Abstract : Information is sent through public networks, and the security of that information has been a top priority. Along with many other common cryptographic approaches, Visual Cryptographic (VC) techniques have also been applied to information and data security. As VC divides the original image into share photos in sequential sequence, a hidden secret image is revealed when the shares are stacked on top of one another. The secure VC technique divides a secret file or image into sharing images to encrypt it. A progression model for large-scale systems is cloud computing. From image recovery through image processing, storage, and retrieval with the progression of data in the medical sector and healthcare systems into the cloud, security for medical image transmission has been an ongoing computational concern. However, the use of cloud computing is restricted to situations where security is guaranteed. This study proposes Modified Leading Diagonal Sorting with Probabilistic Visual Cryptography (MLDS-PVC). The MIAS dataset�s breast cancer mammography images are used in this study�s medical images. The proposed model might allow for the cloud-based storage and transmission of medical image data between hospitals, diagnostic facilities, and other healthcare facilities.

Keywords: Visual Cryptography; Cloud Computing; Modified Leading Diagonal Sorting; Probabilistic Visual Cryptography; MIAS dataset.

Abstract : Automatic speech recognition has become integral to many applications, specifically HMI. Several AI models, supervised/unsupervised, are available in various platforms � MATLAB, LabVIEW, and Python with varying performance metrics. This paper presents a generic AI Model with the following features. The model is specifically built for �digit word utterance detection.� The dataset contains an isolated set of digit utterances from 1 to 10 across speakers of different age groups. The dataset is prepared using established pre-processing steps, trimming each utterance to remove silence in the initial part and end of the utterance using Audacity software and noise removal. Also, single channel selection followed by sampling the speech signal at fs = 48kHz. Statistical variance analysis on the MFCC matrix for each digit utterance is carried out to obtain the feature set. KNN-based and MLP-based AI models are developed for this feature set (resulting from statistical methods). The performance of the developed AI models is analyzed. To reduce the computational complexity of the feature sets, dimensionality reduction has been applied to the extracted MFCC features using the SVD technique. This reduced number of principal components forms a new feature for the same utterances. KNN-based and MLP-based AI models are developed for this new feature set (resulting from the SVD method). The performance analysis is carried out for these models also. Results show that SVD with MLP performs better in classifying the uttered digit.

Keywords: MFCC, PCA, SVD, MLP, KNN, LPC, HMM

Abstract : The frequency of defects in the neurological system and soft tissues is gradually rising, and magnetic resonance imaging (MRI) is the most effective evaluation technique. The brain tumor MR image segmentation functions include qualitative analysis of infected and normal tissues and image reconstruction of the afflicted (diseased tissues). The shape, size, and position of lesion tissues, suitable diagnostic methods, and illness diagnosis affect how accurately an image can be segmented from a medical standpoint. The results of this study rely on the Multi-Perspective Scaling Convolutional Neural Networks (MPS-CNN) model for more precisely and successfully segmenting brain tumors. The suggested CNN model is provided with multi-scale inputs to avoid the need to choose the suitable input scale based on the tumor size, neighbourhood tumor analysis based on scaled images, and adoption towards diverse tumor sizes. Therefore, the segmentation accuracy can be improved using the input multi-scale brain tumor images. Additionally, the multi-scaling segmentation process speeds up, ensuring real-time segmentation. The brain images in the MRI can be successfully segmented using this scaling approach, which improves generalization. It is used to forecast the MRI brain lesion tissue. In the MATLAB environment, the simulation is run. Comparisons are made between the projected MPS-CNN and popular techniques like CNN, FCN, U-Net, SegNet, Deep V3, and Deep FCN, and in contrast to other methods, the MPS-CNN demonstrates a better trade-off.

Keywords: Magnetic resonance imaging (MRI); segmentation, multi-scaling; convolutional neural networks; brain tumor.

Abstract : Disease prediction has emerged as a widespread issue in public health. The most common is cardiac disease, which is caused by unhealthy food habits, poor lifestyle choices, and a lack of health awareness. As a result, diagnosis becomes more difficult, creating an active study topic. Conversely, the quantity of knowledge is thought to deal with making the right choices. The necessity for medical parameters and analysis results in accurate risk analysis prediction. This study examines several well-exposed factors and algorithms in this exploration. Here, we give the performance metrics from a study we conducted using accepted methodologies. The research discusses how well popular algorithms for detecting cardiac attacks may be predicted using reliable criteria.

Keywords: Disease Prediction; Machine Learning Algorithms; Heart Disease; Poor Lifestyle; Health Awareness.

Abstract : Natural Language Processing (NLP) is crucial in the COVID-19 pandemic for identifying the illness, stabilizing intensive care, finding cures, and halting the spread of diseases. In order to help prevent outbreaks during the early phases of coronavirus illness, it energizes chat programs. A multilingual conversation system, chatbots, and deep learning language models have all found success thanks to the advancement of NLP technologies, which have achieved new heights in terms of utility. 13 different languages are supported globally. Health Map and Cobweb platforms, which use NLP-powered AI, track patient requests and carry out event detections. This chapter examines the role of NLP, its technology, difficulties, and potential future applications for crisis management and simpler EHRs in the healthcare sector.

Keywords: Blockchain; Machine Interpretable; Phenotyping; Voice recognition; Cognitive Technologies; Entity Extraction.