Cochin College of Engineering and Technology

Brain tumor classification is one of the most challenging tasks in clinical diagnosis and treatment within medical image analysis. Errors during the diagnostic process can significantly impact patient outcomes, making accuracy paramount. However, many existing techniques overlook critical features that hold substantial relevance for classification, focusing instead on extracting deep but less discriminative features. Deep learning-based categorization of brain tumors using magnetic resonance imaging (MRI) has emerged as a vital research area. This paper introduces an automated deep learning model and an optimal information fusion framework for brain tumor classification from MRI images. The dataset used in this study was imbalanced, posing a significant challenge for training deep learning models, as it biases classifier performance toward the majority class. To address this, we designed a sparse autoencoder network to generate synthetic images, effectively resolving class imbalance. Subsequently, two pretrained neural networks were fine-tuned, with hyperparameters optimized via Bayesian optimization for enhanced training. Deep features were then extracted from the global average pooling layer, yielding a refined set with minimal irrelevant information. To further optimize feature selection, we propose a novel Quantum-Inspired Adaptive Feature Selector (QIAFS) algorithm, which identifies the most discriminative features from both networks and fuses them using a serial-based approach. The fused feature set is then classified using neural network classifiers. Our framework was evaluated on an augmented Figshare dataset, achieving an accuracy of 98.9%, a sensitivity of 99.83%, a false negative rate of 17%, and a precision of 99.73% in average. Comparative and ablation studies demonstrate the significant improvement in performance over existing methods.

The rising dependence on fossil fuels, depleting renewable resources, and increasing oil costs necessitate alternative energy sources. Biofuels, such as pumpkin seed biodiesel, offer environmentally friendly solutions with lower greenhouse gas emissions. This is the first study to integrate pumpkin seed oil–based biodiesel blended with cerium oxide (CeO 2 ) nanoparticles and machine learning (ML) models for optimizing diesel engine performance and emission characteristics. The study uses response surface methodology (RSM) and XGBoost ML to maximize engine performance and predict emissions of carbon monoxide (CO), hydrocarbon (HC), nitrogen oxide (NO x ), and smoke opacity. The optimal blend, achieving a brake thermal efficiency (BTE) of 24.99% with minimal emissions, was 18.32% biodiesel, 63.84% engine load (operating at 75% of maximum capacity), and 48.55 ppm CeO 2 . This study demonstrates the effectiveness of combining RSM and ML, providing new insights into the sustainable optimization of biodiesel blends for compression ignition engines.

This research synthesis the aluminium hybrid composite made with 5-15% of nano SiC and 5% of ZrO2 reinforcement via ultrasonic vibration-assisted stir casting techniques. Influences of hybrid reinforcements on tribological behaviour of composites are evaluated by different sliding spans (1000 to 3000m), load (10-20N), and sliding speed (1.5 to 4.5m/s). The hybrid Al5083/5wt% ZrO2/15wt% SiC composite is noted to have a lower wear rate ranging from 0.0008-0.0016 mm3/Nm and enhanced friction coefficient of 12-0.25 at 4.5m/sec for 3000m under 30N applied load. Moreover, the increased sliding speed and distance lead to an increase in the wear rate and limits the friction coefficient. The tribological behaviour of the hybrid composite Al5083/5wt% ZrO2/15wt% SiC was observed to exhibit diverse surface characteristics due to the occurrence of delamination, wear debris, tribo-chemical wear regimes, and two-body and three-body abrasion wear. Furthermore, the presence of pits and oxide layers was also noticed.

Autism Spectrum Disorder (ASD) is a brain disease that mostly affects communication ability, object identification, cognitive capacity, interpersonal skills, and speech comprehension. Its primary origin is genetics, and intervention and diagnosis in an early stage can alleviate the requirement for expensive medical approaches and lengthy tests for ASD patients. Neuroimaging methods can be used to distinguish the composite biomarkers within the ASD based on functional connectivity abnormalities. However, the identification of ASD adopts symptom-based conditions through medical examination. Traditional automated techniques based on extensive aggregated datasets are likely to attain undependable diagnostic classification. Hence, it is essential to establish an efficient ASD classifier system with a deep learning approach to overcome the limitations of the classical models. The innovation of the developed work lies in the implementation of a novel deep learning approach named ViT-ARDNet-LSTM for classifying ASD utilizing MRI images. This framework integrates the merits of ViT, adaptive residual densenets, and LSTM models for efficiently classifying ASD. The developed model rectifies some issues such as the requirement for effective and accurate ASD diagnosis, the problems of conventional models, and the complexities of validating the complex MRI images. Especially, the suggested work resolves the problems of variability in the MRI images, the requirement of robust feature extraction, and the importance of optimizing the model parameters. By offering an effective and novel solution for ASD classification, the suggested work has the potential to improve the diagnosis accuracy, minimize the diagnosis time, and improve patient care. In the developed work, at first, significant MRI images are accumulated from benchmark resources and it is offered as input to the preprocessing stage. In this phase, Contrast Limited Adaptive Histogram Equalization (CLAHE) and bilateral filtering mechanisms are introduced to pre-process the gathered MRI images. Next, the pre-processed images are offered to the ASD classification stage. In this phase, an implemented mechanism named Vision Transformer-based Adaptive Residual Densenet with Long Short Term Memory layer (ViT-ARDNetLSTM) is utilized to classify the ASD. Moreover, the parameters in ViT-ARDNet-LSTM are optimized using the Modified Zebra Optimization Algorithm (MZOA) for enhancing the functionality of classification. Lastly, the experimental validations are carried out for the developed work. The experimental results displayed that the suggested model attained 94% accuracy, precision, specificity, and sensitivity values when considering the sigmoid activation function. Also, the developed model achieved 5% FPR values. These results elucidate that the designed ASD classification framework outperforms the conventional models and improves timely diagnosis.

ABSTRACT Ion‐sensitive field‐effect transistors (ISFETs) represent a promising technology for the precise detection of ions, particularly within diagnostic and clinical biomedical contexts. This paper details the design and implementation of an ISFET‐based device tailored specifically for efficient and reliable ion detection in biomedical settings. By meticulously selecting materials, configuring sensors, and integrating specialized readout circuitry, the device exhibits exceptional sensitivity and specificity in detecting target ions pertinent to various biomedical assays. The design process involves thorough optimization for sensitivity, durability, and compatibility with biological samples. A comprehensive review of ISFET gate materials is provided, alongside an overview of typical gate structures and signal readout methods for ISFET sensing systems. Additionally, diverse biosensing applications including ions, deoxyribonucleic acid, proteins, and microbes are explored. Rigorous calibration and testing procedures ensure precise and reproducible measurements. The resulting device holds substantial promise for applications such as pH monitoring, ion analysis, and biomarker detection in clinical diagnostics, offering a versatile platform for advancing biomedical research and healthcare practices. In the experimental investigation of pH sensor performance, five distinct measurements of current sensitivity were recorded. The obtained values were 0.025, 0.028, 0.027, 0.026, and 0.030 mA/pH, respectively. These measurements serve as crucial indicators of the sensor's responsiveness to changes in pH levels, providing insights into its effectiveness in detecting subtle variations in acidity or alkalinity. Such precise data are essential for evaluating and optimizing the sensor's design and functionality for various practical applications in analytical chemistry, environmental monitoring, and biomedical diagnostics.