Effects of image processing on nodule detection rates in digitized chest radiographs: ROC study of observer performance

To evaluate the effects of image processing in digitized chest radiographs when high-resolution images are used, an examination was done in which the detection of pulmonary nodules in unprocessed digitized chest radiographs was compared with that in images that had undergone processing with two methods, adaptive filtration and histogram equalization. The processing techniques have been optimized in previous work to selectively enhance the retrocardiac and subdiaphragmatic areas without significant alteration of detail in the lung. Eight observers were shown 150 test radiographs (50 unprocessed, 50 processed with adaptive filtration, 50 processed with histogram equalization) containing 150 nodules. The results indicate a statistically significant (P less than .03) difference, with highest observer performance in the chest radiographs processed with adaptive filtration (median area under ROC curve = 0.78), compared with unprocessed images (median = 0.68) and chest radiographs processed with histogram equalization (median = 0.62). Performance in the lung was not significantly different. Adaptive filtration applied to selectively enhance underexposed areas of film images may improve nodule detection. Histogram equalization provided no improvement in performance.     

Wavelet-based histogram equalization enhancement of gastric sonogram images.

The gray levels of gastric sonogram images are usually concentrated at the zero end of the spectrum, making the image too low in contrast and too dark for the naked eye. Though histogram equalization can enhance the contrast by redistributing the gray levels, it has the drawback that it reduces the information in the processed image. In this paper, a wavelet-based enhancement algorithm post-processor is used to further enhance the image and compensate for the information loss during histogram equalization. Experimental results show that the wavelet-based enhancement algorithm can enhance the contrast and significantly increase the informational entropy of the image. Because the combination of the histogram equalization and wavelet approach can dramatically increase the contrast and maintain information rate in gastric sonograms, it has the potential to improve clinical diagnosis and research.     

基于直方图均衡的DICOM图像增强显示技术初步研究

目的选择合适的医学图像直方图均衡增强算法。方法分别采用全局直方图均衡、局域直方图均衡、自适应领域直方图均衡对医学图像增强,并比较了3种方法的优缺点。结果通过直方图均衡增强能改善医学灰阶图像的层次感,提高图像细节的可识别性和边缘特征。结论对于以局部细节为目的的增强采用局域直方图均衡效果较好,而以整幅图像的边界信息为目的的增强采用自适应领域直方图均衡效果较好。     

Histogram equalization of CT images

Histogram equalization for display of clinical CT images was evaluated. In theory, histogram equalization makes optimal use of an available grey scale to display an image, and its use could circumvent the problem of selecting specific window settings for each image. In several clinical images, the use of a spatially variable histogram equalization technique limited to that portion of the CT image occupied by the patient did appear to increase the visibility of anatomic structures. However, using the technique also increased displayed image noise and artifacts. Although radiologists found this to be objectionable, it did not decrease the detectability of simulated low-contrast liver metastases. Further evaluation of histogram equalization for displaying CT images is being pursued.     

Numerical tissue characterization in MS via standardization of the MR image intensity scale

Image intensity standardization is a recently developed postprocessing method that is capable of correcting the signal intensity variations in MR images. We evaluated signal intensity of healthy and diseased tissues in 10 multiple sclerosis (MS) patients based on standardized dual fast spin-echo MR images using a numerical postprocessing technique. The main idea of this technique is to deform the volume image histogram of each study to match a standard histogram and to utilize the resulting transformation to map the image intensities into standard scale. Upon standardization, the coefficients of variation of signal intensities for each segmented tissue (gray matter, white matter, lesion plaques, and diffuse abnormal white matter) in all patients were significantly smaller (2.3-9.2 times) than in the original images, and the same tissues from different patients looked alike, with similar intensity characteristics. Numerical tissue characterizability of different tissues in MS achieved by standardization offers a fixed tissue-specific meaning for the numerical values and can significantly facilitate image segmentation and analysis.     

Adaptive histogram equalization in digital radiography of destructive skeletal lesions

Adaptive histogram equalization, an image-processing technique that distributes pixel values of an image uniformly throughout the gray scale, was applied to 28 plain radiographs of bone lesions, after they had been digitized. The non-equalized and equalized digital images were compared by two skeletal radiologists with respect to lesion margins, internal matrix, soft-tissue mass, cortical breakthrough, and periosteal reaction. Receiver operating characteristic (ROC) curves were constructed on the basis of the responses. Equalized images were superior to nonequalized images in determination of cortical breakthrough and presence or absence of periosteal reaction. ROC analysis showed no significant difference in determination of margins, matrix, or soft-tissue masses.     

Classification of breast ultrasound images using fractal feature

Fractal analyses have been applied successfully for the image compression, texture analysis, and texture image segmentation. The fractal dimension could be used to quantify the texture information. In this study, the differences of gray value of neighboring pixels are used to estimate the fractal dimension of an ultrasound image of breast lesion by using the fractal Brownian motion. Furthermore, a computer-aided diagnosis (CAD) system based on the fractal analysis is proposed to classify the breast lesions into two classes: benign and malignant. To improve the classification performances, the ultrasound images are preprocessed by using morphology operations and histogram equalization. Finally, the k-means classification method is used to classify benign tumors from malignant ones. The US breast image databases include only histologically confirmed cases: 110 malignant and 140 benign tumors, which were recorded. All the digital images were obtained prior to biopsy using by an ATL HDI 3000 system. The receiver operator characteristic (ROC) area index AZ is 0.9218, which represents the diagnostic performance.     

RPW: A hybrid reverse prediction method for level of detail

In this paper, we propose a way of using multiple domain analysis methodology to speed up the image rendering process. We first apply wavelet transform to the original image, and then compress the wavelets in the frequency domain using histogram equalization techniques. When rendering the image, we uncompress the wavelets and reverse predict the upper level images. This process continues until it reaches a certain criteria. We use two terms—total image size (TIS) and total loading time (TLT) to measure the performance of level of detail (LOD) in a network environment. We compare traditional image-based LOD methods with the new method we are proposing. Experiments show that the proposed method can reduce both TIS and TLT. The image rendering speed on a remote client is approximately 2.5 times faster than the common image compression methods. Applications such as remote diagnostic systems and online museums can use this technique to achieve better real-time animation effects.     

Histogram-directed processing of digital chest images

One of the potential advantages of digital chest imaging is the ability to process these images. However, such processing, when uniformly applied to the entire image, is often unsatisfactory due to the different processing requirements of lung field and mediastinum. Therefore, a method to selectively process these regions based upon the histogram of the original image has been developed. Thirteen conventional chest films were digitized with a laser film scanner. Analysis of individual lung field and mediastinum histograms showed that the chest image histogram is essentially bimodal with significant lung field-mediastinum histogram peak separation; overlap between these peaks is small (9% of the total histogram) and insensitive to minor pathologic change or radiographic technique. Using these histograms, a gray level threshold distinguishing mediastinum from lung field was selected and used to direct the regionally-selective processing of several chest images. This technique may prove especially useful for digital enhancement of the underexposed mediastinum often encountered on conventional chest radiographs.     

Standardized volume-rendering of contrast-enhanced renal magnetic resonance angiography

Purpose: To propose a technique for standardizing volume-rendering technique (VRT) protocols and to compare this with maximum intensity projection (MIP) in regard to image quality and diagnostic confidence in stenosis diagnosis with magnetic resonance angiography (MRA).

Material and Methods: Twenty patients were examined with MRA under suspicion of renal artery stenosis. Using the histogram function in the volume-rendering software, the 95th and 99th percentiles of the 3D data set were identified and used to define the VRT transfer function. Two radiologists assessed the stenosis pathology and image quality from rotational sequences of MIP and VRT images.

Results: Good overall agreement (mean κ = 0.72) was found between MIP and VRT diagnoses. The agreement between MIP and VRT was considerably better than that between observers (mean κ = 0.43). One of the observers judged VRT images as having higher image quality than MIP images.

Conclusion: Presenting renal MRA images with VRT gave results in good agreement with MIP. With VRT protocols defined from the histogram of the image, the lack of an absolute gray scale in MRI need not be a major problem.     

A straightforward algorithm for the quantification of power Doppler signals

RATIONALE AND OBJECTIVES: To document a simple algorithm that quantifies power Doppler signals by computer-based statistical image analysis. METHODS: Personal computer-based software was developed. Color pixels representing power Doppler signals were separated from the background gray pixels in power Doppler images converted in BMP format. To avoid misregistration during this segmentation, a threshold value is required to adequately distinguish between gray and color pixels. Equating the intensity of each color pixel to a Power Value (PV), a histogram of Power Values in power Doppler images was obtained. RESULTS: The software was designed to allow users to adjust two thresholds (Difference Threshold and Intensity Threshold) by observing the segmentation results in real time. The software calculates the following indexes within any region of interest as defined by the operator: mean PV, peak PV, standard deviation of PV, percentage power Doppler area, and integrated PV per unit area. CONCLUSIONS: Power Doppler signals can be objectively quantified using this straightforward algorithm. This simple and practical method can be applied to comparative or longitudinal studies upon vascularity or blood flow.     

Improving image quality around subtle lung nodules by reducing artifacts in similar subtraction imaging

Similar subtraction imaging is useful for the detection of lung nodules; however, some artifacts on similar subtraction images reduce their utility. The authors attempted to improve the image quality of similar subtraction images by reducing artifacts caused by differences in image contrast and sharpness between two images used for similar subtraction imaging. Image contrast was adjusted using the histogram specification technique. The differences in image sharpness were compensated for using a pixel matching technique. The improvement in image quality was evaluated objectively based on the degree of artifacts and the contrast-to-noise ratio (CNR) of the lung nodules. The artifacts in similar subtraction images were reduced in 94% (17/18) of cases, and CNR was improved in 83% (15/18) of cases. The results indicate that the combination of histogram specification and pixel matching techniques is potentially useful in improving image quality in similar subtraction imaging.     

X-ray beam equalization: feasibility and performance of an automated prototype system in a phantom and swine.

PURPOSE: To investigate the feasibility of an automated implementation of a beam equalization technique and to evaluate the experimental performance of the prototype system. MATERIALS AND METHODS: X-ray beam equalization involved the process of low-dose image acquisition, attenuator thickness calculation, mask generation, mask positioning, equalized image acquisition, and mask reshaping. The entire equalization process was performed in approximately 7 seconds. The equalized images were assessed both qualitatively and quantitatively by using a humanoid phantom and a swine animal model. The general image quality was assessed for the ability to visualize arterial branches and other anatomic structures. The level of equalization was quantitatively assessed by segmenting the images into an 8 x 8 matrix of square regions. RESULTS: The ratio of the root-mean-squared variance for the equalized and unequalized images from humanoid phantom and swine animal studies was 0.49 and 0.59, respectively. Furthermore, qualitative assessment of the images showed substantial improvement in image quality and visualization of arterial branches after beam equalization in phantom and animal studies. CONCLUSION: Automated area beam equalization is feasible and improves image quality in previously underpenetrated regions of the image.     

Radiologists' preferences for digital mammographic display. The International Digital Mammography Development Group

PURPOSE: To determine the preferences of radiologists among eight different image processing algorithms applied to digital mammograms obtained for screening and diagnostic imaging tasks. MATERIALS AND METHODS: Twenty-eight images representing histologically proved masses or calcifications were obtained by using three clinically available digital mammographic units. Images were processed and printed on film by using manual intensity windowing, histogram-based intensity windowing, mixture model intensity windowing, peripheral equalization, multiscale image contrast amplification (MUSICA), contrast-limited adaptive histogram equalization, Trex processing, and unsharp masking. Twelve radiologists compared the processed digital images with screen-film mammograms obtained in the same patient for breast cancer screening and breast lesion diagnosis. RESULTS: For the screening task, screen-film mammograms were preferred to all digital presentations, but the acceptability of images processed with Trex and MUSICA algorithms were not significantly different. All printed digital images were preferred to screen-film radiographs in the diagnosis of masses; mammograms processed with unsharp masking were significantly preferred. For the diagnosis of calcifications, no processed digital mammogram was preferred to screen-film mammograms. CONCLUSION: When digital mammograms were preferred to screen-film mammograms, radiologists selected different digital processing algorithms for each of three mammographic reading tasks and for different lesion types. Soft-copy display will eventually allow radiologists to select among these options more easily.     

A comparison of conventional intra-oral radiography and computer imaging techniques for the detection of proximal surface dental caries.

The detection of proximal surface caries by the visual interpretation of bitewing radiographs is known to be relatively inaccurate. The present study was designed to examine whether computer image processing could improve the diagnostic accuracy. A computer-aided, software-driven, TV-based system was used to digitize conventional radiographs and digitally process the images using histogram equalization and grey-scale inversion to enhance the images. The computer-enhanced images were compared with conventional intra-oral radiographs for the detection of proximal surface caries using receiver operating characteristic analysis. The results indicate that the digital image processing techniques used did not improve the diagnostic accuracy of dental radiographs. No significant difference in diagnostic accuracy could be detected between the non-enhanced digital images and conventional film-based images for the detection of proximal surface caries.     

Local histogram correction of MRI spatially dependent image pixel intensity nonuniformity

We describe a computationally straightforward post-hoc statistical method of correcting spatially dependent image pixel intensity nonuniformity based on differences in local tissue intensity distributions. Pixel intensity domains for the various tissues of the composite image are identified and compared to the distributions of local samples. The nonuniformity correction is calculated as the difference of the local sample median from the composite sample median for the tissue class most represented by the sample. The median was chosen to reduce the effect ers on determining the sample statistic and to allow a sample size small enough to accurately estimate the spatial variance of the image intensity nonuniformity. The method was designed for application to two-dimensional images. Simulations were used to estimate optimal conditions of local histogram kernel size and to test the accuracy of the method under known spatially dependent nonuniformities. The method was also applied to correct a phantom image and cerebral MRIs from 15 healthy subjects. Results show that the method accurately models simulated spatially dependent image intensity differences. Further analysis of clinical MR data showed that the variance of pixel intensities within the cerebral MRI slices and the variance of slice volumes within individuals were significantly reduced after nonuniformity correction. Improved brain-cerebrospinal fluid segmentation was also obtained. The method significantly reduced the variance of slice volumes within individuals, whether it was applied to the native images or images edited to remove nonbrain tissues. This statistical method was well behaved under the assumptions and the images tested. The general utility of the method was not determined, but conditions for testing the method under a variety of imaging sequences is discussed. We believe that this algorithm can serve as a method for improving MR image segmentation for clinical and research applications.     

On standardizing the MR image intensity scale.

The lack of a standard image intensity scale in MRI causes many difficulties in image display and analysis. A two-step postprocessing method is proposed for standardizing the intensity scale in such a way that for the same MR protocol and body region, similar intensities will have similar tissue meaning. In the first step, the parameters of the standardizing transformation are "learned" from a set of images. In the second step, for each MR study these parameters are used to map their histogram into the standardized histogram. The method was tested quantitatively on 90 whole-brain studies of multiple sclerosis patients for several protocols and qualitatively for several other protocols and body regions. Measurements using mean squared difference showed that the standardized image intensities have statistically significantly (P < 0.01) more consistent range and meaning than the originals. Fixed gray level windows can be established for the standardized images and used for display without the need of per case adjustment. Preliminary results also indicate that the method facilitates improving the degree of automation of image segmentation. Magn Reson Med 42:1072-1081, 1999.     

Image processing algorithms for digital mammography: a pictorial essay.

Digital mammography systems allow manipulation of fine differences in image contrast by means of image processing algorithms. Different display algorithms have advantages and disadvantages for the specific tasks required in breast imaging-diagnosis and screening. Manual intensity windowing can produce digital mammograms very similar to standard screen-film mammograms but is limited by its operator dependence. Histogram-based intensity windowing improves the conspicuity of the lesion edge, but there is loss of detail outside the dense parts of the image. Mixture-model intensity windowing enhances the visibility of lesion borders against the fatty background, but the mixed parenchymal densities abutting the lesion may be lost. Contrast-limited adaptive histogram equalization can also provide subtle edge information but might degrade performance in the screening setting by enhancing the visibility of nuisance information. Unsharp masking enhances the sharpness of the borders of mass lesions, but this algorithm may make even an indistinct mass appear more circumscribed. Peripheral equalization displays lesion details well and preserves the peripheral information in the surrounding breast, but there may be flattening of image contrast in the nonperipheral portions of the image. Trex processing allows visualization of both lesion detail and breast edge information but reduces image contrast.     

Detection of simulated microcalcifications in fixed mammary tissue: An ROC study of the effect of local versus global histogram equalization

PURPOSE: To investigate whether sliding window adaptive histogram equalization (SWAHE) of digital mammograms improves the detection of simulated calcifications, as compared to images normalized by global histogram equalization (GHE). MATERIAL AND METHODS: Direct digital mammograms were obtained from mammary tissue phantoms superimposed with different frames. Each frame was divided into forty squares by a wire mesh, and contained granular calcifications randomly positioned in about 50% of the squares. Three radiologists read the mammograms on a display monitor. They classified their confidence in the presence of microcalcifications in each square on a scale of 1 to 5. Images processed with GHE were first read and used as a reference. In a later session, the same images processed with SWAHE were read. RESULTS: The results were compared using ROC methodology. When the total areas AZ were compared, the results were completely equivocal. When comparing the high-specificity partial ROC area AZ,0.2 below false-positive fraction (FPF) 0.20, two of the three observers performed best with the images processed with SWAHE. The difference was not statistically significant. CONCLUSION: When the reader's confidence threshold in malignancy is set at a high level, increasing the contrast of mammograms with SWAHE may enhance the visibility of microcalcifications without adversely affecting the false-positive rate. When the reader's confidence threshold is set at a low level, the effect of SWAHE is an increase of false positives. Further investigation is needed to confirm the validity of the conclusions.     

组织均衡技术在DR胸部影像中的应用优势

目的　探讨胸部直接数字化X线摄影 (DR)中 ,采用组织均衡技术的DR图像与标准DR图像的差别。方法　从我院2 0 0 3 -0 5所摄的DR胸部影像中按顺序抽取 10 0例作为分析材料 ,通过运用组织均衡技术对图像进行处理 ,比较组织均衡图像与标准DR图像的差别。结果　运用组织均衡技术的图像 ,同一幅图像上不同体厚部位的细节均可清晰显示 ;标准DR图像需调节不同的窗宽、窗位才能清晰显示不同体厚部位的细节。结论　在直接数字化X线摄影中 ,运用组织均衡技术能明显改善先前受体厚影响而难于观察部分的可视性 ,而又不牺牲其他部分的细节显示