基于特征点匹配的拼接算法研究与实现

针对图像间因具有旋转及光线强度差异等现象而导致的拼接效果不佳及拼接速度慢的问题,提出一种基于特征点的配准算法.该算法首先利用相位相关法确定图像重叠区域,然后采用改进Harris角点检测算法检测角点,再根据相似测度NCC(Normalized Cross Correlation)方法提取出匹配特征点对,最后用渐进渐出的方法实现拼接图像的融合.实验证明,该算法比传统算法在角点数目上减少了四分之一,有效地去除了拼接产生的鬼影现象.能有效地提高图像拼接的速度和精度.     

一种改进的非刚性图像配准算法

非刚性图像配准一直是计算机视觉领域的研究重点. 为解决上述问题, 提出一种改进的光流场模型算法, 以提高光流估计的准确度. 算法首先对原始变分光流模型进行了改进, 提出利用新的各向异性正则项来代替原来的同向扩散函数, 以避免图像模糊, 保留图像的边缘特征与细节特征; 此外, 通过引入包含邻域信息的非局部平滑项来去除光流噪点, 同时增加了一个结合图像结构与光流运动信息的权函数, 以减少过平滑所造成的细节丢失, 提高算法的鲁棒性. 最后, 利用交替最小化与金字塔分层迭代策略相结合的方法求解位移场, 实现非刚性图像的自动配准. 仿真实验结果表明, 与传统方法相比, 本文算法对不同类型的非刚性图像均具有较高的鲁棒性, 取得了理想的图像配准效果.      

基于边缘信息和先验形状的水平集图像分割方法及其在病原菌识别中的应用

研究从多目标图像中自动提取单个目标的图像处理方法.从分析曲线的水平集表示入手,首先探讨了水平集动态轮廓分割和配准模型构建的统计思想和变分方法,然后针对多目标粘连图像的特点,提出了含边缘信息和先验形状的水平集图像分割模型,并将其应用于病原菌的识别.由于引入边缘信息改进对分割的约束,加强了目标边缘对分割轮廓的吸引,同时消除了一些由噪声、阴影和杂质造成的影响.实验表明,改进后的先验形状水平集图像分割方法能直接从多目标粘连图像中提取单个目标,进一步完善了依据显微镜图像识别病原菌的图像处理方法.     

基于FCM-WA联合算法的多种类矿石图像分割

矿石图像分割是基于机器视觉的矿石粒度分布检测的重要组成部分。针对复合矿山中颜色多样、纹理复杂且边缘粘连的多种类矿石图像难以识别与分割的问题,提出了一种基于FCM-WA联合算法的矿石图像分割方法。首先对矿石图像进行形态学优化,利用双边滤波、直方图均衡化和形态学重构来优化矿石图像的几何特征,减少噪声对分割效果的影响,提高图像对比度;然后将模糊C均值聚类（FCM）算法与分水岭（WA）算法相结合,利用FCM算法进行聚类迭代,计算出合适的分割阈值并对矿石图像进行分割,输出二值化图像;再利用基于距离变换的WA算法优化FCM算法的分割结果,对FCM算法输出的矿石图像边缘粘连部分进行分割,以获取最佳的分割图像。研究结果表明：（1）利用形态学优化流程处理矿石图像能够减少噪声并增强边缘信息,从而提高对比度;（2）相比传统的大津法和遗传算法,本文所提FCM-WA方法的稳健性更强、分割效果更好,对多种类的矿石图像像素分割准确率和矿石粒度识别准确率均可达到92%以上;（3）通过试验验证,FCM-WA方法能够精确地分割颜色多样、纹理特征复杂及边缘粘连的多种类矿石图像,分割结果满足粒度分布检测的要求;（4）FCM-...     

专利名称：基于视觉感知的交互式乳腺钼靶图像检索方法

本发明涉及基于视觉感知的交互式乳腺钼靶图像检索方法。本发明方法首先读人一幅待处理的乳腺钼靶图像,利用图像分割算法标志疑似病灶,将疑似病灶区域的特征抽取出来（包括面积、偏心率、紧凑度、不变矩、Gabor特征、分维数）,在病灶数据库中自动寻找一组与疑似病灶区域的特征（包括面积、偏心率、     

基于互信息的复合遗传算法在医学图像配准中的应用

分析当前的医学图像配准技术及算法,选择基于体素相似性的图像配准方法,以互信息作为配准相似性测度,将混合遗传算法、自适应遗传算法和育种算法相结合,提出复合遗传算法,以提高遗传算法的鲁棒性,加速收敛过程,准确地实现多模医学图像配准,再利用此算法对无噪声图像和有噪声图像进行配准实验.实验结果证明了该算法的可行性,是一种值得推广的医学图像配准算法.     

浮选泡沫图像特征提取方法研究进展

机器视觉作为设备操作人员的工具,在泡沫浮选设备的监测中得到了广泛的应用。利用泡沫图像数据集建立预测识别模型,以初级泡沫特征参数为输入,以品位和回收率等浮选指标为输出。根据是否需要手动提取浮选泡沫图像特征,可以将特征提取算法划分为两大类别：一种是基于颜色、形态特征等的传统手动特征提取方法,另一种是基于深度神经网络的自动特征提取方法。本文总结并归纳了近年来浮选泡沫图像特征提取算法领域的研究进展,分析了各种方法的优势和不足,对当前难以人工识别泡沫状态及实现浮选自动化提升浮选效率,具有一定的指导价值。     

基于神经网络实现废钢车辆的精准定位

为实现废钢等级识别全过程无人干预,首先需要通过摄像机自动对废钢车辆进行定位。由于废钢卸料点周围都是废钢、废钢车辆里装载的也是废钢。按照传统方法,很难精准的定位废钢车辆。本文首先对传统的基于阈值的图像分割方法、基于边缘的图像分割方法、基于区域的图像分割方法进行分析,然后提出了基于神经网络的废钢车辆定位方法,与传统图像分割方法相比,准确率有明显的提升,能够精准的实现废钢车辆的定位。     

一种高度自动化的三维CT数据腰椎分割方法

针对现有分割方法,分别存在无法完全分割、不能适应所有数据或过多的手动干预等局限性问题,提出了一种用于CT图像单节腰椎分割的新方法.首先从CT图像中分割整条脊椎,然后再分别断开单个椎体和椎小关节,从而使待分割的一节腰椎与其他部位分离.利用人造数据和真实数据对本方法进行了评估,结果显示本法可以高度自动化地分割任意单节腰椎.本方法是医生进行辅助诊断的重要方法,在相关领域的研究中具有广泛意义.     

基于椭圆对称方向矩的可见光与红外图像配准算法

针对图像匹配制导中异源图像匹配难度大的问题,提出一种基于椭圆对称方向矩的可见光与红外图像配准算法.基于最稳定极值区域提取异源图像中具有尺度和仿射不变特性的椭圆区域,利用聚类分割方法从中自动选取具有异源不变性的同质区域特征,用椭圆对称方向矩描述区域特征边界各方向上的相似程度,通过互相关性指标进行特征匹配,获取匹配特征对,利用匹配矫正策略减少误匹配.实验结果表明:较传统算法,进一步提高了可见光与红外图像关联特征的匹配效率,正确率超过了95%,计算时间缩短了近一半.基本满足图像匹配制导对匹配算法实时性好、匹配正确率高、抗干扰能力强等要求.      

Pubic arch detection in transrectal ultrasound guided prostate cancer therapy

New biopsy techniques, increased life expectancy, and prostate-specific antigen (PSA) screening have contributed to an increase in the reported incidence of prostate cancer. Among several treatment options available to the patients, transperineal prostate brachytherapy has emerged as a medically successful, cost-effective outpatient procedure for treating localized prostate cancer. Transperineal prostate brachytherapy employs transrectal ultrasound (TRUS) as the primary imaging modality to accurately preplan and subsequently execute the placement of radioactive seeds into the prostate. Under TRUS guidance, a needle (preloaded with radioactive seeds) is inserted through a template guide, through the perineum and into a predetermined prostate target. The pubic arch, formed by the central union of pelvic bones, is a potential barrier to the passage of these needles in the prostate. A critical aspect, therefore, in the planning and execution of the brachytherapy procedure is the accurate assessment of pubic arch interference (PAI) in relation to the prostate. Traditionally, the evaluation of PAI has involved computed tomography correlate scanning or crude subjective evaluations. In this paper, we describe a new method of assessing PAI by detecting the pubic arch via image processing on the TRUS images. The PAI detection (PAID) algorithm first uses a technique known as sticks to selectively enhance the contrast of linear features in ultrasound images. Next, the enhanced image is thresholded via percentile thresholding. Finally, we fit a parabola (a model for the pubic arch) recursively to the thresholded image. Our evaluation result from 15 cases indicates that the algorithm can successfully detect the pubic arch with 90% accuracy. Based on this study, we believe that detecting the pubic arch and assessing PAI can be done practically and more accurately in the clinical setting using TRUS rather than the current available methods.     

Nitric oxide superoxide and peroxynitrite modulate osteoclast activity

PURPOSE: To assess whether magnetic resonance (MR) spectroscopic imaging with MR imaging can improve prostate cancer localization in postbiopsy hemorrhage cases. MATERIALS AND METHODS: Records of 175 patients with prostate cancer were retrospectively reviewed; 42 patients (135 hemorrhagic sites) had spatially correlated biopsy data. Patients underwent both phased-array coil-endorectal coil MR imaging and three-dimensional MR spectroscopic imaging within 180 days after transrectal ultrasound (US)-guided biopsy. High-signal-intensity hemorrhage on T1-weighted images and corresponding high- or low-signal-intensity areas on T2-weighted images and the metabolic ratio (choline + creatine)/citrate were recorded. Cancer was identified as a low-signal-intensity area at T2-weighted MR imaging or a metabolite ratio greater than 3 standard deviations above normal at MR spectroscopic imaging. MR imaging, spectroscopic, and biopsy findings were compared. RESULTS: Forty-nine patients had postbiopsy hemorrhage. On T2-weighted images, a higher (P < .01) percentage of hemorrhagic sites demonstrated low signal intensity (80% [108 of 135 sites]), which is similar to the signal intensity seen with cancer. The addition of MR spectroscopic imaging to MR imaging resulted in a significant increase (P < .01) in the accuracy (52% to 75%) and specificity (26% to 66%) of tumor detection. CONCLUSION: The addition of MR spectroscopic imaging to MR imaging significantly improves the ability to determine the presence of prostate cancer and spatial extent when postbiopsy changes hinder interpretation with MR imaging alone.     

Comparison of ultrasound imaging in patients undergoing transperineal and transrectal prostate ultrasound

OBJECTIVES: Prostatic evaluation in men who have undergone prior abdominoperineal resection pose an unusual challenge for the urologist. Neither digital rectal examination nor transrectal ultrasound (TRUS) can be performed. Transperineal ultrasound (TPUS) has been suggested as an alternative means of imaging. This imaging modality was compared directly with the standard TRUS method. METHODS: TPUS was performed with a 4-MHz abdominal probe or biplane multiple frequency probe at a frequency of 5 to 7 MHz followed by TRUS at 7 MHz in 50 consecutive men referred for prostate ultrasound and biopsy who had not undergone prior abdominoperineal resection. Dimensions of the prostate and ultrasound findings such as hypoechoic, anechoic, or hyperechoic areas were noted for each sonographic approach. Volume calculation was performed by the prolate spheroid method. RESULTS: There was good TPUS visualization of the prostate in the transverse plane in 48 (96%) of 50 patients and in the sagittal plane in 45 (90%) of 50 patients. Prostate volume calculation by TPUS correlated well with the volume calculated by TRUS (r=0.876). Twenty-nine patients (58%) were found to have suspicious hypoechoic lesions by TRUS; none were seen by TPUS. Prostatic calcifications were present in 12 patients and were visualized by both TPUS and TRUS in all 12 patients. Six prostate glands demonstrated cystic lesions on TRUS imaging; three of these cystic lesions were also seen with TPUS imaging. CONCLUSIONS: TPUS allows visualization of the prostate with volume determination that is comparable to the volume determination by TRUS. Some intraprostatic findings such as calcifications and cysts can be identified; however, suspicious hypoechoic lesions were not identified by TPUS imaging of the prostate.     

Immuno-hemolytic transfusion reactions. II Physiopathologic basis and diagnosis

Prostatic intraepthelial neoplasia (PIN) is a putative premalignant change that bears a morphological similarity to prostatic cancer and shows increased frequency, severity, and extent in patients with prostate cancer. This article discusses the evidence for PIN as a premalignant lesion, reviews the morphology, terminology, appropriate grading system, and diagnostic significance of PIN, as well as describes management recommendations for further evaluation when PIN is diagnosed in prostate resection and biopsy specimens. Clinical management of high-grade PIN found in transurethral resection of the prostate (TURP) or prostate biopsy specimens should include repeat transrectal ultrasound (TRUS) and prostate biopsy for early detection of prevalent coexistent prostate cancer. In cases of high-grade PIN, increased surveillance methods have the potential to decrease morbidity and mortality by early cancer diagnosis.     

On the future of the speciality of Obstetrics and Gynecology. A Dutch view

Between 4 and 15 ng/ml serum prostate-specific antigen (PSA) has a low specificity for prostate cancer (PCa). One accepted method to enhance this specificity is transrectal ultrasonography (TRUS)-measured PSA-density (PSA-D). We compared this method with a new alternative, transition zone PSA (PSA-TZ). We measured total and transition zone prostatic volumes by TRUS and calculated PSA-D and PSA-TZ in 59 patients with suspicion of PCa and PSA between 4 and 15 ng/ml. All patients then had sextant biopsies of the prostate, 30 were positive for PCa and 29 showed benign tissue. With a cut-off value of 0.35, PSA-TZ had a positive predicted value of 77% for PCa, whereas PSA-D, with a cut-off value of 0.12, had a positive predicted value of 55%. Our data suggest PSA-TZ to be more reliable for avoiding unnecessary biopsies in patients with PCa suspicion and serum PSA below 15 ng/ml. PSA-TZ, calculated by TRUS, enhances the specificity of PSA for needle biopsy diagnosis of PCa.     

Registration of MR and SPECT without using external fiducial markers

The aim of our work is to present, test and validate an automated registration method used for matching brain SPECT scans with corresponding MR scans. The method was applied on a data set consisting of ten brain IDEX SPECT scans and ten T1- and T2-weighted MR scans of the same subjects. Of two subjects a CT scan was also made. (Semi-) automated algorithms were used to extract the brain from the MR, CT and SPECT images. Next, a surface registration technique called chamfer matching was used to match the segmented brains. A perturbation study was performed to determine the sensitivity of the matching results to the choice of the starting values. Furthermore, the SPECT segmentation threshold was varied to study its effect on the resulting parameters and a comparison between the use of MR T1- and T2-weighted images was made. Finally, the two sets of CT scans were used to estimate the accuracy by matching MR to CT and comparing the MR-SPECT match to the SPECT-CT match. The perturbation study showed that for initial perturbations up to 6 cm the algorithm fails in less than 4% of the cases. A variation of the SPECT segmentation threshold over a realistic range (25%) caused an average variation in the optimal match of 0.28 cm vector length. When T2 is used instead of T1 the stability of the algorithm is comparable but the results are less realistic due the large deformations. Finally, a comparison of the direct SPECT-MR match and the indirect match with CT as intermediate yields a discrepancy of 0.4 cm vector length. We conclude that the accuracy of our automatic matching algorithm for SPECT and MR, in which no external markers were used, is comparable to the accuracies reported in the literature for non-automatic methods or methods based on external markers. The proposed method is efficient and insensitive to small variations in SPECT segmentation.     

Reassessment of nonplanimetric transrectal ultrasound prostate volume estimates

OBJECTIVES: To assess the accuracy and reproducibility of nonplanimetric transrectal ultrasound (TRUS) volume estimates because inaccurate volume estimates may potentially undermine the value of serum prostate-specific antigen density (PSAD) in early prostate cancer detection. METHODS: We prospectively evaluated 535 consecutive male patients with two consecutive volume determinations performed by the same ultrasonographer at the time of the same visit. RESULTS: Pearson correlation coefficients between two consecutive gland volume estimates ranged from 0.82 to 0.85 depending on the formula used; however, these correlation coefficients corresponded to an average 25% difference between the first and second gland volume estimates. CONCLUSIONS: Although two consecutive nonplanimetric TRUS volume estimates show statistically good correlation, clinically up to a 25% volume difference should be expected between two such volume estimates. In consequence, nonplanimetric TRUS volume estimates should be interpreted with caution, especially when used for PSAD calculation, in the early detection of prostate cancer.     

A long follow-up on prostate specific antigen density and prostate biopsy

OBJECTIVE: To determine prostate specific antigen density (PSAD) in a risk population without evidence of prostatic cancer, and to assess the long-term usefulness of PSAD as a parameter for determining the need for a prostatic biopsy in patients with a normal digital rectal examination (DRE) and transrectal ultrasound (TRUS). METHODS: The records of 582 patients referred to the clinic between February, 1992 and February, 1994 were studied retrospectively. All these patients with lower urinary tract symptoms (LUTS) were evaluated based on the following parameters: digital rectal examination, serum PSA levels, prostate volume measured using transrectal ultrasound and PSAD. Prostatic biopsy was performed on 431 patients who had a serum PSA level greater than 4.0 ng/mL. A total of 299 patients (69.3%) had PSA levels between 4.0 and 10.0 ng/mL and represented the target population. The study had two parts, in the first one cancer was diagnosed just by one biopsy and in part II, the patients with negative biopsy in part I were followed for a two-year period and required 2 or 3 biopsies for diagnosis. Of the total of patients who had a negative prostate biopsy in part I of the study, 269 were followed for a period of two years with repeated prostate biopsies. RESULTS: Overall prostate cancer was detected in 22/299 (13.9%) patients, 6/105 (5.7%) with PSAD up to 0.15 and 16/194 (8.2%) with PSAD over 0.15 (p = 0.569). CONCLUSION: PSAD is a useful indicator in decreasing the number of negative biopsies in patients with benign prostatic hyperplasia. However, in a long-term follow-up the PSAD (cutoff level 0.15) was unable to predict which patients had a positive biopsy. According to our results, 5.6% of patients with prostate cancer will be missed using the PSAD criteria.     

Iatrogenic urinary tract infection with Pseudomonas cepacia after transrectal ultrasound guided needle biopsy of the prostate

In response to an unexplained development of Pseudomonas cepacia cystoprostatitis after transrectal ultrasound guided prostate biopsy, a retrospective review of records and biopsy protocol was performed at our institution. Between June 5, 1990 and January 9, 1991 no documented infections occurred in 272 patients undergoing transrectal ultrasound and prostate biopsy. During the next 6 months, however, 9 of 110 patients (8.2%) presented again with infectious symptomatology after transrectal ultrasound guided needle biopsy of the prostate. Culture of a majority of the specimens (67%) yielded P. cepacia. Two additional asymptomatic patients became colonized with P. cepacia. Environmental investigations revealed the ultrasound transmission gel as the source of the contamination. The proposed mechanism of infection was direct prostate or bladder seeding of contaminated transmission gel used to prepare the ultrasound transducer probe. Infections developed in immunocompetent patients despite adequate antimicrobial prophylaxis most likely secondary to underlying bladder outlet obstruction and significant direct inoculum of organisms. We currently recommend use of individualized sterile packets of transmission gel in addition to appropriate antimicrobial prophylaxis and povidone-iodine cleansing enemas when performing transrectal sonographic guided biopsies of the prostate.     

Prevalence and predictors of a positive repeat transrectal ultrasound guided needle biopsy of the prostate

PURPOSE: We determined the prevalence of and risk factors for carcinoma in patients with 1 previously negative prostate biopsy. MATERIALS AND METHODS: Transrectal ultrasound guided prostate needle biopsies were repeated in 130 men. Risk factors analyzed included age, pathological result of initial biopsy, inter-biopsy interval, prostate specific antigen (PSA), PSA density, PSA velocity, digital rectal examination, abnormal transrectal ultrasound and family history of prostate cancer. RESULTS: A total of 39 patients (30%) had positive biopsies for cancer. Univariate analysis revealed that PSA more than 20 ng./ml. and abnormal transrectal ultrasound were more frequent in men with positive second biopsies. Using multivariate logistic regression analysis only PSA more than 20 ng./ml. was a significant risk factor (adjusted odds ratio 4.48, 95% confidence interval 1.02 to 20.1). We determined the incidence of carcinoma in patients who represent the lowest risk group as defined by PSA less than 10 ng./ml., PSA density less than 0.15 mg./ml./cm.3, PSA velocity less than 0.75, ng./ml. per year, no prostatic intraepithelial neoplasia plus negative transrectal ultrasound, digital rectal examination and family history. Of 21 patients who fit this cohort 5 (23.8%) had carcinoma on repeat biopsy. CONCLUSIONS: A significant false-negative rate for initial transrectal ultrasound guided prostate biopsies exists. Baseline risk in lowest risk patients is sufficiently high such that one cannot define a subset of patients for whom repeat biopsy is unnecessary. We recommend repeat biopsy in all patients who meet the criteria for a transrectal ultrasound guided biopsy and in whom the initial biopsy is negative.