Suresh Chandra Satapathy

An Improved Mammogram Classification Approach Using Back Propagation Neural Network

Mammograms are generally contaminated by quantum noise, degrading their visual quality and thereby the performance of the classifier in Computer-Aided Diagnosis (CAD). Hence, enhancement of mammograms is necessary to improve the visual quality and detectability of the anomalies present in the breasts. In this paper, a sigmoid based non-linear function has been applied for contrast enhancement of mammograms. The enhanced mammograms are used to define the texture of the detected anomaly using Gray Level Co-occurrence Matrix (GLCM) features. Later, a Back Propagation Artificial Neural Network (BP-ANN) is used as a classification tool for segregating the mammogram into abnormal or normal. The proposed classifier approach has reported to be the one with considerably better accuracy in comparison to other existing approaches.

DWT-PCA Image Fusion Technique to Improve Segmentation Accuracy in Brain Tumor Analysis

Because of its high clinical significance and varied modalities; magnetic resonance (MR) imaging procedures are widely adopted in medical discipline to record the abnormalities arising in a variety of internal organs of human body. Each modality of the MRI, such as T1, T2, T2C, Flair, and DW has its own merit and demerits. Hence, in the proposed work, a unique computer-assisted technique (CAT) is proposed to evaluate the abnormalities in MR images, irrespective of its modalities. Proposed CAT has the following stages: (i) Discrete Wavelet Transform Based Principal Component Averaging (DWT-PCA) image fusion, (ii) Tri-level thresholding based on Social Group Optimization and Shannon’s entropy, and (iii) Watershed segmentation. This approach is experimentally assessed with MICCAI brain cancer segmentation (BRATS 2013) challenge database. Experimental results confirms that the proposed approach is efficient in offering better values of Jaccard (84.33%), Dice (90.86%), sensitivity (99.93%), specificity (90.67%), and accuracy (95.74%) compared with the single modality registered brain MR images. Hence, the proposed work is extremely significant for the segmentation of abnormal region from the brain MR images registered using Flair, T1C, and T2 modalities.

ANN-Based Classification of Mammograms Using Nonlinear Preprocessing

Preprocessing and enhancement of mammograms is necessary to improve the visual quality and detectability of the anomalies present in the breasts. In this work, a nonlinear logistic function has been applied for enhancement (preprocessing) of mammograms. To define the texture of the detected anomaly gray level cooccurrence matrix (GLCM) features are formulated. Lastly, a feedforward artificial neural network (FF-ANN) is used as a classification tool for segregating the mammogram into normal or abnormal. A set of four confusion matrices regarding the learning, testing and validation of the classifier has been computed to analyze the performance of classifier at each stage. The proposed classifier approach has reported of considerably better accuracy in comparison to other existing approaches.

Visible-Infrared Image Fusion Method Using Anisotropic Diffusion

In this paper, Visible and Infrared sensors are used to take complementary images of a targeted scene. Image fusion thus aims to integrate the two images so that maximum information and fewer artifacts are introduced in the fused image. The concept of merging two different multisensor images using the combination of Anisotropic Diffusion (AD) and max–min approach is carried out in this paper. Herein, each of the registered source images are decomposed into approximation and detailed layers using AD filter. Later, max–min fusion rules are applied on detail and approximate layer, respectively, to preserve both spectral as well as structural information. Image-quality assessment of the fused images is made using structural similarity index (SSIM) , fusion factor (FF), and entropy (E) which justifies the effectiveness of proposed method.

Skin Melanoma Assessment Using Kapur’s Entropy and Level Set—A Study with Bat Algorithm

Skin melanoma is considered as a deadliest form of skin malformation originates in human community. Due to its increasing incidence rates, it is necessary to build an accompanying procedure to assist the clinical detection and diagnosis process. Visual examination and the digital dermoscopy are the two common procedures widely adopted by the doctors to detect and verify skin melanoma. This paper proposes a soft-computing assisted tool to investigate the skin melanoma images. In this work, bat algorithm-assisted Kapur’s multithresholding is considered to preprocess the image, and the level set-based segmentation is adopted in the postprocessing stage to mine the skin melanoma section. The experimental investigation is implemented using the benchmark DERMIS dataset. The effectiveness of proposed technique is confirmed by measuring the familiar image similarity measures through a relative study among extracted skin melanoma with the ground truth. The experimental result verifies that the proposed technique is easy to implement and offers superior values of Jaccard (0.8805), Dice (0.9138), sensitivity (0.9927), specificity (0.9177), and accuracy (0.9628).

Kapur’s Entropy and Active Contour-Based Segmentation and Analysis of Retinal Optic Disc

Retinal image scrutiny is essential to detect and supervise a wide variety of retinal infections. Segmentation of region of interest (ROI) from the retinal image is widely preferred to have a clear idea about the infected section. In the proposed work, a new two-stage approach is presented for automatic segmentation of the optic disc (OD) in retinal images. This approach includes the chaotic bat algorithm (CBA) assisted Kapur’s multi-thresholding as the preprocessing stage and active contour (AC) segmentation as the post-processing stage. This method initially identifies the suitable value of threshold to enhance the OD in the chosen retinal image. The enhanced OD is then processed using the gray scale morphological operation, and finally, the OD is extracted using AC segmentation process. To test the proposed approach, optic disc images of different category are acquired from the RIM-ONE database. Experimental results demonstrate that the average Jaccard index, Dice coefficient, precision, sensitivity, specificity, and accuracy are greater than 83.74, 93.66, 98.18, 92.85, 98.43, and 97.28%, respectively. Hence, the proposed work is extremely significant for the segmentation of OD and can be used as the automated screening tool for the OD related retinal diseases.

On the Convergence of Synthesis of Desired Nulls from Circular Arrays Using Flower Pollination Algorithm

In this paper, Synthesis of sum patterns from circular arrays to generate desired nulls with reduced sidelobes using flower pollination algorithm is presented. The array design is first formulated as an optimization problem with the goal of reducing peak sidelobe level with a deep null. The objective of the FPA algorithm is to determine the optimized set of amplitude excitation coefficients to obtain the desired pattern. The patterns are numerically computed for different constraints and the results obtained are compared with those of genetic algorithm in the present paper.

Flying Ad hoc Networks: A Comprehensive Survey

An ad hoc network is the cooperative disposition of a collection of dynamic (mobile) nodes without the necessity of existing infrastructure or any centralized access points. Recently, ad hoc networks have aroused great scientific curiosity and have led to wide-scale research works into this field. In this paper, we provide a complete survey on Flying Ad hoc Networks (FANETS) as an emerging field among Mobile Ad hoc Networks (MANETS) and Vehicular Ad hoc Networks (VANETS). FANET implies creating an ad hoc network between multi-UAV systems, which is connected to the base station. The base station can be remotely ground based or an aircraft. In FANET, communication between UAVs is dependent on node mobility and topological changes. In this paper, we provide a comprehensive survey of the design issues, communication methodologies, and routing protocols of UAVs with open research issues.