Papers by PROF MAHUA BHATTACHARYA
Generation of novel encrypted code using cryptography for multiple level data security for Electronic Patient Record
Current paper represents a new type of encrypted Electronic Patient Record (EPR) code used for da... more Current paper represents a new type of encrypted Electronic Patient Record (EPR) code used for data encryption and content protection. EPR is a collection of several private information related to a patient which needs data authenticity, data security as well as safe and secured transmission. The proposed methodology used both cryptography and image processing techniques to build a new type of encrypted information code in image format which can be transmitted and used like bar code, QR code but even more secure. The success rate of recovery of data is 100% for both short and long messages. The information can also be retrieved at the receiver side exactly same without any loss of information. Use of RSA and DES algorithm consequently, with three keys and followed by some image processing techniques like complement, flip make the proposed algorithm more unbreakable. In this paper a complete Graphical User Interface (GUI) has been developed for both encoder-transmitter and decoder- receiver section.
InTech eBooks, May 16, 2012
Recent history has witnessed the rapid development in information technologies that has given an ... more Recent history has witnessed the rapid development in information technologies that has given an extended and easy access to digital information. Along with several developments it leads to the problem of illegal copying and redistribution of digital media. As a result the integrity and confidentiality of the digital information has come under immense threat. The concept of an emerging technology, digital watermarking came in order to solve the problems related to the intellectual property of media. (P.W. . Digital Watermarking is a technique which allows an individual to add hidden copyright notices or other verification messages or even classified information to digital media. , (Cox I. J., Miller M., Bloom J., 2002). Watermarks can either be visible or invisible. Here in this chapter we utilize the invisible technique. This is used in public information settings such as digital images libraries, museums, and art galleries and also in defense communication where data security is of prime importance. Watermark embedding utilizes two kinds of methods; one is in the spatial domain and the other in the transform domain. In the spatial domain the watermark is directly embedded into the image pixels whereas in the frequency domain the image is decomposed into blocks and then mapped into the transform domain (M. Kutter, F. A. P. . This is basically a process of hiding information in an image known as cover image. Copyright protection is achieved by robust watermarking while image authentication is usually achieved by fragile watermarking techniques. In the fragile watermarking scheme if any alteration of the message is found then it is broken and it can be easily detected as tampered by the provider of the watermark. In general, fragile schemes modify the leastsignificant-bits LSB planes of the original image in an irreversible way. Often a secret key is also used to encrypt the information (P.W. . Invertible watermarking is a new process which enables the exact recovery of the original image upon extraction of the embedded information (M.L. Miller, I.J. Cox, J.M.G. . This work implements both authentication and confidentiality in a reversible manner without affecting the image in any way. Security of images imposes three mandatory characteristics: confidentiality, reliability and availability (J. Fridrich,2002).
American Journal of Biomedical Engineering, Aug 31, 2012
The current paper presents a novel and unique scheme for biomedical image watermarking in wavelet... more The current paper presents a novel and unique scheme for biomedical image watermarking in wavelet domain by hiding multiple copies of the same data in the cover image using bit replacement in the horizontal (HL) and vertical (LH) resolution approximation image components. The proposed scheme uses an approach for recovering the hidden information from the damaged copies due to unauthorized alteration of the data by applying an algorithm to find the closest twin of the embedded information by bit majority algorithm. Experimental results of the proposed watermarking technique show much enhancement in the visual and statistical invisibility of hidden information after data recovery that supports the improvement in performance.
Journal of Cancer Science & Therapy, Jan 21, 2015
An image processing based study of the bone erosion process in osteoporosis using radiographic images : a shape metric approach
Image Processing and Communications, 1999
The Role of Bioinformatics and Imaging Models in Tumorigenesis and Treatment Response of Brain and Spinal Cord Neoplasm
Advances in Experimental Medicine and Biology, 2023
Automatic parameter setting of pulse coupled neural network for image segmentation
PCNN Model is widely used because it simulates the working of visual cortex in cats, but paramete... more PCNN Model is widely used because it simulates the working of visual cortex in cats, but parameter setting in PCNN is hefty affair because of manual adjusting of many initial parameters. Through this paper, we present an efficient optimization approach to reducing the parameter setting in original PCNN model by changing the threshold function to modified Heaviside function. Experimental result reveals that the proposed algorithm is efficient in detecting the segments in medical images.
A novel <scp>DeepML</scp> framework for multi‐classification of breast cancer based on transfer learning
International Journal of Imaging Systems and Technology, May 5, 2022
In the automated diagnosis of breast cancer (BC), microscopic images based on multi‐classificatio... more In the automated diagnosis of breast cancer (BC), microscopic images based on multi‐classification play a prominent role. Multi‐classification of BC means to differentiate among the sub‐categories of BC (papillary carcinoma, ductal carcinoma, fibroadenoma, etc.). However, unpretentious contrasts in various sub‐categories of BC occur due to the wide fluctuation of 1) excessive coherency of malignant cells, 2) high definition image appearance, and 3) excessive heterogeneity in color distribution, which makes the task more crucial. Therefore, the automated sub‐category discrimination using microscopic images has great medical diagnostic significance yet has not much explored. Thus, the present paper proposes a framework based on machine learning (ML) and deep learning (DL) to multi‐classify BC cells into 8 sub‐categories. These 8 sub‐categories comprise four kinds that delineate benigncy, and the other four portray malignancy. More appropriately, both the ML and DL models with the concept of transfer learning have been proposed as DeepML framework to achieve multi‐classification of BC using histopathological images. The DeepML framework has achieved distinguished performance (approx. 98% &amp; 89% average accuracy for 90–10% and 80–20% train‐test split, respectively) on a wide scale dataset, which intimate the quality of the proposed framework among existing approaches.
A Novel Approach to Classify Breast Cancer Tumors Using Deep Learning Approach and Resulting Most Accurate Magnification Factor
Studies in computational intelligence, 2020
On global transform preservation by region based interest points for image registration
There are many methods to find Interest Points (IPs) in images for image registration. However, t... more There are many methods to find Interest Points (IPs) in images for image registration. However, the underlying heuristics for finding them is different for each. Due to this, their behavior towards different image distortions is expected to vary. Through this paper, we attempt to investigate the truth of the following hypothesis - "Global Transform is better preserved by region based IPs as compared to Point based IPs". For this purpose, we make the use of "Speeded up Robust Features (SURF)" based IPs (which are point based) and "Maximally Stable Extremal Region (MSER)" based IPs (region based). Then by using "Random Sample Consensus (RANSAC)" on a standard stereo database (which is globally distorted afterwards), we validate the truth of the hypothesis we have made. The testing of this hypothesis is motivated from the fact that generally in medical images, global (usually affine) distortions are dominant. Local distortions tend to decrease registration accuracy if IPs are located at those sites. However, opposite is the case with temporally separated images (e.g. pictures of highway taken at an interval of 10 seconds, keeping camera fixed). They have dominant local distortion. Hence a proper choice of interest points for registering images is necessary.
Classification of brain MR images using Modified version of Simplified Pulse-Coupled Neural Network and Linear Programming Twin Support Vector Machines
The Journal of Supercomputing, Mar 25, 2022
Computer methods in biomechanics and biomedical engineering. Imaging & visualization, Feb 22, 2019
Robust segmentation of the brain magnetic resonance (MR) images is extremely important for diagno... more Robust segmentation of the brain magnetic resonance (MR) images is extremely important for diagnosing the tissues quantitatively. It is crucial to detect the changes caused by the growth of edema and tumor in healthy tissues for better medical treatment planning. In order to increase the image quality, skull stripping or brain extraction is an essential pre-processing step in neuroimaging before the segmentation process. Hybrid algorithm made up of K-means clustering, and Fuzzy C-Means clustering (KFCM) algorithm offers advantages in the aspect of accuracy on soft tissues of brain MR images. KFCM algorithm clusters the images into the cerebrospinal fluid, white matter, gray matter and abnormal region. The segmented abnormal region has some false positive pixels which can not be removed by low order image processing techniques. In this study, we present the use of Hierarchical Centroid Shape Descriptor (HCSD) on the already segmented abnormal region by the above said method. The HCSD selects the region of interest only, i.e. abnormal region. Our algorithm offers considerable improvement in segmentation accuracy validated by the truth map. The quantitative evaluation and validation of experiments were carried out on 20 high-grade glioma suffering patient and 10 T1-weighted anatomical models of healthy brains.
Segmentation of CA3 Hippocampal Region of Rat Brain Cells Images Based on Bio-inspired Clustering Technique
In the area of clustering, the most common issue of obtaining the optimum number value for the cl... more In the area of clustering, the most common issue of obtaining the optimum number value for the clusters is still an open challenge for different application areas. It is very hard to get the optimal number of clusters because of the lack of prior knowledge. This happens due to having various dimensions of data, clusters having a wide range of shape, size & density, and overlapping exists among groups. Many approaches have been proposed by various researchers which include bio-inspired techniques like genetic algorithm, particle swarm optimization, invasive weed optimization, cat swarm optimization, ant colony optimization, etc., for addressing these issues. Also, various combinations of the hybridization of these techniques have been practices by the researchers. The superiority of evolutionary techniques over the hard clustering techniques such as k-means clustering becomes popular in clustering area. Inspired by this, a novel rat brain cell segmentation approach is proposed using the latest bio-inspired clustering technique known as “Teacher-Learner Based Optimization”. In contrast to most of the well-known clustering techniques, TLBO doesn't require any parameter tuning and is less complex. The proposed approach is validated using NISSL stained rat brain cell dataset. In experimental evaluation performance, comparisons are made, based on quantitative results as well as qualitative results. The overall result analysis shows that the proposed approach is much more capable in segmenting the cells in comparison to the other well-known clustering techniques.
MR brain tumor detection employing Laplacian Eigen maps and kernel support vector machine
An innovative and robust image segmentation approach has been proposed for magnetic resonance (MR... more An innovative and robust image segmentation approach has been proposed for magnetic resonance (MR) brain tumor extraction. We have proposed a novel technique to classify a given MR brain image as benign or malignant. In order to extract the features from given MR brain tumor image, we have first employed wavelet transform which is then followed by Laplacian Eigen maps (LE) so as to curtail the dimensions of extracted features. These reduced features are now given to kernel support vector machine (K-SVM). Once we are done with classification, the next logical step remains image segmentation. We have the proposed algorithm with Gaussian Radial Basis (GRB) kernel owing to the fact that it achieves higher efficiency. Moreover, we have adopted the Leave-one-out cross validation (LOOCVCV) strategy so as to enhance generalization of K-SVM. Experimental findings reveal that our proposed algorithm outperformed the existing brain tumor extraction techniques in terms of computational and qualitative aspect. It could serve doctors to examine whether the tumors is benign or malignant.
Classification of breast tumors as benign and malignant using textural feature descriptor
In this paper we have presented an automated diagnosis of breast cell cancer using histopathologi... more In this paper we have presented an automated diagnosis of breast cell cancer using histopathological images on the basis of different textural descriptors. In the proposed technique, the images being preprocessed using extended adaptive-top-bottom transform (EAHE-TBhat) and segmented the nuclei regions from the non-nuclei regions using region growing segmentation. The nuclei regions are then used to extract features and provides texture descriptors using parameter free version of threshold adjacency statistics (PFTAS). The feature vector obtained are then classified as benign tumor feature and malignant tumor features using Rotation Forest (RF) classifier. The proposed technique compared with the other four combination of conventional texture techniques and classifiers. The experimental results and performance metrics values shows that the proposed technique is better than the other conventional techniques.
An automated computer-aided diagnosis system for classification of MR images using texture features and gbest-guided gravitational search algorithm
Biocybernetics and Biomedical Engineering, Apr 1, 2020
Abstract The segmentation and classification of brain magnetic resonance (MR) images are the cruc... more Abstract The segmentation and classification of brain magnetic resonance (MR) images are the crucial and challenging task for radiologists. The conventional methods for analyzing brain images are time-consuming and ineffective in decision-making. Thus, to overcome these limitations, this work proposes an automated and robust computer-aided diagnosis (CAD) system for accurate classification of normal and abnormal brain MR images. The proposed CAD system has the ability to assist the radiologists for diagnosis of brain MR images at an early stage of abnormality. Here, to improve the quality of images before their segmentation, contrast limited adaptive histogram equalization (CLAHE) is employed. The segmentation of the region of interest is obtained using the multilevel Otsu's thresholding algorithm. In addition, the proposed system selects the most significant and relevant features from the texture and multiresolution features. The multiresolution features are extracted using discrete wavelet transform (DWT), stationary wavelet transform (SWT), and fast discrete curvelet transform (FDCT). Moreover, the Tamura and local binary pattern (LBP) are used to extract the texture features from the images. These features are used to classify the brain MR images using feedforward neural network (FNN) classifier, where different meta-heuristic optimization algorithms, e.g., genetic algorithm (GA), particle swarm optimization (PSO), gravitational search algorithm (GSA), and gbest-guided gravitational search algorithm (GG-GSA) are employed for optimizing the weights and biases of FNN. The extensive experimental results on DS-195, DS-180, and three standard datasets show that the classification accuracy of GG-GSA based FNN classifier outperforms all mentioned meta-heuristic-based classifiers and several state-of-the-art methods.
Automated Semantic Segmentation of Chest X-ray images using Deep Learning Model
2021 IEEE Bombay Section Signature Conference (IBSSC), Nov 18, 2021
Chest Radiography proves to be a faster, cheaper, and less invasive diagnosis mode for respirator... more Chest Radiography proves to be a faster, cheaper, and less invasive diagnosis mode for respiratory diseases like pneumonia and viral infections like the coronavirus. The utilization of AI based strategies for programmed finding or imaging are pretty prevalent. In this work, a deep learning model is proposed for automatically segmenting chest x-ray images. The model comprises 20 fully convolutional layers that simplify images to precisely section the lung lobes from the x-ray images. The utilization of transposed convolution offers a lesser computational overhead than traditional methods. The proposed model achieves an accuracy of 97%, with an average Dice coefficient of 0.95 and an average Jaccard (IoU) score of 0.90. The proposed model is trained and tested on publicly available Montgomery County (MC) and Shenzen Hospital (SH) datasets. The segmentation ability of the proposed model can be used as input for predictive models, achieving better accuracy and faster convergence.
Segmentation of Brain MR Images Using Fuzzy-C Means and Markov Random Field Approaches
IPCV, 2010
Abstract: Segmentation of brain MR image has become more significant in research and medical appl... more Abstract: Segmentation of brain MR image has become more significant in research and medical applications. It is a process of extraction of various cortical tissues which is a key issue in neuroscience to detect early neural disorders. The aim of this study is to segment the brain MR image ...
Development of Plant-Leaf Disease Classification Model using Convolutional Neural Network
2022 IEEE 4th International Conference on Cybernetics, Cognition and Machine Learning Applications (ICCCMLA)
A Novel Computational Approach based on 3D Reconstruction and WEKA Tool to analyse the morphology of Golgi-Cox stained Rat Brain Cells
2019 4th International Conference on Information Systems and Computer Networks (ISCON)
Inspite the many significant researches has been done in the area of neurological field based on ... more Inspite the many significant researches has been done in the area of neurological field based on detection and extraction of morphology of the dendritic spines. Still the current researches and automated tools have various shortcomings against the detection and analysis of the dendritic spines. The detection and extraction of neuronal morphology of spines is especially challenging due to spines compounded with lower image resolution & contrast levels. To address this issue, a novel approach is proposed to detect and quantify the dendritic spines of the rat brain cells. The overall approach includes: (i) a novel 3D reconstruction algorithm to enhance the spines in 3D space, (ii) the mathematical description of proposed algorithm and (iii) trainable WEKA tool to classify the spines as stubby, thin & mushroom using random forest classifier and to calculate the number of each spine type. This approach enables the computation of volumteric representation of the raw spines in 3D space and morphological analysis of each spine type. We illustrate the proposed approach using Golgi-Cox stained dataset images. This dataset consists of two types of images i.e. EMF exposed images (with radiation) and SHAM exposed (without radiation) images of rat brain cells. The proposed approach achieved the accuracy of 98% (approx) in classifying the dendritic spines of rat brain cells. The proposed approach has successfully conveyed the degeneracy in the maturational shift of spine types for EMF exposure spine images as compared to SHAM exposure spine images during development in the rat primary visual cortex. Moreover, this approach is responsible for finding the defects in the morphology and size of the spines which leads to several neurological disorders.
Uploads
Papers by PROF MAHUA BHATTACHARYA