Sign languages are one of those mediums for hearing-impaired people. These languages transmit meaning by visual-manual treatment, or more simply, hand movement. Currently, there are only 95 sign language interpreters registered with the Malaysian Federation of the Deaf as of 2020, compared to 40,389 hearing-impaired individuals with disabilities registered with the welfare department which is a problem. Therefore, with the use of deep-learning technology, this paper proposes to alleviate the scarcity of Malaysian Sign Language interpreters for the benefit of hearing-impaired persons. The paper aims to test and report a sequenced 3D keypoint hand pose estimation model for Malaysian Sign Language Recognition and evaluate the implementation of action model in decoding basic poses of Malaysian Sign Language. According to the findings, the detecting of 3D keypoints and incorporating into LSTM models using deep learning machine learning platform and framework like TensorFlow and MediaPipe enables the detection of Malaysian sign language 3D hand posture estimation. The results demonstrated that 3D hand posture estimation may be utilised to estimate sign language in real time, providing for a better interpretation approach for the deaf community.

No abstract

Automatic segmentation solution is the process of detecting and extracting information to simplify the representation of Cardiac Magnetic Resonance images (CMRI) of Left Ventricle (LV) contour. This segmented information, using CMR images, helps to reduce the segmentation error between expert and automatic segmented contours. The error represents missing region values calculated in percentages after segmenting a cardiac LV contour. This review paper will discuss the major three segmentation approaches, namely manual approach, semi-automatic, and fully automatic, along with the segmentation models, namely image-based models, region-based models, edge-based models, deformable-based models, active shape-based models (ASM), active contour-based models (ACM), level set-based models (LSM), and Variational LSM (VLSM). The review deeply explains the performance of segmentation models using different techniques. Furthermore, the review compares 122 studies on segmentation model approaches, i.e., 16 from 2004 to 2010, 40 from 2011 to 2016, and 63 from 2017 to 2021, and 3 other related studies were conducted LV contour segmentation, cardiac function, area-at-risk (AAR) identification, scar tissue classification, oedema tissue classification, and identification via presence, size, and location. Given the large number of articles on CMR-LV images that have been published, this review conducted a critical analysis and found a gap for researchers in the areas of LV localization, LV contour segmentation, cardiac function, and oedoema tissue classification and segmentation. Regarding critical analysis, this paper summrised a research gap and made useful suggestions for new CMR-LV researchers. Although a timely reviewed study can lead to cardiac segmentation challenges, which will be discussed in each review section.

An optimized approach aiming to improve classification accuracy of wrist movements via electromyography (EMG) signals is presented here. EMG signals of the different types of wrist movements are obtained from the NINAPRO database. Useful features are extracted from the EMG signals via the waveform length method. The developed optimized classification system contains two main modules, known here as (i) optimized neural network module and (ii) movement prediction module. The optimized neural network module is made up of multiple 2-class neural networks. During Stage 1 Classification, a group of neural network (named NNG_S1) is formed after analyzing the sensitivity computed from the training outcomes of each neural network. A new group of neural network (named NNG_S2) is later formed in Stage 2 Classification after initial elimination via Stage 1 Classification. Further analysis is performed via the movement prediction module to predict the final outcome of the classification. The overall average classification accuracy achieved via the optimized classification system is 8.3% higher than the conventional neural network. The results validate that the optimized classification system performs better than the conventional neural network, providing more accurate signals for manipulating of exoskeleton for rehabilitation purposes.

No abstract