Muhammad Kashif

RESEARCH OUTPUT

Selected publications

1. NeuroFusion: A Unified Framework for Generalized Visual Stimulus Decoding from fMRI Across Datasets and Subjects.Neuroinformatics · 2026
   Recent advancements in neural decoding have shown promising results in reconstructing visual experiences from brain activity. However, existing approaches focus primarily on decoding within a single dataset or subject, which limits generalization across various sources of neuroimaging. In this work, we propose a novel framework for the decoding of visual stimuli between subjects and between data sets, integrating neural recordings from multiple publicly available fMRI datasets. To address inherent intersubject and interdataset variability, we introduce a contrastive learning-based alignment strategy using image embeddings from a pre-trained IP-Adapter model. Our approach learns a shared latent space by aligning subject-specific neural representations with image features, enabling generalized decoding across both subjects and datasets. In addition, we propose a simple yet effective data augmentation method using ridge regression. This method synthesizes realistic fMRI-like signals from novel images by predicting voxel activity and injecting learned noise distributions, thus enhancing training diversity and model robustness. To the best of our knowledge, while several recent studies have explored cross-subject decoding, we extend recent cross-subject decoding efforts by training a single unified framework jointly across multiple public fMRI datasets and subjects, enabling cross-dataset transfer in addition to cross-subject generalization. We distinguish this multi-dataset unified training setting, where each dataset contributes training data, from a stricter leave-one-dataset-out transfer setting in which the target dataset is excluded from source pretraining and used only for lightweight alignment-layer adaptation. Empirically, our unified model achieves strong semantic reconstruction across datasets (e.g., up to 94.8% CLIP similarity on NSD (AUG) and 0.403 SSIM on BOLD5000 after lightweight finetuning), demonstrating robust cross-subject and cross-dataset transfer.

   Read abstract
   PublisherPubMedScopus
2. Multilevel Feedback Queue: Efficient Scheduling and Implementation by Using Dynamic QuantumAIP Conference Proceedings · 2023

   Abstract not available.

   Publisher
3. Odor clustering using a gas sensor array system of chicken meat based on temperature variations and storage timeSensing and Bio-Sensing Research · 2022

   Abstract not available.

   Publisher
4. Gas Array Sensors based on Electronic Nose for Detection of Tuna (Euthynnus Affinis) Contaminated by Pseudomonas AeruginosaJournal of Medical Signals and Sensors · 2022
   BACKGROUND: Fish is a food ingredient that is consumed throughout the world. When fishes die, their freshness begins to decrease. The freshness of the fish can be determined by the aroma it produces. The purpose of this study is to monitor the odor of fish using a collection of gas sensors that can detect distinct odors. METHODS: The sensor was tested with three kinds of samples, namely Pseudomonas aeruginosa, tuna, and tuna that was contaminated with P. aeruginosa bacteria. During the process of collecting sensor data, all samples were placed in a vacuum so that the gas or aroma produced was not contaminated with other aromas. Eight sensors were used which were designed and implemented in an electronic nose (E-nose) device that can withstand aroma. The data collection process was carried out for 48 h, with an interval of 6 h for each data collection. Data processing was performed by using the principal component analysis and support vector machine (SVM) methods to obtain a plot score visualization and classification and to determine the aroma pattern of the fish. RESULTS: The results of this study indicate that the E-nose system is able to smell fish based on the hour with 95% of the cumulative variance of the main component in the classification test between fresh tuna and tuna fish contaminated with P. aeruginosa. CONCLUSION: The SVM classifier was able to classify the healthy and unhealthy fish with an accuracy of 99%. The sensors that provided the highest response are the TGS 825 and TGS 826 sensors.

   Read abstract
   Publisher
5. Development of a multi-epitope spike glycoprotein vaccine to combat SARS-CoV-2 using the bioinformatics approachJournal of Pharmacy and Pharmacognosy Research · 2022
   Context: The current COVID-19 pandemic has significantly impacted health and socio-economic status worldwide. The only way to combat this situation is to develop an effective vaccine and immunize people around the globe. Aims: To construct a multi-epitope spike glycoprotein-based vaccine from the SARS-CoV-2 Surabaya isolate using a bioinformatics approach. Methods: The spike protein was submitted to IEDB, VaxiJen, AllerTOP, and ToxinPred webservers to predict antigenic, non-allergic, non-toxic, B- and T-cell epitopes. To develop a multi-epitope vaccine, an adjuvant cholera toxin B subunit was linked to B-cell and B-cell with T-cell through EAAAK and GPGPG linkers, respectively. The designed vaccine 3D structure development, refinement, and validation were done through PHYRE2, Galaxy Refine, and RAMPAGE webservers. Moreover, the Cluspro-2.0 webserver was used for the molecular docking of the vaccine designed with TLR3. The vaccine+TLR3 complex was docked with Surfactant protein A as a control to validate the docking results. Finally, immune-simulation and in silico cloning of the vaccine were carried out by C-ImmSim webserver and SnapGene software, respectively. Results: A multi-epitopic vaccine containing B and T-cell was developed using 392 amino acids with a molecular weight of 40825.59 Da. The docking and immunogenicity results of the vaccine met all established parameters for constructing a quality vaccine. Furthermore, the optimized sequence of the vaccine was successfully cloned in expression vector pET 28 a (+) that yielded a colon of 2724 bp. Conclusions: The vaccine’s immunogenicity demonstrates its effectiveness against SARS-CoV-2 infection. Further confirmatory testing may therefore be performed as soon as possible in the public interest.

   Read abstract
   Publisher
6. An improved Kalman Filter in Photoplethysmography DC Component Denoising for cardiorespiratory analysisRawal Medical Journal · 2022

   Abstract not available.

   Publisher
7. Gas sensor array to classify the chicken meat with E. coli contaminant by using random forest and support vector machineBiosensors and Bioelectronics: X · 2021

   Abstract not available.

   Publisher