0000000000400388
AUTHOR
Alina Sultana
Automatic detection of cervical cells in Pap-smear images using polar transform and k-means segmentation
We introduce a novel method of cell detection and segmentation based on a polar transformation. The method assumes that the seed point of each candidate is placed inside the nucleus. The polar representation, built around the seed, is segmented using k-means clustering into one candidate-nucleus cluster, one candidate-cytoplasm cluster and up to three miscellaneous clusters, representing background or surrounding objects that are not part of the candidate cell. For assessing the natural number of clusters, the silhouette method is used. In the segmented polar representation, a number of parameters can be conveniently observed and evaluated as fuzzy memberships to the non-cell class, out of …
Automatic detection of hemangiomas using unsupervised segmentation of regions of interest
In this paper we compare the performances of three automatic methods of identifying hemangioma regions in images: 1) unsupervised segmentation using the Otsu method, 2) Fuzzy C-means clustering (FCM) and 3) an improved region growing algorithm based on FCM (RG-FCM). For each image, the starting point of the algorithms is a rectangular region of interest (ROI) containing the hemangioma. For computing the performances of each method, the ROIs had been manually labeled in 2 classes: pixels of hemangioma and pixels of non-hemangioma. The computed scores are given separately for each image, as well as global performances across all ROIs for both classes. The best classification of non-hemangioma…
Automatic Detection of Hemangioma through a Cascade of Self-organizing Map Clustering and Morphological Operators
Abstract In this paper we propose a method for the automatic detection of hemangioma regions, consisting of a cascade of algorithms: a Self Organizing Map (SOM) for clustering the image pixels in 25 classes (using a 5x5 output layer) followed by a morphological method of reducing the number of classes (MMRNC) to only two classes: hemangioma and non-hemangioma. We named this method SOM-MMRNC. To evaluate the performance of the proposed method we have used Fuzzy C-means (FCM) for comparison. The algorithms were tested on 33 images; for most images, the proposed method and FCM obtain similar overall scores, within one percent of each other. However, in about 18% of the cases, there is a signif…
Automatic Monitoring System for the Evolution of the Hemangiomas
In this paper we describe an automatic monitoring system for the evolution of infantile hemangiomas using a fuzzy logic system based on two parameters: area and redness. To follow the evolution, we have used for each subject pairs of images at different moments of time. The starting points of the algorithm are the rectangular regions of interest (ROI), manually selected for each of the two images, and automatically segmented using Otsu’s method in combination with different preprocessing methods. Using the results of segmentation, we could compute the evolution of the area and the evolution of the redness of hemangioma. These two parameters were used as input for the fuzzy logic system, obt…
Automatic Detection of Infantile Hemangioma using Convolutional Neural Network Approach
Infantile hemangioma is the most common tumor of childhood. This study proposes an automatic detection as a preliminary step for a further accurate monitoring tool to evaluate the clinical status of hemangioma. For the detection of hemangioma pixels, a convolutional neural network (CNN) was trained on patches of two classes (hemangioma and nonhemangioma) from the train dataset, and then it was used to classify all the pixels of the region of interest from the test dataset. In order to evaluate the results of segmentation obtained with CNN, the region of interest of the test dataset was also segmented using two classical methods of segmentation: fuzzy c-means clustering (FCM) and segmentatio…
Automatic monitoring system for the detection and evaluation of the evolution of hemangiomas
In this paper we introduce an automatic monitoring system for the detection and the evaluation of the evolution of hemangiomas using a fuzzy logic system based on two parameters: area and redness. We have considered pairs of images (from two different moments in time) that show hemangiomas either evolving, stationary or regressing. The starting points of the algorithm are the rectangular regions of interest (ROI), manually selected for each of the two images, and automatically segmented using Fuzzy C-means. Using the area and the redness of the hemagiomas extracted with Fuzzy C-means, for the same patient, at different moments of time, the algorithm decides whether the hemangioma is evolvin…
Infantile Hemangioma Detection using Deep Learning
Infantile hemangiomas are the most common type of benign tumor which appear in the first weeks of life. As currently there is no robust protocol to monitor and assess the hemangioma status, this study proposes a preliminary method to detect the lesion. Therefore, in this paper we describe a hemangiomas classifier based on a linear convolutional neural network architecture. The challenge was to achieve a good classification using a relatively small internal database of 240 images from 40 different patients. The results are promising as the CNN performance evaluation showed a level of accuracy on the test set of 93.84%. Five metrics were calculated to assess the proposed model performances: a…