Matthew T. Freedman

Computerized radiographic mass detection. I. Lesion site selection by morphological enhancement and contextual segmentation

This paper presents a statistical model supported approach for enhanced segmentation and extraction of suspicious mass areas from mammographic images. With an appropriate statistical description of various discriminate characteristics of both true and false candidates from the localized areas, an improved mass detection may be achieved in computer-assisted diagnosis (CAD). In this study, one type of morphological operation is derived to enhance disease patterns of suspected masses by cleaning up unrelated background clutters, and a model-based image segmentation is performed to localize the suspected mass areas using a stochastic relaxation labeling scheme. We discuss the importance of model selection when a finite generalized Gaussian mixture is employed, and use the information theoretic criteria to determine the optimal model structure and parameters. Examples are presented to show the effectiveness of the proposed methods on mass lesion enhancement and segmentation when applied to mammographical images. Experimental results demonstrate that the proposed method achieves a very satisfactory performance as a preprocessing procedure for mass detection in CAD.

Automatic lung nodule detection using profile matching and back-propagation neural network techniques.

The potential advantages of using digital techniques instead of film-based radiography have been discussed extensively for the past 10 years. A major future application of digital techniques is computer-assisted diagnosis: the use of computer techniques to assist the radiologist in the diagnostic process. One aspect of this assistance is computer-assisted detection. The detection of small lung nodule has been recognized as a clinically difficult task for many years. Most of the literature has indicated that the rate for finding lung nodules (size range from 3 mm to 15 mm) is only approximately 65%, in those cases in which the undetected nodules could be found retrospectively. In recent published research, image processing techniques, such as thresholding and morphological analysis, have been used to enhance true-positive detection. However, these methods still produce many false-positive detections. We have been investigating the use of neural networks to distinguish true-positives nodule detections among those areas of interest that are generated from a signal enhanced image. The initial results show that the trained neural networks program can increase true-positive detections and moderately reduce the number of false-positive detections. The program reported here can perform three modes of lung nodule detection: thresholding, profile matching analysis, and neural network. This program is fully automatic and has been implemented in a DEC 5000/200 (Digital Equipment Corp, Maynard, MA) workstation. The total processing time for all three methods is less than 35 seconds. In this report, key image processing techniques and neural network for the lung nodule detection are described and the results of this initial study are reported.

A multiple circular path convolution neural network system for detection of mammographic masses

A multiple circular path convolution neural network (MCPCNN) architecture specifically designed for the analysis of tumor and tumor-like structures has been constructed. We first divided each suspected tumor area into sectors and computed the defined mass features for each sector independently. These sector features were used on the input layer and were coordinated by convolution kernels of different sizes that propagated signals to the second layer in the neural network system. The convolution kernels were trained, as required, by presenting the training cases to the neural network. In this study, randomly selected mammograms were processed by a dual morphological enhancement technique. Radiodense areas were isolated and were delineated using a region growing algorithm. The boundary of each region of interest was then divided into 36 sectors using 36 equi-angular dividers radiated from the center of the region. A total of 144 Breast Imaging-Reporting and Data System-based features (i.e., four features per sector for 36 sectors) were computed as input values for the evaluation of this newly invented neural network system. The overall performance was 0.78-0.80 for the areas (A/sub z/) under the receiver operating characteristic curves using the conventional feed-forward neural network in the detection of mammographic masses. The performance was markedly improved with A/sub z/ values ranging from 0.84 to 0.89 using the MCPCNN. This paper does not intend to claim the best mass detection system. Instead it reports a potentially better neural network structure for analyzing a set of the mass features defined by an investigator.

Probabilistic principal component subspaces: a hierarchical finite mixture model for data visualization

Visual exploration has proven to be a powerful tool for multivariate data mining and knowledge discovery. Most visualization algorithms aim to find a projection from the data space down to a visually perceivable rendering space. To reveal all of the interesting aspects of multimodal data sets living in a high-dimensional space, a hierarchical visualization algorithm is introduced which allows the complete data set to be visualized at the top level, with clusters and subclusters of data points visualized at deeper levels. The methods involve hierarchical use of standard finite normal mixtures and probabilistic principal component projections, whose parameters are estimated using the expectation-maximization and principal component neural networks under the information theoretic criteria.We demonstrate the principle of the approach on several multimodal numerical data sets, and we then apply the method to the visual explanation in computer-aided diagnosis for breast cancer detection from digital mammograms.

Optimization of wavelet decomposition for image compression and feature preservation

A neural-network-based framework has been developed to search for an optimal wavelet kernel that can be used for a specific image processing task. In this paper, a linear convolution neural network was employed to seek a wavelet that minimizes errors and maximizes compression efficiency for an image or a defined image pattern such as microcalcifications in mammograms and bone in computed tomography (CT) head images. We have used this method to evaluate the performance of tap-4 wavelets on mammograms, CTs, magnetic resonance images, and Lena images. We found that the Daubechies wavelet or those wavelets with similar filtering characteristics can produce the highest compression efficiency with the smallest mean-square-error for many image patterns including general image textures as well as microcalcifications in digital mammograms. However, the Haar wavelet produces the best results on sharp edges and low-noise smooth areas. We also found that a special wavelet (whose low-pass filter coefficients are 0.32252136, 0.85258927, 0.38458542, and -0.14548269) produces the best preservation outcomes in all tested microcalcification features including the peak signal-to-noise ratio, the contrast and the figure of merit in the wavelet lossy compression scheme. Having analyzed the spectrum of the wavelet filters, we can find the compression outcomes and feature preservation characteristics as a function of wavelets. This newly developed optimization approach can be generalized to other image analysis applications where a wavelet decomposition is employed.

Technology improvements for image-guided and minimally invasive spine procedures

This paper reports on technology developments aimed at improving the state of the art for image-guided minimally invasive spine procedures. Back pain is a major health problem with serious economic consequences. Minimally invasive procedures to treat back pain are rapidly growing in popularity due to improvements in technique and the substantially reduced trauma to the patient versus open spinal surgery. Image guidance is an enabling technology for minimally invasive procedures, but technical problems remain that may limit the wider applicability of these techniques. The paper begins with a discussion of low back pain and the potential shortcomings of open back surgery. The advantages of minimally invasive procedures are enumerated, followed by a list of technical problems that must be overcome to enable the more widespread dissemination of these techniques. The technical problems include improved intraoperative imaging, fusion of images from multiple modalities, the visualization of oblique paths, percutaneous spine tracking, mechanical instrument guidance, and software architectures for technology integration. Technical developments to address some of these problems are discussed next. The discussion includes intraoperative computerized tomography (CT) imaging, magnetic resonance imaging (MRI)/CT image registration, three-dimensional (3-D) visualization, optical localization, and robotics for percutaneous instrument placement. Finally, the paper concludes by presenting several representative clinical applications: biopsy, vertebroplasty, nerve and facet blocks, and shunt placement. The program presented here is a first step to developing the physician-assist systems of the future, which will incorporate visualization, tracking, and robotics to enable the precision placement and manipulation of instruments with minimal trauma to the patient.

Image-guided robotic delivery system for precise placement of therapeutic agents

The effectiveness of conventional solid tumor treatment is limited by the systemic toxicity and lack of specificity of chemotherapeutic agents. Present treatment modalities are frequently insufficient to eliminate competent cancer cells without exceeding the limits of toxicity to normal tissue. The coming generation of cancer therapeutics depends on the precise targeting and sustained release of antitumor agents to overcome these limitations. We are developing an image-guided, robotic system for precise intratumoral placement of anticancer drugs and sustained release devices to advance this new treatment paradigm. The robotic system will use intraoperatively obtained computed tomographic (CT) images from a mobile CT scanner for guidance. The concept is to track patient anatomy and localize instruments using currently available optical tracking technology. Tracking will also be used to register patient anatomy with the images. The physician can then use the registered image to select an appropriate tumor target and entry location and to plan the instrument path. This path will then be transmitted to the robot, which orients and drives the instrument to the desired target under physician control. Achievement of the target is confirmed via intraoperative CT. This system will provide instrument guidance that is precise, direct, and controllable. Error due to poor target visualization and hand unsteadiness should be reduced greatly. The basic components of the system (robot, mobile CT, tracking) have been demonstrated in our laboratory, and the integration of the components is in progress. In future work, we plan to fuse preoperative PET imaging with intraoperative CT to allow functional as well as anatomic image guidance.

Computerized radiographic mass detection. II. Decision support by featured database visualization and modular neural networks

For pt.I see ibid., vol.20, no.4, p.289-301 (2001). Based on the enhanced segmentation of suspicious mass areas, further development of computer-assisted mass detection may be decomposed into three distinctive machine learning tasks: (1) construction of the featured knowledge database; (2) mapping of the classified and/or unclassified data points in the datahase; and (3) development of an intelligent user interface. A decision support system may then be constructed as a complementary machine observer that should enhance the radiologists performance in mass detection, We adopt a mathematical feature extraction procedure to construct the featured knowledge database from all the suspicious mass sites localized by the enhanced segmentation. The optimal mapping of the data points is then obtained by learning the generalized normal mixtures and decision boundaries, where a probabilistic modular neural network (PMNN) is developed to carry out both soft and hard clustering. A visual explanation of the decision making is further invented as a decision support, based on an interactive visualization hierarchy through the probabilistic principal component projections of the knowledge database and the localized optimal displays of the retrieved raw data. A prototype system is developed and pilot tested to demonstrate the applicability of this framework to mammographic mass detection.

Classification of microcalcifications in digital mammograms using trend-oriented radial basis function neural network

Abstract We proposed some novel classification features for the microcalcification of mammograms, and selected the effective combined features using Karhunen–Loeve (KL) transformation followed by the restricted Euclidean distance measure, and finally applied the proposed trend-oriented radial basis function neural network (TRBF-NN) to distinguish the benign group from the malignant group and evaluate the performance with the round-robin method. The two-dimensional KL features were more distinguishable than the raw two-dimensional features. The TRBF-NN was able to define the more generalized distribution than those distributions defined by the conventional RBF-NNs. According to the receiver operating characteristic analysis, the proposed system performed better than two trained radiologists.

A preliminary study of content-based mammographic masses retrieval

The purpose of this study is to develop a Content-Based Image Retrieval (CBIR) system for mammographic computer-aided diagnosis. We have investigated the potential of using shape, texture, and intensity features to categorize masses that may lead to sorting similar image patterns in order to facilitate clinical viewing of mammographic masses. Experiments were conducted within a database that contains 243 masses (122 benign and 121 malignant). The retrieval performances using the individual feature was evaluated, and the best precision was determined to be 79.9% when using the curvature scale space descriptor (CSSD). By combining several selected shape features for retrieval, the precision was found to improve to 81.4%. By combining the shape, texture, and intensity features together, the precision was found to improve to 82.3%.