3-D Reconstruction of Microcalcification Clusters Using Stereo Imaging: Algorithm and Mammographic Unit Calibration

A simple multi-feature based stereoscopic medical image retrieval system

This paper describes a method of retrieving stereoscopic medical images from the database that consists of feature extraction, similarity measure, and re-ranking of retrieved images. This method retrieves similar images of the query image from the database and re-ranks them according to the disparity map. The performance is evaluated using the metrics namely average retrieval precision (APR) and average retrieval rate (ARR). According to the performance outcomes, the multi-feature based image retrieval using Mahalanobis distance measure has produced better result compared to other distance measures namely Euclidean, Minkowski, the sum of absolute difference (SAD) and the sum of squared absolute difference (SSAD). Therefore, the stereo image retrieval systems presented has high potential in biomedical image storage and retrieval systems.

Three-Dimensional Modeling From Endoscopic Video Using Geometric Constraints Via Feature Positioning

The endoscope is a popular imaging modality used in many preevaluations and surgical treatments, and is also one of the essential tools in minimally invasive surgery. However, regular endoscopes provide only 2-D images. Even though stereoendoscopy systems can display 3-D images, the real anatomical structure of the observed lesion is unavailable and can only be judged by the surgeon's imagination. In this paper, we present a constraint-based factorization method for reconstructing 3-D structures registered to the patient, from 2-D endoscopic images. The proposed method incorporates the geometric constraints from the tracked surgical instrument into the traditional factorization method based on frame-to-frame feature motion on the endoscopically viewed scene. Experiments with real and synthetic data demonstrate good real-scale 3-D extraction, with greater accuracy than is available from traditional methods. The reconstruction process can also be accomplished in a few seconds, making it suitable for on-line surgical applications to provide surgeons with additional 3-D shape information, critical distance monitoring and warnings.

Shape analysis of microcalcifications using Radon transform

Microcalcifications are one of the early signs of breast cancer, and they are of great importance for an early diagnosis. Moreover, the spatial distribution and the shape of the microcalcifications have a significant impact in medical practice to evaluate the probability of malignancy of the tumor. In this work a method, performing computer-aided classification of the shape of calcifications accordingly to the classification scheme proposed by Le Gal, is presented. In the first stage, in order to remove mammographic background, the image is preprocessed with a matched filter, designed by modeling the microcalcifications as Gaussian spots and the image as a Fractional Brownian Motion. Afterwards, morphology of spots has been evaluated using two different sets of parameters. The first set utilizes the moments of inertia of the second and third order to compute a set of features, which are invariant to rotations and translations of the image. The second set of parameters is derived from the evaluation of the Radon transform, as computed along eight axes. The results of the Radon transform are used to associate to each lesion a set of features, which are invariant to rotation and scaling of the image. In the final stage, a multilayer neural network has been used to assign each microcalcification to the classes introduced by Le Gal. The topology of the neural network is the same for both sets of descriptors, in order to allow comparison of the discriminative power of the two feature sets. Experimental results obtained with the proposed method from a set of digitized mammograms are reported and discussed.