MTMohd Izzuddin Mohd Tamrin

Mohd Izzuddin Mohd Tamrin

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Mohd Izzuddin Mohd Tamrin has a total of 69 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Design of a Wearable EMG–IMU Gesture Recognition System for Real-Time Sign Language Translation in Assistive Communication Applications

    by Muhammad Nazmi Fazail, Muhammad Hamdi Mohd Najib, Ahmad Anwar Bin Zainuddin, Mohd Izzuddin Mohd Tamrin, Krishnan Subramaniam, Siti Fairuz Nurr Sadikan, Saidatul Izyanie Kamarudin

    Communication barriers have existed between the speech-and-hearing-impaired people and rest of the world due to the non-mainstream nature of sign language (SL). While there are proposed options to translate, most of them are costly and undeployable. The paper aims to create an affordable system to provide impaired users ability to communicate with normal people easily using text-to-speech (TTS) technology with the power of artificial intelligence (AI) and Internet of Things (IoT), powered by ESP32 to accurately predict SL. This is supported by dual channel electromyography (EMG) sensors and 6 axis gyrometer that is processed through filtering and feature extraction to give out high quality data for prediction. This system allows seamless communication enabled by MQTT and Wi-Fi connection from a wearable glove to collect required data that is used to train a Random Forest (RF) model. On a held-out validation set the system attained 92% accuracy; however, in on-field deployment this fell to about 53% (38–70% across the three gestures), so the system is presented as a feasibility prototype rather than a deployment-ready aid. As a starting point for this approach, future work will expand the dataset with more participants and gestures, complete a subject-independent evaluation, and refine the wearable device.

  • Open Access

    Machine learning or morphometric scaling? A systematic review of methodological confounds and the generalizability of sex classification in neuroimaging

    by Abdul Halim Sapuan, Iqbal Jamaludin, Zafri Azran Abdul Majid, Mohd Izzuddin Mohd Tamrin, Mohd Zulfaezal Che Azemin, Sherzod Turaev

    Background: This systematic review critically evaluates whether machine learning (ML) identifies biologically meaningful sex-related brain architecture or merely exploits methodological artifacts and allometric scaling. While ML models achieve high classification accuracies, it remains unclear if these reflect stable, mechanistically informative dimorphism or are driven by confounds such as total intracranial volume (TIV) and site-specific noise. We examine how imaging modalities, algorithms, and population strata influence both classification outcomes and biological interpretability. Methods: Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, we searched Web of Science, PubMed, and Scopus through January 2024. Included studies [healthy humans, 3T magnetic resonance imaging (MRI), ML-based sex classification] were assessed for risk of bias, focusing on data leakage, validation strategies, and confound management. Results: Thirty-five studies (n > 110,000) were included. While reported accuracies reached 98.06% for T1-weighted MRI, 96.0% for diffusion MRI (dMRI), and 94.72% for functional MRI (fMRI), performance was highly dependent on population characterization and age. Deep learning consistently outperformed traditional ML (TML) but showed high sensitivity to methodological artifacts. Notably, studies failing to correct for TIV reported potentially inflated accuracies, suggesting that many models identify physical scale rather than intrinsic neuroanatomical dimorphism. Discussion: High classification accuracies are often bolstered by methodological confounds and a lack of cross-site validation. There is a significant discrepancy between ML-driven predictive power and biological inference validity. Current pipelines do not yet allow for robust, generalizable inference about brain sex. To move beyond statistical separation toward mechanistic understanding, the field must prioritize TIV-corrected benchmarks and diverse non-WEIRD (Western, Educated, Industrialized, Rich, Democratic) datasets. We conclude that while ML is a powerful pattern detector, its results must be interpreted with caution regarding biological dimorphism.

  • Lightweight Texture-Based Classification of Huffaz Status from Structural MRI Using GLCM Correlation and Brodmann-Area VOIs

    by Mohd Izzuddin Mohd Tamrin, Iqbal Jamaludin, Abdul Halim Sapuan, Mohd Zulfaezal Che Azemin, Asadullah Shah

    Previous neuroimaging studies suggest that intensive Quran memorization may be associated with structural brain differences, but simple and interpretable MRI-based biomarkers remain limited. This study investigated whether a single interpretable radiomics feature, gray-level co-occurrence matrix (GLCM) Correlation, extracted from predefined Brodmann Area (BA) volumes of interest (VOIs), is associated with Huffaz status.A cross-sectional case-control design was used involving 47 participants (23 Huffaz, 24 non-Huffaz). Structural MRI scans were pre-processed using a standard SPM pipeline to generate modulated, warped, and smoothed tissue-class images (smwc*.nii). Inferential analyses were restricted to five literature-driven candidate regions (BA22, BA24, BA32, BA40, and BA46) to reduce multiplicity. Best-subset logistic regression was performed with sex forced into all models and Bayesian Information Criterion (BIC) used for model selection. Bootstrap resampling was applied to assess feature-selection stability.BA46 GLCM Correlation emerged as the lowest-BIC predictor set. Bootstrap resampling showed higher selection stability for BA46 (approximately 0.68 selection frequency) than for the other candidate regions (approximately 0.13–0.15), supporting a reproducible single-region signal within the tested set. These findings suggest that BA46 texture organization, as captured by GLCM Correlation, may differentiate Huffaz from non-Huffaz after adjustment for sex.The modest sample size and lack of external validation limit generalizability. Nevertheless, the findings support the potential of an interpretable single-feature radiomics biomarker and motivate further validation in larger independent cohorts, as well as robustness analyses across preprocessing and discretization settings.


  • Open Access

    Exploring Sex-Specific Brain Complexity in Quran Memorizers: A Fractal Dimension Approach

    by Abdul Halim Sapuan, Zafri Azran Abdul Majid, Mohd Izzuddin Mohd Tamrin, Mohd Zulfaezal Che Azemin

    This study explores sex-specific variations in brain complexity among Quran memorizers (Huffaz) and non-Huffaz using fractal dimension (FD) analysis, a novel method for quantifying structural intricacies of the brain. By employing both box-counting and Fourier FD techniques, we analysed T1-weighted MRI scans from 47 healthy young adults, investigating whether intensive cognitive practices, such as Quran memorization, impact brain structure differently in males and females. The findings reveal significant sex-related differences in FD across multiple brain regions, with the declive, lingual gyrus, and medial frontal gyrus emerging as key areas of interest. These differences persisted even after controlling for Huffaz status, suggesting that both inherent sexual dimorphism and cognitive engagement contribute to brain complexity. The study underscores the potential of FD as a tool for understanding the nuanced interplay between sex, cognition, and brain structure, offering insights that may inform future research on brain morphology and sex-based neurological differences. VIDEO EXPLAINER AUDIO PODCAST

  • Open Access

    Optimized Retinal Vessel Segmentation Using IS-Net and High-Resolution Dataset

    by Mohd Zulfaezal Che Azemin, Mohd Izzuddin Mohd Tamrin

    Segmentation of the retinal vessels is extremely useful and very important in the diagnosis and management of various diseases associated with the eye, including diabetic retinopathy and glaucoma. The work has presented an improved methodology using an IS-Net model trained on the high-resolution FIVES dataset, including 800 annotated images of the retina. This paper therefore resolves the proposed approach by pre-processing, which consists of normalizing and performing horizontal flipping, followed by enhancement using IS-Net and histogram-based thresholding criteria for vessel structure binarization. The IS-Net architecture is designed with multi-scale RSU blocks to capture both fine and broad vessel details comprehensively for segmentation. Results have shown that IS-Net achieves a good balance in recall and specificity, with the F1 score high enough to outperform other models in terms of specificity by reducing false positives. These findings underlined the effectiveness of IS-Net for clinical applications and emphasized the value of high-resolution data for refinement in the performance of segmentation.


  • Open Access

    Smart Attendance in Classroom (CObot): IoT and Facial Recognition for Educational and Entrepreneurial Impact

    by Ahmad Anwar Zainuddin, Rizal Mohd Nor, Dini Handayani, Mohd. Izzuddin Mohd. Tamrin, Krishnan Subramaniam, Siti Fairuz Nurr Sadikan

    Current attendance methods, though simple, are prone to manipulation and can be time consuming for both educators and students. For instance, manual systems and QR code based methods allow students to register attendance on behalf of others due to the lack of unique identification. While calling names individually improves security, it disrupts the learning process by consuming significant time. This study addresses these issues by developing an autonomous robot, CObot, equipped with a facial recognition system powered by a Raspberry Pi microcontroller. CObot navigates classrooms autonomously, avoiding obstacles, and efficiently records attendance without requiring movement from students or educators. The use of facial recognition ensures that only registered individuals can mark attendance, creating a secure and tamper-proof system. Additionally, the integration of Internet of Things (IoT) technology enables real-time data transfer to Google Sheets, simplifying record-keeping and reducing educators administrative workload. A 3D-printed, customizable car structure enhances the robot design, while the Raspberry Pi 5 was selected over alternatives like the ESP32-S3 for its superior processing power and faster data transfer speeds, ensuring smoother operations. In testing with 60 participants, the Raspberry Pi 5 demonstrated a 99% accuracy rate in facial recognition, outperforming the ESP32-S3 90% accuracy. By saving time, improving security, and reducing manual effort, CObot enhances the classroom environment, benefiting both students and educators. While the improvement in attendance systems may appear incremental, CObot represents a meaningful step toward fostering a more efficient and effective learning environment.


  • Open Access

    Assessing the Efficacy of StyleGAN 3 in Generating Realistic Medical Images with Limited Data Availability

    by Mohd Zulfaezal Che Azemin, Mohd Izzuddin Mohd Tamrin, Mohd Radzi Hilmi, Khairidzan Mohd Kamal

    In this study, we leveraged StyleGAN 3 to synthesize high-fidelity images of pterygium, achieving significant strides in image realism as evidenced by low Fréchet Inception Distance (FID) scores. Our results demonstrate that StyleGAN 3 can intricately capture the textural nuances and vascular patterns distinctive to pterygium, with color tones and variations that closely mirror clinical photography. The generated images exhibit high equivariance to transformations, retaining their realism under various manipulations. Clinician reviews, expressed through confusion matrices, validated the authenticity of the synthetic images, although variations in individual assessments highlighted the challenges in differentiating between generated and real images. Ultimately, our findings confirm the efficacy of StyleGAN 3 in producing synthetic medical images that could potentially expand datasets for medical research and training, while also underscoring the necessity for diversity in training data and model tuning to achieve optimal realism.


  • Open Access

    Modelling Intraocular Lens Design Based on Image Analysis

    by Wan Anis Mardhiah Wan Mohd Fadlah, Mohd Zulfaezal Che Azemin, Mohd Izzuddin Mohd Tamrin, Mohamed Jalaldeen Mohamed Razi

    Intraocular lenses (IOLs) are commonly used to treat ocular diseases such as cataracts. The International Organization for Standardization (ISO 11979) recommends examining IOLs using real model eyes with the lenses in situ, although this process can also be simulated in silico with an accurate IOL haptic design. One way to create this design is by using software such as Abaqus, which generates a finite element mesh configuration with a linear element set up and directly tests it in the model eye using the same program. While this process is like those reported in other studies and includes most of the software necessary for generating the result automatically, there is an alternative method for generating the IOL haptic design — microscopic image analysis. The purpose of this study was to evaluate the image quality of IOL using microscopic analysis. Several IOL images were captured using a microscope and the converted into vector images and the steps were repeated after two weeks. The structural changes in the images were measured using the Structural Similarity Index Measure (SSIM). All the tested IOLs have SSIM values greater than 0.7. The greatest value was extracted from the simplest IOL shape design. Our result suggests that image analysis for IOL modelling is a reliable method and best for simple IOL shape design.

  • Open Access

    Volume of Interest-Based Fractal Analysis of Huffaz’s Brain

    by Iqbal Jamaludin, Mohd Zulfaezal Che Azemin, Mohd Izzuddin Mohd Tamrin, Abdul Halim Sapuan

    The robust process in memorising the Quran is expected to cause neuroplasticity changes in the brain. To date, the analysis of neuroplasticity is limited in binary images because greyscale analysis requires the usage of more robust processing techniques. This research work aims to explore and characterise the complexity of textual memorisation brain structures using fractal analysis between huffaz and non-huffaz applying global box-counting, global Fourier fractal dimension (FFD), and volume of interest (VOI)-based analysis. The study recruited 47 participants from IIUM Kuantan Campus. The huffaz group had their 18 months of systematic memorisation training. The brain images were acquired by using MRI. Global box-counting and FFD analysis were conducted on the brain. Magnetic resonance imaging (MRI) found no significant statistical difference between brains of huffaz and non-huffaz. VOI-based analysis found nine significant areas: two for box-counting analysis (angular gyrus and medial temporal gyrus), six for FFD analysis (BA20, BA30, anterior cingulate, fusiform gyrus, inferior temporal gyrus, and frontal lobe), and only a single area (BA33) showed significant volume differences between huffaz and non-huffaz. The results have highlighted the sensitivity of VOI-based analysis because of its local nature, as compared to the global analysis by box-counting and FFD.


  • Open Access

    Postsurgery Classification of Best-Corrected Visual Acuity Changes Based on Pterygium Characteristics Using the Machine Learning Technique

    by Fatin Nabihah Jais, Mohd Zulfaezal Che Azemin, Mohd Radzi Hilmi, Mohd Izzuddin Mohd Tamrin, Khairidzan Mohd Kamal

    Introduction. Early detection of visual symptoms in pterygium patients is crucial as the progression of the disease can cause visual disruption and contribute to visual impairment. Best-corrected visual acuity (BCVA) and corneal astigmatism influence the degree of visual impairment due to direct invasion of fibrovascular tissue into the cornea. However, there were different characteristics of pterygium used to evaluate the severity of visual impairment, including fleshiness, size, length, and redness. The innovation of machine learning technology in visual science may contribute to developing a highly accurate predictive analytics model of BCVA outcomes in postsurgery pterygium patients. Aim. To produce an accurate model of BCVA changes of postpterygium surgery according to its morphological characteristics by using the machine learning technique. Methodology. A retrospective of the secondary dataset of 93 samples of pterygium patients with different pterygium attributes was used and imported into four different machine learning algorithms in RapidMiner software to predict the improvement of BCVA after pterygium surgery. Results. The performance of four machine learning techniques were evaluated, and it showed the support vector machine (SVM) model had the highest average accuracy (94.44±5.86), specificity (100), and sensitivity (92.14±8.33). Conclusion. Machine learning algorithms can produce a highly accurate postsurgery classification model of BCVA changes using pterygium characteristics.