: A total of 221 consecutive patients were included. In the high risk group according to the ATA guideline, while 45.5% of demonstrated Scan(+) Tg(+), 27.3% of patients demonstrated Scan(-) Tg(-); in the intermediate group, the figures were 2.3% and 90.0% while in the low risk group, they were 0.6% and 96.4%. In 9 of 11 patients with metastases (81.8%), stimulated serum Tg level at the time of radioiodine ablation therapy was over 10, however in 1 patient (9.1%) it was <2ng/mL and in one patient it was 2-10ng/mL (p=0.005). Aggressive subtypes of DTC were found in 8 of 221 patients and serum Tg levels were ${\leq}2ng/ml$ in 4 of these 8. Conclusions: We conclude that TSH-stimulated serum thyroglobulin level at the time of ablation may not determine risk of recurrence. Therefore, DxWBS should be performed at 8-12 months after ablation therapy.