Abstract: A computer-aided diagnosis (CAD) method for detection of clustered microcalcifications in digital mammograms based on an image reconstruction using a substantially optimally weighted wavelet transform. Weights at individual scales of the wavelet transform are optimized based on a supervised learning method. In the learning method, an error function represents a difference between a desired output and a reconstructed image obtained from weighted wavelet coefficients of the wavelet transform for a given mammogram. The error function is then minimized by modifying the weights by means of a conjugate gradient algorithm. Performance of the optimally weighted wavelets was evaluated by means of receiver-operating characteristic (ROC) analysis which indicated that the present invention outperformed both a difference-image technique and partial reconstruction method currently used in CAD methods.

Abstract: A method of computerized analysis of temporally sequential digital images, including (a) determining first shift values between pixels of a first digital image and corresponding pixels of a second digital image; (b) warping the second digital image based on the first shift values to obtain a first warped image in which spatial locations of pixels are varied in relation to the first shift values; (c) determining second shift values between pixels of the first digital image and pixels of the first warped image; and (d) warping the first warped image based on the second shift values to obtain a second warped image in which spatial locations of pixels of the first warped image are varied in relation to the second shift values.

Abstract: A computer-aided method for detecting, classifying, and displaying candidate abnormalities, such as microcalcifications and interstitial lung disease in digitized medical images, such as mammograms and chest radiographs, a computer programmed to implement the method, and a data structure for storing required parameters, wherein in the classifying method candidate abnormalities in a digitized medical image are located, regions are generated around one or more of the located candidate abnormalities, features are extracted from at least one of the located candidate abnormalities within the region and from the region itself, the extracted features are applied to a classification technique, such as an artificial neural network (ANN) to produce a classification result (i.e., probability of malignancy in the form of a number and a bar graph), and the classification result is displayed along with the digitized medical image annotated with the region and the candidate abnormalities within the region.

Abstract: A computerized method for the detection and characterization of disease in an image derived from a chest radiograph, wherein an image in the chest radiograph is processed to determine the ribcage boundary, including lung top edges, right and left ribcage edges, and right and left hemidiaphragm edges. Texture measures including RMS variations of pixel values within regions of interest are converted to relative exposures and corrected for system noise existing in the system used to produce the image. Texture and/or geometric pattern indices are produced. A histogram(s) of the produced index (indices) is produced and values of the histograms) are applied as inputs to a trained artificial neural network, which classifies the image as normal or abnormal.

Abstract: A method and system for displaying medical images and computer output from various CAD schemes on the images. Images are loaded into the display system and displayed in a main viewing area. The system can display various CAD schemes, such as the automated detection of lung nodules, interstitial infiltrates, and heart size, in the case of a chest image, on the image. The method and system also display interval change (temporal subtraction) between images. Individual abnormality from CAD schemes can be viewed by clicking buttons with minified (postage stamp size) images with CAD annotation rendered into them. The images and results are then shown on a high-speed monitor. Interval change between the current and a previous image can also be chosen by clicking buttons containing minified images. Interval change between any other pair of images for a patient can be selected by choosing from minified subtraction images presented in a two-dimensional array format.

Abstract: A method and computerized automated initial image matching technique for enhancing detection of interval changes between temporally subsequent radiographic images via image subtraction. The method includes the steps of digitizing images, normalizing density and contrast in the digital images, correcting for lateral inclination in the digital images, detecting edges of a same feature in each image, converting the images into low resolution matrices, blurring the low resolution images, segmenting portions of the blurred low resolution matrices based on the detected edges, matching the digital images based on a cross-correlation match between the segmented portions, performing non-linear warping to further match Regions of Interest (ROI), and performing image subtraction between the matched digital images. The low resolution matrices are greater than 64.times.64 in size and are produced by averaging.

Abstract: A method and system for the computerized registration of radionuclide images with radiographic images, including generating image data from radiographic and radionuclide images of the thorax. Techniques include contouring the lung regions in each type of chest image, scaling and registration of the contours based on location of lung apices, and superimposition after appropriate shifting of the images. Specific applications are given for the automated registration of radionuclide lungs scans with chest radiographs. The method in the example given yields a system that spatially registers and correlates digitized chest radiographs with V/Q scans in order to correlate V/Q functional information with the greater structural detail of chest radiographs. Final output could be the computer-determined contours from each type of image superimposed on any of the original images, or superimposition of the radionuclide image data, which contains high activity, onto the radiographic chest image.

Abstract: A computerized method and system for the radiographic analysis of bone structure and risk of future fracture with or without the measurement of bone mass. Techniques including texture analysis for use in quantitating the bone structure and risk of future fracture. The texture analysis of the bone structure incorporates directionality information, for example in terms of the angular dependence of the RMS variation and first moment of the power spectrum of a ROI in the bony region of interest. The system also includes using dual energy imaging in order to obtain measures of both bone mass and bone structure with one exam. Specific applications are given for the analysis of regions within the vertebral bodies on conventional spine radiographs. Techniques include novel features that characterize the power spectrum of the bone structure and allow extraction of directionality features with which to characterize the spatial distribution and thickness of the bone trabeculae.

Abstract: A method and system for the automated detection of lesions in computed tomographic images, including generating image data from at least one selected portion of an object, for example, from CT images of the thorax. The image data are then analyzed in order to produce the boundary of the thorax. The image data within the thoracic boundary is then further analyzed to produce boundaries of the lung regions using predetermined criteria. Features within the lung regions are then extracted using multi-gray-level thresholding and correlation between resulting multi-level threshold images and between at least adjacent sections. Classification of the features as abnormal lesions or normal anatomic features is then performed using geometric features yielding a likelihood of being an abnormal lesion along with its location in either the 2-D image section or in the 3-D space of the object.

Abstract: An automated computer-aided diagnosis (CAD) method and system using artificial neural networks (ANNs) for the quantitative analysis of image data. Three separate ANNs were applied for detection of interstitial disease on digitized two-dimensional chest images. The first ANN was trained with horizontal profiles in regions of interest (ROIs) selected from normal and abnormal chest radiographs. The second ANN was trained using vertical output patterns obtained from the 1.sup.st ANN for each ROI. The output value of the 2.sup.nd ANN was used to distinguish between normal and abnormal ROIS with interstitial infiltrates. If the ratio of the number of abnormal ROIs to the total number of all ROIs in a chest image was greater than a certain threshold level, the chest image was considered abnormal. In addition, the third ANN was applied to distinguish between normal and abnormal chest images where the chest image was not clearly normal or abnormal.

Abstract: A method and system for the automated detection and classification of masses in mammograms. These method and system include the performance of iterative, multi-level gray level thresholding, followed by a lesion extraction and feature extraction techniques for classifying true masses from false-positive masses and malignant masses from benign masses. The method and system provide improvements in the detection of masses include multi-gray-level thresholding of the processed images to increase sensitivity and accurate region growing and feature analysis to increase specificity. Novel improvements in the classification of masses include a cumulative edge gradient orientation histogram analysis relative to the radial angle of the pixels in question; i.e., either around the margin of the mass or within or around the mass in question. The classification of the mass leads to a likelihood of malignancy.

Abstract: A computerized method for the detection and characterization of disease in an image derived from a chest radiograph, wherein an image in the chest radiograph is processed to determine the ribcage boundary, including lung top edges, right and left ribcage edges, and right and left hemidiaphragm edges. Texture measures including RMS variations of pixel values within regions of interest are converted to relative exposures and corrected for system noise existing in the system used to produce the image. Texture and/or geometric pattern indices are produced. A histogram(s) of the produced index (indices) is produced and values of the histogram(s) are applied as inputs to a trained artificial neural network, which classifies the image as normal or abnormal.

Abstract: A method for automated detection of abnormal anatomic regions, wherein a mammogram is digitized to produce a digital image and the digital image is processed using local edge gradient analysis and linear pattern analysis in addition to feature extraction routines to identify abnormal anatomic regions. Noise reduction filtering and pit-filling/spike-removal filtering techniques are also provided. Multiple difference imaging techniques are also used in which difference images employing different filter characteristics are obtained and processing results logically OR'ed to identify abnormal anatomic regions. In another embodiment the processing results with and without noise reduction filtering are logically AND'ed to improve detection sensitivity. Also, in another embodiment the wavelet transform is utilized in the identification and detection of abnormal regions.

Abstract: A computerized method and system using a shift-invariant artificial neural network (SIANN) for the quantitative analysis of image data. A series of digitized medical images are used to train an artificial neural network to differentiate between diseased and normal tissue. The sum of the weights in groups between layers is constrained to be substantially zero so as to avoid saturation of layers which would otherwise be saturated by low frequency background noise. The method and system also include utilizing training-free zones to exclude from training the center portions of microcalcifications in the digitized images. The method and system further include rule-based selection criteria for providing a more accurate diagnosis.

Abstract: A method for automated detection of abnormal anatomic regions, wherein a mammogram is digitized to produce a digital image and the digital image is processed using local edge gradient analysis and linear pattern analysis in addition to feature extraction routines to identify abnormal anatomic regions. Noise reduction filtering and pit-filling/spike-removal filtering techniques are also provided. Multiple difference imaging techniques are also used in which difference images employing different filter characteristics are obtained and processing results logically OR'ed to identify abnormal anatomic regions. In another embodiment the processing results with and without noise reduction filtering are logically AND'ed to improve detection sensitivity. Also, in another embodiment the wavelet transform is utilized in the identification and detection of abnormal regions.

Abstract: A method and system for the detection of anatomical features in a digital chest radiograph, and in particular rib edges and pneumothorax, wherein vertical profiles are defined in regions of the lungs in the radiograph, and a model function is fitted to the profiles to obtain initial estimates of the locations of rib edges. Gradient-histogram analysis is performed by determining edge gradient and their corresponding orientations in regions of interest (ROIs) defined in the radiograph in the vicinity of the initially estimated rib edges, determining the maximum edge gradient in each of the ROIs, and forming histograms which define accurate detections of rib edges. The edges are fitted with an elliptical function and a representation of the entire rib cage structure is obtained. The method and system are also applied to the detection of pneumothorax where ROIs are defined in the apical lung region and edge gradients and their orientations are detected in the ROIs.

Abstract: A method for automated detection of abnormal anatomic regions, wherein a mammogram is digitized to produce a digital image and the digital image is processed using local edge gradient analysis and linear pattern analysis in addition to feature extraction routines to identify abnormal anatomic regions. Noise reduction filtering and pit-filling/spike-removal filtering techniques are also provided. Multiple difference imaging techniques are also used in which difference images employing different filter characteristics are obtained and processing results logically OR'ed to identify abnormal anatomic regions. In another embodiment the processing results with and without noise reduction filtering are logically AND'ed to improve detection sensitivity. Also, in another embodiment the wavelet transform is utilized in the identification and detection of abnormal regions.

Abstract: A method and system for computer-aided differential diagnosis of diseases, and in particular, computer-aided differential diagnosis using neural networks. A first embodiment of the neural network distinguishes between a plurality of interstitial lung diseases on the basis of inputted clinical parameters and radiographic information. A second embodiment distinguishes between malignant and benign mammographic cases based upon similar inputted clinical and radiographic information. The neural networks were first trained using a hypothetical data base made up of hypothetical cases for each of the interstitial lung diseases and for malignant and benign cases. The performance of the neural network was evaluated using receiver operating characteristics (ROC) analysis. The decision performance of the neural network was compared to experienced radiologists and achieved a high performance comparable to that of the experienced radiologists.

Abstract: A method for automated detection of abnormal anatomic regions, wherein a mammogram is digitized to produce a digital image and the digital image is processed using local edge gradient analysis and linear pattern analysis in addition to feature extraction routines to identify abnormal anatomic regions. Noise reduction filtering and pit-filling/spike-removal filtering techniques are also provided. Multiple difference imaging techniques are also used in which difference images employing different filter characteristics are obtained and processing results logically OR'ed to identify abnormal anatomic regions. In another embodiment the processing results with and without noise reduction filtering are logically AND'ed to improve detection sensitivity. Also, in another embodiment the wavelet transform is utilized in the identification and detection of abnormal regions.

Abstract: A method of computerized detection of clustered microcalcifications in digital mammograms, including obtaining a digitized mammogram, deriving a difference image signal from the digitized mammogram, performing global grey-level thresholding, area filtering, and local grey-level thresholding on the difference image, in that order, performing a texture discrimination of the signal extracted from the previous step, performing a cluster filtering technique on the texture discriminated signals to identify locations in the digitized mammogram corresponding to candidate clustered microcalcifications, performing a feature extraction step in which the area, contrast and background pixel values of signals corresponding to the candidate clustered microcalcifications in the original image are extracted, performing thresholding tests based on the extracted features to eliminate from the candidate clustered microcalcifications particular candidate clustered microcalcification identified as corresponding to false-positive i