Bipin C. Desai

A Framework for Medical Image Retrieval Using Machine Learning and Statistical Similarity Matching Techniques With Relevance Feedback

A content-based image retrieval (CBIR) framework for diverse collection of medical images of different imaging modalities, anatomic regions with different orientations and biological systems is proposed. Organization of images in such a database (DB) is well defined with predefined semantic categories; hence, it can be useful for category-specific searching. The proposed framework consists of machine learning methods for image prefiltering, similarity matching using statistical distance measures, and a relevance feedback (RF) scheme. To narrow down the semantic gap and increase the retrieval efficiency, we investigate both supervised and unsupervised learning techniques to associate low-level global image features (e.g., color, texture, and edge) in the projected PCA-based eigenspace with their high-level semantic and visual categories. Specially, we explore the use of a probabilistic multiclass support vector machine (SVM) and fuzzy c-mean (FCM) clustering for categorization and prefiltering of images to reduce the search space. A category-specific statistical similarity matching is proposed in a finer level on the prefiltered images. To incorporate a better perception subjectivity, an RF mechanism is also added to update the query parameters dynamically and adjust the proposed matching functions. Experiments are based on a ground-truth DB consisting of 5000 diverse medical images of 20 predefined categories. Analysis of results based on cross-validation (CV) accuracy and precision-recall for image categorization and retrieval is reported. It demonstrates the improvement, effectiveness, and efficiency achieved by the proposed framework

Privacy-preserving trajectory data publishing by local suppression

The pervasiveness of location-aware devices has spawned extensive research in trajectory data mining, resulting in many important real-life applications. Yet, the privacy issue in sharing trajectory data among different parties often creates an obstacle for effective data mining. In this paper, we study the challenges of anonymizing trajectory data: high dimensionality, sparseness, and sequentiality. Employing traditional privacy models and anonymization methods often leads to low data utility in the resulting data and ineffective data mining. In addressing these challenges, this is the first paper to introduce local suppression to achieve a tailored privacy model for trajectory data anonymization. The framework allows the adoption of various data utility metrics for different data mining tasks. As an illustration, we aim at preserving both instances of location-time doublets and frequent sequences in a trajectory database, both being the foundation of many trajectory data mining tasks. Our experiments on both synthetic and real-life data sets suggest that the framework is effective and efficient to overcome the challenges in trajectory data anonymization. In particular, compared with the previous works in the literature, our proposed local suppression method can significantly improve the data utility in anonymous trajectory data.

Correlated network data publication via differential privacy

With the increasing prevalence of information networks, research on privacy-preserving network data publishing has received substantial attention recently. There are two streams of relevant research, targeting different privacy requirements. A large body of existing works focus on preventing node re-identification against adversaries with structural background knowledge, while some other studies aim to thwart edge disclosure. In general, the line of research on preventing edge disclosure is less fruitful, largely due to lack of a formal privacy model. The recent emergence of differential privacy has shown great promise for rigorous prevention of edge disclosure. Yet recent research indicates that differential privacy is vulnerable to data correlation, which hinders its application to network data that may be inherently correlated. In this paper, we show that differential privacy could be tuned to provide provable privacy guarantees even in the correlated setting by introducing an extra parameter, which measures the extent of correlation. We subsequently provide a holistic solution for non-interactive network data publication. First, we generate a private vertex labeling for a given network dataset to make the corresponding adjacency matrix form dense clusters. Next, we adaptively identify dense regions of the adjacency matrix by a data-dependent partitioning process. Finally, we reconstruct a noisy adjacency matrix by a novel use of the exponential mechanism. To our best knowledge, this is the first work providing a practical solution for publishing real-life network data via differential privacy. Extensive experiments demonstrate that our approach performs well on different types of real-life network datasets.

Medical image retrieval and registration: towards computer assisted diagnostic approach

A large number of medical images in digital format are generated by hospitals and clinics every day. Such images constitute an important source of anatomical and functional information for diagnosis of diseases, medical research and education. These images include high-resolution 2-D or 3-D spatial and temporal data of various modalities (CT, MRI, X-ray, ultrasound etc.) require effective processing and organization techniques. Two of the most important techniques along this direction are; image retrieval and image registration. Image retrieval is the technique to find similar images from an image archive by their textual and visual contents and image registration is the establishment of correspondences between images or between image and physical space. Characteristics of medical images differ significantly from other general-purpose images and hence require special treatment. This article presents a highlight of recent research in medical image retrieval and registration, main techniques and trends, key issues and limitations. We also point out some promising research directions and propose a system architecture for computer aided diagnosis, where both technologies can be integrated and complement each other.

A unified image retrieval framework on local visual and semantic concept-based feature spaces

This paper presents a learning-based unified image retrieval framework to represent images in local visual and semantic concept-based feature spaces. In this framework, a visual concept vocabulary (codebook) is automatically constructed by utilizing self-organizing map (SOM) and statistical models are built for local semantic concepts using probabilistic multi-class support vector machine (SVM). Based on these constructions, the images are represented in correlation and spatial relationship-enhanced concept feature spaces by exploiting the topology preserving local neighborhood structure of the codebook, local concept correlation statistics, and spatial relationships in individual encoded images. Finally, the features are unified by a dynamically weighted linear combination of similarity matching scheme based on the relevance feedback information. The feature weights are calculated by considering both the precision and the rank order information of the top retrieved relevant images of each representation, which adapts itself to individual searches to produce effective results. The experimental results on a photographic database of natural scenes and a bio-medical database of different imaging modalities and body parts demonstrate the effectiveness of the proposed framework.

Image Retrieval-Based Decision Support System for Dermatoscopic Images

This paper presents a content-based image retrieval system for dermatoscopic images as a diagnostic aid to the dermatologists for skin cancer recognition. In this context, the ultimate aim is to support decision making by locating, retrieving and displaying relevant past cases along with diagnostic reports. However, most challenging aspect in this domain is to extract local lesion specific image features and define the relevance between query and database images for retrieval. A fast and automatic segmentation method to detect the lesion from background healthy skin is proposed. This method first transforms a color image into an intensity image by utilizing domain specific image properties and NBS color distance in HVC color space. Lesion mask is detected by fusing individually segmented images based on iterative thresholding. Lesion specific local color and texture features are extracted and represented in the form of mean and variance-covariance of color channels and in a reduced PCA sub-space. Finally, for effective image retrieval, a similarity matching function is defined based on the fusion of a Bhattacharyya and Euclidean distance metric. The performance of the retrieval system is evaluated using average precision on a collection of 358 images, which demonstrates effectiveness of the proposed approach

Using semantic templates for a natural language interface to the CINDI virtual library

In this paper, we present our work in building a template-based system for translating English sentences into SQL queries for a relational database system. The input sentences are syntactically parsed using the Link Parser, and semantically parsed through the use of domain-specific templates. The system is composed of a pre-processor and a run-time module. The pre-processor builds a conceptual knowledge base from the database schema using WordNet. This knowledge base is then used at run time to semantically parse the input and create the corresponding SQL query. The system is meant to be domain independent and has been tested with the CINDI database that contains information on a virtual library.

Supporting discovery in virtual libraries

It is well known that selectivity leaves a lot to be desired in searching for information resources on the Internet with existing search systems (Desai, 1995c). This has prompted a number of researchers to turn their attention to the development and implementation of models for indexing and searching information resources on the Internet. In this article, 1 we examine briefly the results of a simple query on a number of existing search systems and then discuss two proposed index metadata structures for indexing and supporting search and discovery: The Dublin Core Elements List and the Semantic Header. We also present an indexing and discovery system based on the Semantic Header.

Privacy protection for RFID data

Radio Frequency IDentification (RFID) is a technology of automatic object identification. Retailers and manufacturers have created compelling business cases for deploying RFID in their supply chains. Yet, the uniquely identifiable objects pose a privacy threat to individuals. In this paper, we study the privacy threats caused by publishing RFID data. Even if the explicit identifying information, such as name and social security number, has been removed from the published RFID data, an adversary may identify a target victims record or infer her sensitive value by matching a priori known visited locations and timestamps. RFID data by default is high-dimensional and sparse, so applying traditional K-anonymity to RFID data suffers from the curse of high dimensionality, and would result in poor data usefulness. We define a new privacy model, develop an anonymization algorithm to accommodate special challenges on RFID data, and evaluate its performance in terms of data quality, efficiency, and scalability. To the best of our knowledge, this is the first work on anonymizing high-dimensional RFID data.

A multiple expert-based melanoma recognition system for dermoscopic images of pigmented skin lesions

This paper presents an integrated decision support system for an automated melanoma recognition of dermoscopic images based on multiple expert fusion. In this context, the ultimate aim is to support decision making by predicting image categories (e.g., melanoma, benign and dysplastic nevi) by combining outputs from different classifiers. A fast and automatic segmentation method to detect the lesion from the background healthy skin is proposed and lesion-specific local color and texture-related features are extracted. For the classification, combining experts which are classifiers with different structures, are examined as alternative solution instead of an individual classifier. In this approach, probabilistic outputs of the experts are combined based on the combination rules that are derived by following Bayespsila theorem. The category label with the highest confidence score is considered to be the class of a test image. Experimental results on a collection of 358 dermoscopic images demonstrate the effectiveness of the proposed expert fusion-based approach.