Microscopic Cell Nuclei Segmentation Based on Adaptive Attention Window

This paper presents an adaptive attention window (AAW)-based microscopic cell nuclei segmentation method. For semantic AAW detection, a luminance map is used to create an initial attention window, which is then reduced close to the size of the real region of interest (ROI) using a quad-tree. The purpose of the AAW is to facilitate background removal and reduce the ROI segmentation processing time. Region segmentation is performed within the AAW, followed by region clustering and removal to produce segmentation of only ROIs. Experimental results demonstrate that the proposed method can efficiently segment one or more ROIs and produce similar segmentation results to human perception. In future work, the proposed method will be used for supporting a region-based medical image retrieval system that can generate a combined feature vector of segmented ROIs based on extraction and patient data.     

An Automated Neural-Fuzzy Approach to Malignant Tumor Localization in 2D Ultrasonic Images of the Prostate

In this paper, a new neural-fuzzy approach is proposed for automated region segmentation in transrectal ultrasound images of the prostate. The goal of region segmentation is to identify suspicious regions in the prostate in order to provide decision support for the diagnosis of prostate cancer. The new automated region segmentation system uses expert knowledge as well as both textural and spatial features in the image to accomplish the segmentation. The textural information is extracted by two recurrent random pulsed neural networks trained by two sets of data (a suspicious tissues’ data set and a normal tissues’ data set). Spatial information is captured by the atlas-based reference approach and is represented as fuzzy membership functions. The textural and spatial features are synthesized by a fuzzy inference system, which provides a binary classification of the region to be evaluated.     

Prostate cancer multi-feature analysis using trans-rectal ultrasound images

This note focuses on extracting and analysing prostate texture features from trans-rectal ultrasound (TRUS) images for tissue characterization. One of the principal contributions of this investigation is the use of the information of the images' frequency domain features and spatial domain features to attain a more accurate diagnosis. Each image is divided into regions of interest (ROIs) by the Gabor multi-resolution analysis, a crucial stage, in which segmentation is achieved according to the frequency response of the image pixels. The pixels with a similar response to the same filter are grouped to form one ROI. Next, from each ROI two different statistical feature sets are constructed; the first set includes four grey level dependence matrix (GLDM) features and the second set consists of five grey level difference vector (GLDV) features. These constructed feature sets are then ranked by the mutual information feature selection (MIFS) algorithm. Here, the features that provide the maximum mutual information of each feature and class (cancerous and non-cancerous) and the minimum mutual information of the selected features are chosen, yielding a reduced feature subset. The two constructed feature sets, GLDM and GLDV, as well as the reduced feature subset, are examined in terms of three different classifiers: the condensed k-nearest neighbour (CNN), the decision tree (DT) and the support vector machine (SVM). The accuracy classification results range from 87.5% to 93.75%, where the performance of the SVM and that of the DT are significantly better than the performance of the CNN.     

Automatic segmentation of left ventricle cavity from short-axis cardiac magnetic resonance images

In this paper, a computational framework is proposed to perform a fully automatic segmentation of the left ventricle (LV) cavity from short-axis cardiac magnetic resonance (CMR) images. In the initial phase, the region of interest (ROI) is automatically identified on the first image frame of the CMR slices. This is done by partitioning the image into different regions using a standard fuzzy c-means (FCM) clustering algorithm where the LV region is identified according to its intensity, size and circularity in the image. Next, LV segmentation is performed within the identified ROI by using a novel clustering method that utilizes an objective functional with a dissimilarity measure that incorporates a circular shape function. This circular shape-constrained FCM algorithm is able to differentiate pixels with similar intensity but are located in different regions (e.g. LV cavity and non-LV cavity), thus improving the accuracy of the segmentation even in the presence of papillary muscles. In the final step, the segmented LV cavity is propagated to the adjacent image frame to act as the ROI. The segmentation and ROI propagation are then iteratively executed until the segmentation has been performed for the whole cardiac sequence. Experiment results using the LV Segmentation Challenge validation datasets show that our proposed framework can achieve an average perpendicular distance (APD) shift of 2.23 ± 0.50 mm and the Dice metric (DM) index of 0.89 ± 0.03, which is comparable to the existing cutting edge methods. The added advantage over state of the art is that our approach is fully automatic, does not need manual initialization and does not require a prior trained model.     

An observer study comparing spot imaging regions selected by radiologists and a computer for an automated stereo spot mammography technique

We are developing an automated stereo spot mammography technique for improved imaging of suspicious dense regions within digital mammograms. The technique entails the acquisition of a full-field digital mammogram, automated detection of a suspicious dense region within that mammogram by a computer aided detection (CAD) program, and acquisition of a stereo pair of images with automated collimation to the suspicious region. The latter stereo spot image is obtained within seconds of the original full-field mammogram, without releasing the compression paddle. The spot image is viewed on a stereo video display. A critical element of this technique is the automated detection of suspicious regions for spot imaging. We performed an observer study to compare the suspicious regions selected by radiologists with those selected by a CAD program developed at the University of Michigan. True regions of interest (TROIs) were separately determined by one of the radiologists who reviewed the original mammograms, biopsy images, and histology results. We compared the radiologist and computer-selected regions of interest (ROIs) to the TROIs. Both the radiologists and the computer were allowed to select up to 3 regions in each of 200 images (mixture of 100 CC and 100 MLO views). We computed overlap indices (the overlap index is defined as the ratio of the area of intersection to the area of interest) to quantify the agreement between the selected regions in each image. The averages of the largest overlap indices per image for the 5 radiologist-to-computer comparisons were directly related to the average number of regions per image traced by the radiologists (about 50% for 1 region/image, 84% for 2 regions/image and 96% for 3 regions/image). The average of the overlap indices with all of the TROIs was 73% for CAD and 76.8% +/- 10.0% for the radiologists. This study indicates that the CAD determined ROIs could potentially be useful for a screening technique that includes stereo spot mammography imaging.     

A completely automated CAD system for mass detection in a large mammographic database

Mass localization plays a crucial role in computer-aided detection (CAD) systems for the classification of suspicious regions in mammograms. In this article we present a completely automated classification system for the detection of masses in digitized mammographic images. The tool system we discuss consists in three processing levels: (a) Image segmentation for the localization of regions of interest (ROIs). This step relies on an iterative dynamical threshold algorithm able to select iso-intensity closed contours around gray level maxima of the mammogram. (b) ROI characterization by means of textural features computed from the gray tone spatial dependence matrix (GTSDM), containing second-order spatial statistics information on the pixel gray level intensity. As the images under study were recorded in different centers and with different machine settings, eight GTSDM features were selected so as to be invariant under monotonic transformation. In this way, the images do not need to be normalized, as the adopted features depend on the texture only, rather than on the gray tone levels, too. (c) ROI classification by means of a neural network, with supervision provided by the radiologist's diagnosis. The CAD system was evaluated on a large database of 3369 mammographic images [2307 negative, 1062 pathological (or positive), containing at least one confirmed mass, as diagnosed by an expert radiologist]. To assess the performance of the system, receiver operating characteristic (ROC) and free-response ROC analysis were employed. The area under the ROC curve was found to be Az = 0.783 +/- 0.008 for the ROI-based classification. When evaluating the accuracy of the CAD against the radiologist-drawn boundaries, 4.23 false positives per image are found at 80% of mass sensitivity.     

Incorporation of an iterative, linear segmentation routine into a mammographic mass CAD system

In previous research, we have developed a computer-aided detection (CAD) system designed to detect masses in mammograms. The previous version of our system employed a simple but imprecise method to localize the masses. In this research, we present a more robust segmentation routine for use with mammographic masses. Our hypothesis is that by more accurately describing the morphology of the masses, we can improve the CAD system's ability to distinguish masses from other mammographic structures. To test this hypothesis, we incorporated the new segmentation routine into our CAD system and examined the change in performance. The developed iterative, linear segmentation routine is a gray level-based procedure. Using the identified regions from the previous CAD system as the initial seeds, the new segmentation algorithm refines the suspicious mass borders by making estimates of the interior and exterior pixels. These estimates are then passed to a linear discriminant, which determines the optimal threshold between the interior and exterior pixels. After applying the threshold and identifying the object's outline, two constraints on the border are applied to reduce the influence of background noise. After the border is constrained, the process repeats until a stopping criterion is reached. The segmentation routine was tested on a study database of 183 mammographic images extracted from the Digital Database for Screening Mammography. Eighty-three of the images contained 50 malignant and 50 benign masses; 100 images contained no masses. The previously developed CAD system was used to locate a set of suspicious regions of interest (ROIs) within the images. To assess the performance of the segmentation algorithm, a set of 20 features was measured from the suspicious regions before and after the application of the developed segmentation routine. Receiver operating characteristic (ROC) analysis was employed on the ROIs to examine the discriminatory capabilities of each individual feature before and after the segmentation routine. A statistically significant performance increase was found in many of the individual features, particularly those describing the mass borders. To examine how the incorporation of the segmentation routine affected the performance of the overall CAD system, free-response ROC (FROC) analysis was employed. When considering only malignant masses, the FROC performance of the system with the segmentation routine appeared better than the previous system. When detecting 90% of the malignant masses, the previous system achieved 4.9 false positives per image (FPpI) compared to the post-segmentation system's 4.2 FPpI. At 80% sensitivity, the respective FPpI were 3.5 and 1.6.     

Implementation and evaluation of a new workflow for registration and segmentation of pulmonary MRI data for regional lung perfusion assessment

Recently it has been shown that regional lung perfusion can be assessed using time-resolved contrast-enhanced magnetic resonance (MR) imaging. Quantification of the perfusion images has been attempted, based on definition of small regions of interest (ROIs). Use of complete lung segmentations instead of ROIs could possibly increase quantification accuracy. Due to the low signal-to-noise ratio, automatic segmentation algorithms cannot be applied. On the other hand, manual segmentation of the lung tissue is very time consuming and can become inaccurate, as the borders of the lung to adjacent tissues are not always clearly visible. We propose a new workflow for semi-automatic segmentation of the lung from additionally acquired morphological HASTE MR images. First the lung is delineated semi-automatically in the HASTE image. Next the HASTE image is automatically registered with the perfusion images. Finally, the transformation resulting from the registration is used to align the lung segmentation from the morphological dataset with the perfusion images. We evaluated rigid, affine and locally elastic transformations, suitable optimizers and different implementations of mutual information (MI) metrics to determine the best possible registration algorithm. We located the shortcomings of the registration procedure and under which conditions automatic registration will succeed or fail. Segmentation results were evaluated using overlap and distance measures. Integration of the new workflow reduces the time needed for post-processing of the data, simplifies the perfusion quantification and reduces interobserver variability in the segmentation process. In addition, the matched morphological data set can be used to identify morphologic changes as the source for the perfusion abnormalities.     

Breast Ultrasound Image Classification Based on Multiple-Instance Learning

Breast ultrasound (BUS) image segmentation is a very difficult task due to poor image quality and speckle noise. In this paper, local features extracted from roughly segmented regions of interest (ROIs) are used to describe breast tumors. The roughly segmented ROI is viewed as a bag. And subregions of the ROI are considered as the instances of the bag. Multiple-instance learning (MIL) method is more suitable for classifying breast tumors using BUS images. However, due to the complexity of BUS images, traditional MIL method is not applicable. In this paper, a novel MIL method is proposed for solving such task. First, a self-organizing map is used to map the instance space to the concept space. Then, we use the distribution of the instances of each bag in the concept space to construct the bag feature vector. Finally, a support vector machine is employed for classifying the tumors. The experimental results show that the proposed method can achieve better performance: the accuracy is 0.9107 and the area under receiver operator characteristic curve is 0.96 (p < 0.005).     

Automated localization of implanted seeds in 3D TRUS images used for prostate brachytherapy

An algorithm has been developed in this paper to localize implanted radioactive seeds in 3D ultrasound images for a dynamic intraoperative brachytherapy procedure. Segmentation of the seeds is difficult, due to their small size in relatively low quality of transrectal ultrasound (TRUS) images. In this paper, intraoperative seed segmentation in 3D TRUS images is achieved by performing a subtraction of the image before the needle has been inserted, and the image after the seeds have been implanted. The seeds are searched in a "local" space determined by the needle position and orientation information, which are obtained from a needle segmentation algorithm. To test this approach, 3D TRUS images of the agar and chicken tissue phantoms were obtained. Within these phantoms, dummy seeds were implanted. The seed locations determined by the seed segmentation algorithm were compared with those obtained from a volumetric cone-beam flat-panel micro-CT scanner and human observers. Evaluation of the algorithm showed that the rms error in determining the seed locations using the seed segmentation algorithm was 0.98 mm in agar phantoms and 1.02 mm in chicken phantoms.     

Boundary delineation in transrectal ultrasound image for prostate cancer

This paper presents a new advanced automatic edge delineation model for the detection and diagnosis of prostate cancer on transrectal ultrasound (TRUS) images. The proposed model is to improve prostate boundary detection system by modifying a set of preprocessing algorithms including tree-structured nonlinear filter (TSF), directional wavelet transforms (DWT) and tree-structured wavelet transform (TSWT). The model consists of a preprocessing module and a segmentation module. The preprocessing module is implemented for noise suppression, image smoothing and boundary enhancement. The active contours model is used in the segmentation module for prostate boundary detection in two-dimensional (2D) TRUS images. Experimental results show that the addition of the preprocessing module improves the accuracy and sensitivity of the segmentation module, compared to the implementation of the segmentation module alone. It is believed that the proposed automatic boundary detection module for the TRUS images is a promising approach, which provides an efficient and robust detection and diagnosis strategy and acts as "second opinion" for the physician's interpretation of prostate cancer.     

Computer-aided diagnosis of prostate cancer in the peripheral zone using multiparametric MRI

This study evaluated a computer-assisted diagnosis (CADx) system for determining a likelihood measure of prostate cancer presence in the peripheral zone (PZ) based on multiparametric magnetic resonance (MR) imaging, including T2-weighted, diffusion-weighted and dynamic contrast-enhanced MRI at 1.5 T. Based on a feature set derived from grey-level images, including first-order statistics, Haralick features, gradient features, semi-quantitative and quantitative (pharmacokinetic modelling) dynamic parameters, four kinds of classifiers were trained and compared: nonlinear support vector machine (SVM), linear discriminant analysis, k-nearest neighbours and na?ve Bayes classifiers. A set of feature selection methods based on t-test, mutual information and minimum-redundancy-maximum-relevancy criteria were also compared. The aim was to discriminate between the relevant features as well as to create an efficient classifier using these features. The diagnostic performances of these different CADx schemes were evaluated based on a receiver operating characteristic (ROC) curve analysis. The evaluation database consisted of 30 sets of multiparametric MR images acquired from radical prostatectomy patients. Using histologic sections as the gold standard, both cancer and nonmalignant (but suspicious) tissues were annotated in consensus on all MR images by two radiologists, a histopathologist and a researcher. Benign tissue regions of interest (ROIs) were also delineated in the remaining prostate PZ. This resulted in a series of 42 cancer ROIs, 49 benign but suspicious ROIs and 124 nonsuspicious benign ROIs. From the outputs of all evaluated feature selection methods on the test bench, a restrictive set of about 15 highly informative features coming from all MR sequences was discriminated, thus confirming the validity of the multiparametric approach. Quantitative evaluation of the diagnostic performance yielded a maximal area under the ROC curve (AUC) of 0.89 (0.81-0.94) for the discrimination of the malignant versus nonmalignant tissues and 0.82 (0.73-0.90) for the discrimination of the malignant versus suspicious tissues when combining the t-test feature selection approach with a SVM classifier. A preliminary comparison showed that the optimal CADx scheme mimicked, in terms of AUC, the human experts in differentiating malignant from suspicious tissues, thus demonstrating its potential for assisting cancer identification in the PZ.     

Optimization of reference library used in content-based medical image retrieval scheme

Building an optimal image reference library is a critical step in developing the interactive computer-aided detection and diagnosis (I-CAD) systems of medical images using content-based image retrieval (CBIR) schemes. In this study, the authors conducted two experiments to investigate (1) the relationship between I-CAD performance and size of reference library and (2) a new reference selection strategy to optimize the library and improve I-CAD performance. The authors assembled a reference library that includes 3153 regions of interest (ROI) depicting either malignant masses (1592) or CAD-cued false-positive regions (1561) and an independent testing data set including 200 masses and 200 false-positive regions. A CBIR scheme using a distance-weighted K-nearest neighbor algorithm is applied to retrieve references that are considered similar to the testing sample from the library. The area under receiver operating characteristic curve (Az) is used as an index to evaluate the I-CAD performance. In the first experiment, the authors systematically increased reference library size and tested I-CAD performance. The result indicates that scheme performance improves initially from Az= 0.715 to 0.874 and then plateaus when the library size reaches approximately half of its maximum capacity. In the second experiment, based on the hypothesis that a ROI should be removed if it performs poorly compared to a group of similar ROIs in a large and diverse reference library, the authors applied a new strategy to identify "poorly effective" references. By removing 174 identified ROIs from the reference library, I-CAD performance significantly increases to Az = 0.914 (p < 0.01). The study demonstrates that increasing reference library size and removing poorly effective references can significantly improve I-CAD performance.     

Region of interest motion compensation for PET image reconstruction

A motion-incorporated reconstruction (MIR) method for gated PET imaging has recently been developed by several authors to correct for respiratory motion artifacts in PET imaging. This method however relies on a motion map derived from images (4D PET or 4D CT) of the entire field of view (FOV). In this study we present a region of interest (ROI)-based extension to this method, whereby only the motion map of a user-defined ROI is required and motion incorporation during image reconstruction is solely performed within the ROI. A phantom study and an NCAT computer simulation study were performed to test the feasibility of this method. The phantom study showed that the ROI-based MIR produced results that are within 1.26% of those obtained by the full image-based MIR approach when using the same accurate motion information. The NCAT phantom study on the other hand, further verified that motion of features of interest in an image can be estimated more efficiently and potentially more accurately using the ROI-based approach. A reduction of motion estimation time from 450 s to 30 and 73 s was achieved for two different ROIs respectively. In addition, the ROI-based approach showed a reduction in registration error of 43% for one ROI, which effectively reduced quantification bias by 44% and 32% using mean and maximum voxel values, respectively.     

Automatic noise robust registration of radiographs for subtraction using strategic local correlation: an application to radiographs of dental implants

Most of digital subtraction methods in dental radiography are based on registration using manual landmarks. We have developed an automatic registration method without using the manual selection of landmarks. By restricting a geometrical matching of images to a region of interest (ROI), we compare the cross-correlation coefficient only between the ROIs. The affine or perspective transform parameters satisfying maximum of cross-correlation between the local regions are searched iteratively by a fast searching strategy. The parameters are searched on the 14 scale image coarsely and then, the fine registration is performed on the original scale image. The developed method can match the images corrupted by Gaussian noise with the same accuracy for the images without any transform simulation. The registration accuracy of the perspective method shows a 17% improvement over the manual method. The application of the developed method to radiographs of dental implants provides an automatic noise robust registration with high accuracy in almost real time.     

Automated selection of trabecular bone regions in knee radiographs

Osteoarthritic (OA) changes in knee joints can be assessed by analyzing the structure of trabecular bone (TB) in the tibia. This analysis is performed on TB regions selected manually by a human operator on x-ray images. Manual selection is time-consuming, tedious, and expensive. Even if a radiologist expert or highly trained person is available to select regions, high inter- and intraobserver variabilities are still possible. A fully automated image segmentation method was, therefore, developed to select the bone regions for numerical analyses of changes in bone structures. The newly developed method consists of image preprocessing, delineation of cortical bone plates (active shape model), and location of regions of interest (ROI). The method was trained on an independent set of 40 x-ray images. Automatically selected regions were compared to the "gold standard" that contains ROIs selected manually by a radiologist expert on 132 x-ray images. All images were acquired from subjects locked in a standardized standing position using a radiography rig. The size of each ROI is 12.8 x 12.8 mm. The automated method results showed a good agreement with the gold standard [similarity index (SI) = 0.83 (medial) and 0.81 (lateral) and the offset =[-1.78, 1.27]x[-0.65,0.26] mm (medial) and [-2.15, 1.59]x[-0.58, 0.52] mm (lateral)]. Bland and Altman plots were constructed for fractal signatures, and changes of fractal dimensions (FD) to region offsets calculated between the gold standard and automatically selected regions were calculated. The plots showed a random scatter and the 95% confidence intervals were (-0.006, 0.008) and (-0.001, 0.011). The changes of FDs to region offsets were less than 0.035. Previous studies showed that differences in FDs between non-OA and OA bone regions were greater than 0.05. ROIs were also selected by a second radiologist and then evaluated. Results indicated that the newly developed method could replace a human operator and produces bone regions with an accuracy that is sufficient for fractal analyses of bone texture.     

Glioma grading using apparent diffusion coefficient map: application of histogram analysis based on automatic segmentation

The accurate diagnosis of glioma subtypes is critical for appropriate treatment, but conventional histopathologic diagnosis often exhibits significant intra‐observer variability and sampling error. The aim of this study was to investigate whether histogram analysis using an automatically segmented region of interest (ROI), excluding cystic or necrotic portions, could improve the differentiation between low‐grade and high‐grade gliomas. Thirty‐two patients (nine low‐grade and 23 high‐grade gliomas) were included in this retrospective investigation. The outer boundaries of the entire tumors were manually drawn in each section of the contrast‐enhanced T₁‐weighted MR images. We excluded cystic or necrotic portions from the entire tumor volume. The histogram analyses were performed within the ROI on normalized apparent diffusion coefficient (ADC) maps. To evaluate the contribution of the proposed method to glioma grading, we compared the area under the receiver operating characteristic (ROC) curves. We found that an ROI excluding cystic or necrotic portions was more useful for glioma grading than was an entire tumor ROI. In the case of the fifth percentile values of the normalized ADC histogram, the area under the ROC curve for the tumor ROIs excluding cystic or necrotic portions was significantly higher than that for the entire tumor ROIs (p < 0.005). The automatic segmentation of a cystic or necrotic area probably improves the ability to differentiate between high‐ and low‐grade gliomas on an ADC map. Copyright © 2014 John Wiley & Sons, Ltd.     

Identification of masses in digital mammogram using gray level co-occurrence matrices

Digital mammogram has become the most effective technique for early breast cancer detection modality. Digital mammogram takes an electronic image of the breast and stores it directly in a computer. The aim of this study is to develop an automated system for assisting the analysis of digital mammograms. Computer image processing techniques will be applied to enhance images and this is followed by segmentation of the region of interest (ROI). Subsequently, the textural features will be extracted from the ROI. The texture features will be used to classify the ROIs as either masses or non-masses. In this study normal breast images and breast image with masses used as the standard input to the proposed system are taken from Mammographic Image Analysis Society (MIAS) digital mammogram database. In MIAS database, masses are grouped into either spiculated, circumscribed or ill-defined. Additional information includes location of masses centres and radius of masses. The extraction of the textural features of ROIs is done by using gray level co-occurrence matrices (GLCM) which is constructed at four different directions for each ROI. The results show that the GLCM at 0o, 45o, 90o and 135o with a block size of 8X8 give significant texture information to identify between masses and non-masses tissues. Analysis of GLCM properties i.e. contrast, energy and homogeneity resulted in receiver operating characteristics (ROC) curve area of Az = 0.84 for Otsu's method, 0.82 for thresholding method and Az = 0.7 for K-mean clustering. ROC curve area of 0.8-0.9 is rated as good results. The authors' proposed method contains no complicated algorithm. The detection is based on a decision tree with five criterions to be analysed. This simplicity leads to less computational time. Thus, this approach is suitable for automated real-time breast cancer diagnosis system.     

Regional Context-Sensitive Support Vector Machine Classifier to Improve Automated Identification of Regional Patterns of Diffuse Interstitial Lung Disease

We propose the use of a context-sensitive support vector machine (csSVM) to enhance the performance of a conventional support vector machine (SVM) for identifying diffuse interstitial lung disease (DILD) in high-resolution computerized tomography (HRCT) images. Nine hundred rectangular regions of interest (ROIs), each 20 × 20 pixels in size and consisting of 150 ROIs representing six regional disease patterns (normal, ground-glass opacity, reticular opacity, honeycombing, emphysema, and consolidation), were marked by two experienced radiologists using consensus HRCT images of various DILD. Twenty-one textual and shape features were evaluated to characterize the ROIs. The csSVM classified an ROI by simultaneously using the decision value of each class and information from the neighboring ROIs, such as neighboring region feature distances and class differences. Sequential forward-selection was used to select the relevant features. To validate our results, we used 900 ROIs with fivefold cross-validation and 84 whole lung images categorized by a radiologist. The accuracy of the proposed method for ROI and whole lung classification (89.88 ± 0.02%, and 60.30 ± 13.95%, respectively) was significantly higher than that provided by the conventional SVM classifier (87.39 ± 0.02%, and 57.69 ± 13.31%, respectively; paired t test, p < 0.01, and p < 0.01, respectively). We conclude that our csSVM provides better overall quantification of DILD.     

Assessment of Performance and Reliability of Computer-Aided Detection Scheme Using Content-Based Image Retrieval Approach and Limited Reference Database

Content-based image retrieval approach was used in our computer-aided detection (CAD) schemes for breast cancer detection with mammography. In this study, we assessed CAD performance and reliability using a reference database including 1500 positive (breast mass) regions of interest (ROIs) and1500 normal ROIs. To test the relationship between CAD performance and the similarity level between the queried ROI and the retrieved ROIs, we applied a set of similarity thresholds to the retrieved similar ROIs selected by the CADschemes for all queried suspicious regions, and used only the ROIs that were above the threshold for assessing CAD performance at each threshold level. Using the leave-one-out testing method, we computed areas under receiver operating characteristic (ROC) curves (A _Z ) to assess CAD performance. The experimental results showed that as threshold increase, (1) less true positive ROIs can be referenced in the database than normal ROIs and (2) the A _Z value was monotonically increased from 0.854±0.004 to 0.932±0.016. This study suggests that (1) in order to more accurately detect and diagnose subtle masses, a large and diverse database is required, and (2) assessing the reliability of the decision scores based on the similarity measurement is important in application of the CBIR-based CAD schemes when the limited database is used.