This research focuses on the depression states classification of EEG signals using EEGNet model optimized with Optuna. The purpose was to increase performance by combining data from healthy and depressed subjects, which ensured robustness across datasets. methodology comprised construction a preprocessing pipeline, included noise filtering, artifact removal, signal segmentation. Additive extraction time frequency domains further captured important features signals. developed merged dataset...

Glaucoma is one of the leading causes irreversible blindness around world and remains asymptomatic until its later stages. Therefore, early diagnosis crucial importance. The detection glaucoma in stages (from color fundus images) a challenging task, since clinical signs retinal images are very subtle go undetected most time by human eye. Convolutional neural networks have proven to provide good results for automatic features from images. In this work we explore possibility using residual...

Detection of pectoral muscle in mammograms is an important pre-processing segmentation step. The one the few anatomical features that appears clearly and reliably medio-lateral oblique view mammograms. This new method overcomes limitation straight-line representation considered our initial investigation using Hough transform.

Dermoscopy is the primary tool used for pigmented skin lesion diagnosis. Despite use of this relative new clinical method dermoscopy based, diagnose still subjective and diagnosis detection accuracy about 75-80%. In paper we present several enhancement pre-processing techniques applied on dermatoscopic images, such as black frame removal hair in an automatically manner. We have tested our algorithms 45 dermoscopic images compared automated methods results with other existing methods.

The Babes-Papanicolaou test (also known as Pap smear) is a method of cervical cancer screening used to detect abnormal cells which are or can become cancerous. Since the visual inspection pap smears very time consuming, need for automatic methods required. This paper presents an algorithm detection nuclei within images. relies in highly effective mean-shift filtering enhances contrast areas. segmentation consists region growing with starting points taken from image gradient map. Size and...

Medical imaging is an area of great interest in terms accuracy, speed and capacity integration. In order to improve results ease the physicians' task, some feature enhancement image processing should be done automatically lead features that allow automatic classification images. This paper presents original approach construct melanoma detection system, based on employing natural computing methods for preprocessing, extraction classification. Among these we rely cellular automata, reaction...

Total hip replacement is a common procedure in today orthopedics, with high rate of long-term success. Failure prevention based on regular follow-up aimed at checking the prosthes ...

Infantile hemangiomas are the most common types of tumors that found in infants and have an incidence approximately 10% population. Although infantile self-involuting, due to their fast proliferation they may threaten vital anatomical structures physiological functions; also, involution process take up several years. An accurate monitoring progress hemangioma growth regression is essential. We thus suggest using a computer aided follow-up these lesions by automatic detection quantifying...

We introduce a novel method of cell detection and segmentation based on polar transformation. The assumes that the seed point each candidate is placed inside nucleus. representation, built around seed, segmented using k-means clustering into one candidate-nucleus cluster, candidate-cytoplasm cluster up to three miscellaneous clusters, representing background or surrounding objects are not part cell. For assessing natural number silhouette used. In parameters can be conveniently observed...

In this paper we propose a method for the automatic detection of hemangioma regions, consisting cascade algorithms: Self Organizing Map (SOM) clustering image pixels in 25 classes (using 5x5 output layer) followed by morphological reducing number (MMRNC) to only two classes: and non-hemangioma. We named SOM-MMRNC. To evaluate performance proposed have used Fuzzy C-means (FCM) comparison. The algorithms were tested on 33 images; most images, FCM obtain similar overall scores, within one...

In this paper we compare the performances of three automatic methods identifying hemangioma regions in images: 1) unsupervised segmentation using Otsu method, 2) Fuzzy C-means clustering (FCM) and 3) an improved region growing algorithm based on FCM (RG-FCM). For each image, starting point algorithms is a rectangular interest (ROI) containing hemangioma. computing ROIs had been manually labeled 2 classes: pixels non-hemangioma. The computed scores are given separately for as well global...

On-board detection of driver's emotions has become a task high importance for car manufacturers, as negative appear to be one the major risks accidents. Deep neural networks have over last years state art methods computer vision and image classification. Yet, their success depends upon being trained on comprehensive database, which should cover all real-life situations that may arise in practice. Most in-car driver monitoring cameras capture images near infra-red (NIR) domain therefore needs...

A novel method for the detection and segmentation of nuclei cells in Pap smear images is introduced. The based on a geometric analysis iso- edge-contours. For we employ isocontours taken at different levels intensity report best (object) recall values as well precision values. cell outline detection, traditional edge-based contours show that with simple radial one can detect even agglomerated - which so far has been approached rather hesitantly. system was tested three databases.

Infantile hemangiomas (IH) are benign vascular tumors, most of them appearing in the first weeks and developing until six months age. The evaluation lesion size is usually made by physician through manual measurement, which inaccurate. This paper presents an algorithm for automatic segmentation hemangioma region, relying on Maximum a Posteriori (MAP) classification method. result improved regularization with discrete Markov fields (MAP-Markov). Then, further improvement performed, eliminated...

Skin cancer is one of the most common types cancer, and it caused by a variety dermatological conditions. Identifying abnormalities from skin images an important pre-diagnostic step to assist physicians in determining patient's condition. Thus, aid dermatologists diagnosis process, we proposed five CNN-based classification approaches namely ResNet-101, DenseNet-121, GoogLeNet, VGG16, MobileNetV2 architectures on which transfer learning process was applied. The HAM10000-N database consisting...

According to the World Health Organization, breast cancer is most common suffered by women in world, which during last two decades, has increased mortality developing countries. Mammography best method used for screening; problem of detecting possible areas very complex due, on one hand, diversity shape ill tissue and, other poorly defined border between healthy and cancerous zone. An automated technique alignment right left images been developed use computerized analysis bilateral images....

Detection and characterization of microcalcification clusters in mammograms is vital daily clinical practice. The problem detecting possible cancer areas very complex due, on one hand, to the diversity shape ill tissue and, other poorly defined border between healthy cancerous zone. Even though it has been studied for many years, there are still remaining challenges directions future research, such as developing better enhancement segmentation algorithms. In this paper, we propose a...

Infantile hemangiomas are the most common type of benign tumor which appear in first weeks life. As currently there is no robust protocol to monitor and assess hemangioma status, this study proposes a preliminary method detect lesion. Therefore, paper we describe classifier based on linear convolutional neural network architecture. The challenge was achieve good classification using relatively small internal database 240 images from 40 different patients. results promising as CNN performance...

Infantile hemangiomas are the most common types of tumors with an incidence approximately 10% in population. An accurate monitoring progress hemangioma growth and regression is essential for effective treatment. This study presents automatic evaluation evolution on a follow-up series images based color area features. A constancy approach applied to correct variation ambient illumination. The segmentation uses two-level thresholding some post-processing methods. proposed method has been...

This paper proposes an extension of the widely used logarithmic image processing (LIP) models by means parametrization. The mathematical structure mentioned is that cone or vector space. Once investigation defined boundaries these structures, parametric extensions known are straight-forward. It has been showed implementation Laplacian edge detector techniques under LIP model yields superior performance. In this we shall prove parametrization not only adds flexibility, but may also lead to quality.

In this paper we introduce an automatic monitoring system for the detection and evaluation of evolution hemangiomas using a fuzzy logic based on two parameters: area redness. We have considered pairs images (from different moments in time) that show either evolving, stationary or regressing. The starting points algorithm are rectangular regions interest (ROI), manually selected each images, automatically segmented Fuzzy C-means. Using redness hemagiomas extracted with C-means, same patient,...

Reaction-diffusion cellular neural networks have been studied because of their properties regarding processing time and capability integration. The possibility using them in image preprocessing and/or the advantages disadvantages doing that are analyzed. main objective is to find suitable sets parameters (genes) can enhance some useful a dermatoscopic image. These should later help classifying images into healthy or suspect.