Valério Ramos Batista

LBP operators on curvelet coefficients as an algorithm to describe texture in breast cancer tissues

We present a method based on curvelet transform, LBP, ANOVA and PL classifier.We validate the proposed approach considering the metrics accuracy and AUC.The features was evaluated by applying the DT, RaF, SVM and PL classifiers.The proposed approach achieved AC values among 91% and 100%.The method was tested on the datasets: DDSM, BCDR-FMR, BCDR-DMR and UCSB-BB. In computer-aided diagnosis one of the crucial steps to classify suspicious lesions is the extraction of features. Texture analysis methods have been used in the analysis and interpretation of medical images. In this work we present a method based on the association among curvelet transform, local binary patterns, feature selection by statistical analysis and distinct classification methods, in order to support the development of computer aided diagnosis system. The similar features were removed by the statistical analysis of variance (ANOVA). The understanding of the features was evaluated by applying the decision tree, random forest, support vector machine and polynomial (PL) classifiers, considering the metrics accuracy (AC) and area under the ROC curve (AUC): the rates were calculated on images of breast tissues with different physical properties (commonly observed in clinical practice). The datasets were the Digital Database for Screening Mammography, Breast Cancer Digital Repository and UCSB biosegmentation benchmark. The investigated groups were normal-abnormal and benign-malignant. The association of curvelet transform, local binary pattern and ANOVA with the PL classifier achieved higher AUC and AC values for all cases: the obtained rates were among 91% and 100%. These results are relevant, specially when we consider the difficulties of clinical practice in distinguishing the studied groups. The proposed association is useful as an automated protocol for the diagnosis of breast tissues and may contribute to the diagnosis of breast tissues (mammographic and histopathological images).

A limit-method for solving period problems on minimal surfaces

We introduce a new technique to solve period problems on minimal surfaces called limit-method. If a family of surfaces has Weierstrass-data converging to the data of a known example, and this presents a transversal solution of periods, then the original family contains a sub-family with closed periods.

An interactive programme for weighted Steiner trees

We introduce a fully written programmed code with a supervised method for generating weighted Steiner trees. Our choice of the programming language, and the use of well- known theorems from Geometry and Complex Analysis, allowed this method to be implemented with only 764 lines of effective source code. This eases the understanding and the handling of this beta version for future developments.

A software tool based on the Surface Evolver for precise location of tumours as a preoperative procedure to partial mastectomy

We present a fast and reliable program that gives precise location of breast tumours for a partial mastectomy. Our program is fully implemented in the Surface Evolver, which is a general-purpose simulator of physical experiments. By starting from the mammograms that show a tumour one takes its 2D coordinates in each view (CC and MLO). These coordinates, together with some measurements of the patients breast, are given as input to our simulator. From this point on the simulator reproduces all main steps of taking mammography with a virtual transparent breast that matches the patients. The virtual mammography procedure is graphically displayed on the computer screen, so that users can track the virtual tumour inside the breast. As output we have the coordinates of the tumour position when the woman lies on the operating table for the surgery. With these coordinates the surgeon can make a small incision into the breast and reach the tumour for its removal. The whole structure of the breast is preserved after a simple plastic correction.

Simulating CC and MLO compressions with the Surface Evolver

Mammographies are X-ray images of the breast under external compressions called Craniocaudal (CC) and Mediolateral Oblique (MLO). Together they increase the chances of detecting cancer but the breast is shown in strongly deformed shapes. Cancer location is highly uncertain for the surgery and so the breast is commonly taken out entirely, a serious trauma for the patient. In this paper we present a fully virtual mammography procedure that faithfully reproduces all shapes of the breast and in its inside tracks the cancer at any step. The cancer is then precisely located for the surgery and can be removed through a small incision. So the whole structure is preserved and cured as an integral benefit to the patient.

Human-computer interfaces applied to numerical solution of the Plateau problem

In this work we present a code in Matlab to solve the Problem of Plateau numerically, and the code will include human-computer interface. The Problem of Plateau has applications in areas of knowledge like, for instance, Computer Graphics. The solution method will be the same one of the Surface Evolver, but the difference will be a complete graphical interface with the user. This will enable us to implement other kinds of interface like ocular mouse, voice, touch, etc. To date, Evolver does not include any graphical interface, which restricts its use by the scientific community. Specially, its use is practically impossible for most of the Physically Challenged People.

Simulating external compressions of the breast with the Surface Evolver

In this paper we introduce a computational modelling that reproduces the breast compression processes used to obtain the mammogram. The main result is a programme in which one can track the first steps of virtual mammography. On the one hand, our modelling enables addition of structures that represent different tissues, muscles and glands in the breast. On the other hand, we shall validate and implement it by means of laboratory tests with phantoms. To the best of our knowledge, these two characteristics do confer originality to our research. This is because their interrelation seems not to be properly established elsewhere yet. We conclude that our model reproduces the same shapes and measurements really taken by the volunteers breasts.

Using wavelet sub-band and fuzzy 2-partition entropy to segment chronic lymphocytic leukemia images

Abstract Histological images analysis is an important procedure to diagnose different types of cancer. One of them is the chronic lymphocytic leukemia (CLL), which can be identified by applying image segmentation techniques. This study presents an unsupervised method to segment neoplastic nuclei in CLL images. Firstly, deconvolution, histogram equalization and mean filter were applied to enhance nuclear regions. Then, a segmentation technique based on a combination of wavelet transform, fuzzy 2-partition entropy and genetic algorithm was used, followed by removal of false positive regions, and application of valley-emphasis and morphological operations. In order to evaluate the proposed algorithm H&E-stained histological images were used. In the accuracy metric, the proposed method attained more than 80%, which can surpass similar methods. This proposal presents spatial distribution that has a good consistency with a manual segmentation and lower overlapping rate than other techniques in the literature.

EV: A First Version of the Surface Evolver with a Human-Computer Interface

The Surface Evolver is a general-purpose software with which one can simulate innumerous physical experiments. Since its first release in 1989, its sole developer and maintainer K. Brakke prioritised the technical part. This made Evolver applicable in many Science Areas, but its input is only through data files and command lines. Here we present EV, the first human-computer interface for Evolver totally written in Matlab. By means of eye tracker or head mouse the user can access virtual buttons and perform simulations compatible with Evolver. Although still restricted to the numerical solution of the Plateau problem, its Matlab code is at least 90% shorter than the corresponding C-code of Evolver. This will facilitate both the maintenance and the use of EV, specially if we consider physically challenged people in academia.

Stree: An Interactive Program for Weighted Steiner Trees

Here we present the extended version of a previous short publication about stree, which is a fully written program with a semi-automatic method that generates weighted Steiner trees. Our choice of the programming language, and the use of well-known theorems from Geometry and Complex Analysis, allowed this method to be implemented with only 764 lines of effective source code. This simplifies the understanding and the handling of this beta version for future developments.