共查询到20条相似文献,搜索用时 31 毫秒
1.
LP Tourne 《Canadian Metallurgical Quarterly》1996,66(6):433-440
A commercially available digitizer has been tested for accuracy. The various areas of the digitizing tablet show degrees of precision that can differ for the x- and y-coordinates. The tablet has a mean absolute error of 0.016 mm for the x-coordinate and 0.09 mm for the y-coordinate. Error tended to increase toward the sides of the tablet. Partitioning the error into systematic and random components revealed that the x-error is due mainly to errors in linearity of the digitizer. When compared with the error involved in locating cephalometric landmarks, one can conclude that this type of digitizer is suited for orthodontic purposes. This article describes a new methodological approach to locating and measuring errors in linearity. 相似文献
2.
The accuracy of disk diffusion susceptibility testing of lomefloxacin was evaluated using 1,555 recent clinical isolates from ten medical centers. The isolates were rapidly growing nonfastidious aerobic species (1,501 isolates) and Haemophilus species (54 isolates), each found as an indicated species in the product package insert. Applying the recently proposed modification of disk diffusion interpretive criteria (susceptible at > or = 20 mm and resistant at < or = 16 mm), absolute categorical agreement for nonfastidious aerobes (1,501 strains) was 95.5% with 0.5% very major and 0.1% major errors (error rates calculated using all tested strains as the denominator). The intermethod discord (MIC vs disk diffusion) was 0.5%. This contrasts to the current NCCLS recommended criteria (susceptible at > or = 22 mm and resistant at < or = 18 mm) where the absolute categorical agreement was significantly less (89.6%) with 0.2% very major, and 0.3% major errors, and the intermethod discord was 7.1%. For Haemophilus species (54 strains), intermethod agreement was complete using either the current NCCLS interpretive criteria or the modified criteria. These multicenter (ten laboratories) data support the acceptance of the proposed modification of disk diffusion interpretive criteria for 10-microgram lomefloxacin disks. 相似文献
3.
To investigate the precision achieved by human measurement on a digitizing board, 100 healthy volunteers (46 women, mean age 36 +/- 12 years) were asked to measure 15 times on artificial pattern composed of 15 points. A high precision digitizing board (programmed to the technical accuracy of +/- 50 microns) was used, and mean and maximum errors in measuring the same distance repeatedly and relocalizing the same point repeatedly were obtained for each volunteer. A median mean and maximum error of 0.2 mm and 1.0 mm were found for repeated distance measurement. When simulating QT dispersion measurement (measuring the same distance 12 times), median value of 20 ms was obtained for ECGs of 25 mm/s paper speed. The study concludes that human precision of operating a digitizing board is rather poor. A recommendation is given to use either a computer screen for manual measurement of ECGs or to provide an operator of the digitizing board with an immediate feedback of the precision and measurement stability achieved so that erroneous measurement can be actively rejected. 相似文献
4.
Real-time ultrasonic measurements of 10th-rib backfat (BF10) and loin muscle area (LMA) were made by a single technician at four mean BW (67.4, 80.3, 93.4, and 104.9 kg) on live hogs to assess the accuracy of predicting carcass measurements before and at slaughter weight. Records were evaluated on 655 purebred barrows and 472 purebred gilts in two tests. Residual correlations-accounting for test, sex, and breed effects, among and between scans and carcass measurements--were moderate to high for BF10 (r=.69 to .82) and LMA (r=.57 to .68), with the largest correlations at 104.9 kg of live weight. Ultrasonic BF10 and LMA were within +/-4 mm and +/-6.45 cm2, respectively, of the corresponding carcass measurement 75.9 and 89.8% of the time. Sex differences for LMA bias were significant (P < .001); ultrasonic LMA was overestimated in barrows by .75 cm2 and underestimated in gilts by .91 cm2. Breed differences were significant (P < .001) for BF10 and LMA bias. Standard errors of prediction (SEP) for BF10 and LMA across the two tests were 3.46 mm and 4.04 cm2, respectively. The SEP for BF10 were 3.60 mm for barrows and 3.19 mm for gilts. The SEP for LMA were 3.77 cm2 for barrows and 4.22 cm2 for gilts. The SEP for BF10 within breeds ranged from 3.25 to 3.72 mm, and for LMA, ranged from 2.98 cm2 to 4.90 cm2. Ultrasound measurements overestimated the carcass measurement by .57 mm for carcasses measuring < 24.1 mm and underestimated by 2.81 mm carcasses with BF10 > 30.3 mm. Ultrasonic LMA overestimated the carcass by 2.35 cm2 in carcasses measuring < 32.5 cm2 and underestimated by 2.29 cm2 in carcasses measuring greater than 37.9 cm2. Results indicate that the magnitude of the carcass measurement affects bias and accuracy of prediction for real-time ultrasonic measurements of BF10 and LMA. The SEP statistic is more consistent in evaluating accuracy of ultrasonic measurement than bias, absolute deviations, and percentage of absolute deviation. 相似文献
5.
OBJECTIVE: To investigate a modified angioplasty balloon catheter, which uses a novel balloon polymer to produce luminal moulds. DESIGN: The catheter was tested in polyurethane phantoms of diameter 1.5 to 4.0 mm. Inflations were to 1.4 atmospheres for 20 seconds at 37 degrees C. The moulds were viewed by reinflating the balloon to 0. 34 atmospheres and quantified using macrophotography and caliper measurement. RESULTS: Evidence of systematic error was found with lumen diameters = 2.0 mm, accuracy being 0.32 to 0.80 mm and precision 0.23 to 0.24 mm. However, between 2.5 and 4.0 mm the accuracy of measuring luminal diameters was 0.01 mm, the precision 0. 06 mm, and the absolute mean error 0.05 mm. The results for percentage diameter stenosis were -1.15%, 0.86%, and 1.21%, and for lesion length they were 0.20, 0.60, and 0.41 mm, respectively. CONCLUSIONS: The prototype catheter shows a high degree of accuracy and precision in phantoms of diameter 2.5 to 4.0 mm-the range within which most interventional work is performed. Further work on this technique as a method of direct three dimensional moulding of the coronary artery lumen is warranted. 相似文献
6.
A simple instrument has been developed for measurement of axial length and anterior chamber depth. It is pocket sized, and the digital output is easy to interpret. The long term accuracy of 0.2 mm is ideal in helping to eliminate errors of lens selection. 相似文献
7.
AC Colchester J Zhao KS Holton-Tainter CJ Henri N Maitland PT Roberts CG Harris RJ Evans 《Canadian Metallurgical Quarterly》1996,1(1):73-90
VISLAN is an integrated neurosurgical planning and guidance system. New segmentation and rendering techniques have been incorporated. A stereo video system is used intra-operatively and fulfils four roles. First, the video display is overlaid with graphical outlines showing the position of the planned craniotomy or the target (enhanced reality displays). Second, a skin surface patch is reconstructed from the stereo video images using patterned light (mean errors of surface point location are < 0.15 mm). Third, a freely mobile, hand-held localizer is tracked in real time (position errors are < 0.5 mm and with improved calibration < 0.2 mm), with its position superimposed on the pre-operative patient representation to assist surgical guidance. Fourth, markers fixed to the skull bone next to the cranial opening are used to detect intra-operative movement and to update registration. Initial results from phantom experiments show an overall system accuracy of better than 0.9 mm for intra-operative localization of features defined in pre-operative images. The prototype system has been tested during six neurosurgical operations with very good results. 相似文献
8.
We measured both the just-noticeable difference in time to collision (TTC) with an approaching object, and the absolute accuracy in estimating TTC in the following cases: only binocular information available; only monocular information available; both binocular and monocular information available as in the everyday situation. Observers could discriminate trial-to-trial variations in TTC on the basis of binocular information alone: the just-noticeable difference in TTC (5.1-9.8%) was the same for a small (0.03 deg) target and for a large (0.7 deg) target. In line with previous reports, when only monocular information was available, the just-noticeable difference in TTC was 5.8-12% for the large target. However, observers could not reliably discriminate trial-to-trial variations in TTC with the small target when only monocular information was available. When both binocular and monocular information was available, the just-noticeable difference in TTC for the large target was not significantly different from when only binocular or only monocular information was available. Observers could make reliable estimates of absolute TTC using binocular information only. Errors ranged from 2.5 to 10% for the large target, and 2.6 to 3.0% for the small target, all being overestimates. Errors for the small target were the same or lower than errors for the large target. Observers could make reliable estimates of TTC with the large target using monocular information only. Errors ranged from 2.0 to 12%, all being underestimates. Since monocular information did not provide a basis for reliable estimates of absolute TTC with the small target we conclude that, in everyday conditions, accurate estimates of TTC with small targets are based on binocular information when the object is small and is no more than a few metres away. Errors in estimating absolute TTC were lower in the case where both binocular and monocular information were available (as in the everyday situation) than when only binocular information or only monocular information was available. Errors ranged from 1.3 to 2.7%. An error of 1.3% approaches the accuracy required to explain the +/- 2.0-2.5 msec accuracy with which top sports players can estimate the instant of impact between bat and ball. 相似文献
9.
AIM: To determine the clinical value of common diagnostic tests for bacterial infection in early life. METHODS: A Medline search (1966-95) was undertaken to identify studies that reported the assessment of a diagnostic "test," predicting the presence or absence of bacterial infection in infants up to 90 days of age. The quality of each selected study was assessed using defined criteria. Data were extracted twice to minimise errors. RESULTS: Six hundred and seventy articles were identified. Two independent investigators agreed that 194 studies met the inclusion criteria (kappa = 0.85), 52 of which met primary quality criteria; 23 studies reported data on (a) haematological indices, (b) C reactive protein evaluation, and (c) surface swab assessment. For haematological indices, the likelihood ratios for individual tests ranged from 20.4 (95% confidence interval 7.3 to 56.8) for a white cell count < 7000/mm3 to 0.12 (0.04 to 0.37) for an immature:total (I:T) white cell ratio < 0.2. For C reactive protein evaluation, the likelihood ratios ranged from 12.56 (0.79 to 199.10) for a value of > 6 mg/l to 0.22 (0.08 to 0.65) for a negative value. For surface swab assessment, the likelihood ratios ranged from 33.6 (2.1 to 519.8) for a positive gastric aspirate culture to 0.08 (0.006 to 1.12) for microscopy of ear swab material that did not show any neutrophils. Likelihood ratios for combinations of these individual tests ranged from 10.17 (3.64 to 28.41) to 0.47 (0.22 to 1.00). CONCLUSIONS: The methodological quality of studies assessing the accuracy of diagnostic tests is generally poor. Even in rigorous studies, the reported accuracy of the tests varies enormously and they are of limited value in the diagnosis of infection in this population. 相似文献
10.
JS Fleming JH Conway ST Holgate EA Moore AH Hashish AG Bailey TB Martonen 《Canadian Metallurgical Quarterly》1998,43(8):2423-2429
Planar images of known, theoretical distributions of radioaerosol in the lung have been simulated using lung models derived from magnetic resonance studies on human subjects. Total lung activity was evaluated from the simulated images together with the absolute penetration index (PI) and a relative value expressed as a fraction of that in a simulated ventilation image. The accuracy and precision of these measurements were calculated by comparison with the true values used in the simulation. Total activity was assessed with systematic errors within 5% and precision within 6.5%. Measured PIs varied only slowly with true PI and inter-model variation masked changes between measurements on the different distributions. The relative PI reduced inter-model variation and provided significant differences between all the distributions. PI was significantly affected by misalignment of the lung region of interest. The conducting airways deposition fraction (CADF) used in the simulation correlated linearly with the fractional activity in a central lung region, allowing CADF to be estimated with a precision of 21%. 相似文献
11.
PURPOSE: To assess the accuracy of a conventional simulation procedure in radiotherapy of age-related macular degeneration. METHODS AND MATERIALS: A computed tomographic (CT) extension attached to the treatment simulator was used to acquire CT images immediately after conventional simulation in 18 patients referred for treatment of age-related macular degeneration. Analysis was performed on 16 one-sided treatment cases for whom images were obtained. Error was estimated by the displacement between the observed treatment isocenter and the intended isocenter based on reconstructed eye geometry. RESULTS: Based on single slice measurements, the mean error amplitude was 2.3 mm (range 0.2-5.6). Based on three-dimensional eye globe reconstruction, the mean error amplitude was 2.8 mm (range 0.8-5.3). An incidental finding previously unreported was the lower image quality at the center of the simulator-CT image acquisition field. CONCLUSIONS: Small but significant errors from conventional simulation were noted. The integrated simulation-CT procedure may help correct the errors to improve the accuracy of simulation setup. The lower image quality at the center of image acquisition field requires adaptation of the simulation-CT procedure. 相似文献
12.
An automated high-performance liquid chromatography (HPLC) method for the determination of formoterol in human plasma with improved sensitivity has been developed and validated. Formoterol and CGP 47086, the internal standard, were extracted from plasma (1 ml) using a cation-exchange solid-phase extraction (SPE) cartridge. The compounds were eluted with pH 6 buffer solution-methanol (70:30, v/v) and the eluate was further diluted with water. An aliquot of the extract solution was injected and analyzed by HPLC. The extraction, dilution, injection and chromatographic analysis were combined and automated using the automate (ASPEC) system. The chromatographic separations were achieved on a 5 microm, Hypersil ODS analytical column (200 mm x 3 mm I.D.), using (pH 6 phosphate buffer, 0.035 M + 20 mg/l EDTA)-MeOH-CH3CN (70:25:5, v/v/v) as the mobile phase at a flow-rate of 0.4 ml/min. The analytes were detected with electrochemical detection at an operating potential of +0.63 V. Intra-day accuracy and precision were assessed from the relative recoveries of calibration/quality control plasma samples in the concentration range of 7.14 to 238 pmol/l of formoterol base. The accuracy over the entire concentration range varied from 81 to 105%, and the precision (C.V.) ranged from 3 to 14%. Inter-day accuracy and precision were assessed in the concentration range of 11.9 to 238 pmol/l of formoterol base in plasma. The accuracy over the entire concentration range varied from 98 to 109%, and precision ranged from 8 to 19%. At the limit of quantitation (LOQ) of 11.9 pmol/l for inter-day measurements, the recovery value was 109% and C.V. was 19%. As shown from intra-day accuracy and precision results, favorable conditions (a newly used column, a newly washed detector cell and moderate residual cell current level) allowed us to reach a LOQ of 7.14 pmol/l of formoterol base (3 pg/ml of formoterol fumarate dihydrate). Improvement of the limit of detection by a factor of about 10 was reached as compared to the previously described methods. The method has been applied for quantifying formoterol in plasma after 120 microg drug inhalation to volunteers. Formoterol was still measurable at 24 h post-dosing in most subjects and a slow elimination of formoterol from plasma beyond 6-8 h after inhalation was demonstrated for the first time thanks to the sensitivity of the method. 相似文献
13.
D Yan E Ziaja D Jaffray J Wong D Brabbins F Vicini A Martinez 《Canadian Metallurgical Quarterly》1998,41(3):715-720
PURPOSE: Adaptive Radiation Therapy (ART) is a feedback treatment process that optimizes a patient's treatment according to the patient specific information measured during the course of treatment. Utilizing an electronic portal imaging device (EPID) and a computer-controlled multileaf collimator (MLC), the ART process is currently being implemented in our clinic to improve the treatment accuracy by compensating for the treatment setup error. A prospective study was conducted to evaluate the feasibility and efficacy of the ART process for clinical use. METHODS AND MATERIALS: The prospective study included 20 patients who underwent conventional radiotherapy on a linear accelerator equipped with an EPID and a MLC. No specific changes were made in the routine clinical procedures except daily portal images were obtained for each treatment field. Two-dimensional setup error for each treatment field was then measured offline using a software tool. The measured setup errors from initial treatment days were used to predict the systematic and random setup errors for each treatment field. An adjustment decision was made if the predicted systematic error was larger than or equal to 2 mm. Furthermore, the treatment field was extended if the predicted random setup error could not be effectively compensated by the predefined treatment setup margin. Instead of the conventional approach of patient repositioning, setup adjustment was implemented by reshaping the MLC field. The entire process from measuring setup error to reshaping the MLC field was performed offline through a computer network. After completion of a patient's treatment, the systematic and random setup errors after adjustment were compared with those predicted prior to the adjustment. The accuracy of the adjustment, and the reliability and stability of the process were analyzed. RESULTS: Treatment fields of 13 patients were modified to correct for systematic errors. The mean systematic error was 4 mm with a range of 2 to 7 mm before adjustment. It was reduced to 0.5 mm with a range of 0.2 to 1.4 mm after adjustment. There was no significant difference in random setup errors before and after adjustment. The ART process was found to be stable, as more than 95% of patient specific setup margins were predictable within 1 mm using the first four to nine fractions of treatment, confirming the feasibility of treatment plan reoptimization with the ART process. CONCLUSIONS: The prospective study demonstrates that the ART process can be effectively implemented in routine clinical practice to improve treatment accuracy. This process is also ready to be further extended to reoptimize the treatment plan by incorporating the predicted patient specific setup variation. 相似文献
14.
MR stereotaxic procedures are being increasingly used, particularly in functional neurosurgery where very high levels of localization accuracy are required. Whilst many studies have investigated intrinsic causes of non-linearity, potential errors due to an extrinsic cause are not generally appreciated. It is not uncommon to find objects such as hair clips, paper clips and pins inside high-field magnets. They can remain undetected for long periods because they can reach positions not open to visual inspection and because they often do not produce observable deterioration in routine image quality. In this study we measured the maximum absolute positional shifts caused by such objects and found that these can be significant (> 1 mm, even up to 200 mm from one such object). Additional measurements were performed using an MR compatible Leksell stereotaxic frame to calculate actual stereotaxic coordinate errors. The encompassing nature of the frame is such that some degree of compensation for such nonlinearities is inherent, and so errors for areas of the brain more proximal to the object are found to be reduced but not eliminated. Stereotaxic coordinate errors will not be reduced in nonencompassing designs and in frameless stereotaxy. The prevalence of such objects in clinical systems and the measures required to detect their presence are discussed. The need for quality control testing before each stereotaxic procedure is highlighted. 相似文献
15.
T Narisawa Y Fukaura M Hasebe S Nomura S Oshima H Sakamoto T Inakuma Y Ishiguro J Takayasu H Nishino 《Canadian Metallurgical Quarterly》1998,89(10):1003-1008
A liquid chromatographic/tandem mass spectrometric (LC/MS/MS) assay was developed for the quantitative determination of olanzapine (LY170053, OLZ) in human plasma and serum. Bond Elut C2 solid-phase extraction cartridges (single cartridge or 96-well format), in conjunction with a positive pressure manifold, were used to extract OLZ and its internal standard, LY170222, from the biological matrix. Chromatographic resolution of OLZ from endogenous plasma interferences and its metabolites was accomplished with a MetaChem monochrom HPLC (4.6 x 150 mm, dp 5 microns). Detection was effected with a Perkin-Elmer SCIEX API III Plus mass spectrometer using positive ion atmospheric pressure chemical ionization and a multiple reaction monitoring protocol. The linear dynamic range was from 250 pg ml-1 to 50 ng ml-1 of human plasma/serum using a 0.5 ml aliquot. The inter-day precision (relative standard deviation) and accuracy (relative error) in plasma ranged from 6.26 to 7.66% and from -3.54 to 7.52%, respectively. The intra-day precision and accuracy in serum ranged from 3.46 to 8.76% and from -8.06 to 12.46%, respectively. This assay is sensitive and selective, and will be used to support both human clinical and toxicological analyses. Furthermore, using the 96-well solid-phase extraction format, sample preparation can be easily automated. 相似文献
16.
Automated devices have regularly replaced manual sphygmomanometry for the determination of blood pressure not only in homes and clinics, but also in emergency and critical care settings. Few studies exist that correctly assess the accuracy of these devices, and even fewer that specifically compare commercially available units that rely on different physiologic events for measurement. Six hundred pressure measurements were obtained from 120 subjects using 1 of 3 randomly selected blood pressure monitors. In addition, central arterial pressure measurements were obtained simultaneously and directly from the ascending aorta of each subject. Overall, these devices tended to overestimate diastolic (+2.5 mm Hg, p < 0.0001) and mean (+3.8 mm Hg, p < 0.0001) pressures, but not systolic (+0.7 mm Hg, p = NS) pressure. Compared with the other 2 devices, device I, relying on oscillometric detection, demonstrated a significantly smaller mean absolute error for diastolic pressure (4.9 +/- 3.0 vs 7.0 +/- 4.8 and 6.2 +/- 5.3 mm Hg, p < 0.0001) and mean pressure (4.0 +/- 3.2 vs 7.8 +/- 5.9 and 8.6 +/- 7.5 mm Hg, p < 0.0001), and a trend toward smaller error with systolic pressure (6.8 +/- 6.5 vs 7.3 +/- 6.8 and 8.0 +/-5.6 mm Hg, p = 0.19). Clinically significant (+/-10 mm Hg) errors were common with each device (24.8% overall), but serious (+/-20 mm Hg) errors were unusual (3.2%) and did not occur at all with device I during diastolic and mean pressure measurement. All of the devices tested could be expected to perform satisfactorily in most clinical settings provided that an average error of 4.0 to 8.6 mm Hg is tolerable. This level of accuracy typically extended throughout the range of pressures anticipated in most noncritical clinical situations. As implemented in the devices tested, noninvasive measurement by oscillometry with stepped deflation is more accurate than automated auscultation. 相似文献
17.
18.
This study was designed to estimate the contribution of repositioning inaccuracies to measured intracranial lesion volumes in multiple sclerosis (MS). Five patients with MS were each scanned 10 times, using spin-echo imaging (2,000/34/90), contiguous 5-mm slices, and different scanning positions. The maximum displacements from baseline were +/-4 degrees (AP rotation) and 3 mm (slice offset). Lesion volume was measured twice for each scan using a semiautomated contour technique. Measured lesion volumes ranged from 4,328 mm3 to 164,831 mm3. The mean intrarater coefficient of variation (CV) calculated for individual patients ranged from 1.1 to 4% (median, 1.7%). Using analysis of variance, the overall variance and CV due to altered scan position were greater than that due to rater error (repositioning CV 4.0%, intrarater CV 3.5%). The worst-case difference between volumes in the same patient ranged from 8.9 to 32% (median, 9.9%). Both rater and repositioning errors were greater for smaller lesion volumes. The maximum potential error due to repositioning inaccuracies is of a similar magnitude to the 5 to 10% expected change in lesion volume over 1 year. This study justifies continued careful attention to accuracy in repositioning for serial MR studies in patients with MS. 相似文献
19.
CD Steward SA Stocker JM Swenson CM O'Hara JR Edwards RP Gaynes JE McGowan FC Tenover 《Canadian Metallurgical Quarterly》1999,37(3):544-547
Fluoroquinolone resistance appears to be increasing in many species of bacteria, particularly in those causing nosocomial infections. However, the accuracy of some antimicrobial susceptibility testing methods for detecting fluoroquinolone resistance remains uncertain. Therefore, we compared the accuracy of the results of agar dilution, disk diffusion, MicroScan Walk Away Neg Combo 15 conventional panels, and Vitek GNS-F7 cards to the accuracy of the results of the broth microdilution reference method for detection of ciprofloxacin and ofloxacin resistance in 195 clinical isolates of the family Enterobacteriaceae collected from six U.S. hospitals for a national surveillance project (Project ICARE [Intensive Care Antimicrobial Resistance Epidemiology]). For ciprofloxacin, very major error rates were 0% (disk diffusion and MicroScan), 0.9% (agar dilution), and 2.7% (Vitek), while major error rates ranged from 0% (agar dilution) to 3.7% (MicroScan and Vitek). Minor error rates ranged from 12.3% (agar dilution) to 20.5% (MicroScan). For ofloxacin, no very major errors were observed, and major errors were noted only with MicroScan (3.7% major error rate). Minor error rates ranged from 8.2% (agar dilution) to 18.5% (Vitek). Minor errors for all methods were substantially reduced when results with MICs within +/-1 dilution of the broth microdilution reference MIC were excluded from analysis. However, the high number of minor errors by all test systems remains a concern. 相似文献
20.
《钛工业进展》2019,36(5):1-7
模型处理是影响3D打印钛合金零件成形精度的关键因素之一。以电子束3D打印几种典型形状TC4钛合金零件为研究对象,通过对成形零件进行精度检测,分析其在模型处理过程中引起误差的来源,提出了几项改进措施:对于圆柱体轴类零件,将模型精度设置提高到0. 02 mm,可有效提高零件的成形精度;对于点阵多孔零件,提高STL数据的精度并优化切层算法,保持切层前后模型原貌是解决该问题的途径;对于针状细圆柱零件,将相同零件按照同一方向摆放,可以减少摆放角度对偏差的随机性影响;对于异形件,应在满足成形精度的前提下合理布局并减少支撑;对于薄壁零件,增加支撑密度能够有效提高零件悬空部分的成形精度。 相似文献