Photosensitizer-Loaded PVA/CMC Nanofibers to Combat Antibiotic-Resistant Bacteria
Student: Jarin Tasnim Maisha
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
The climbing incidence of antibiotic-resistant bacteria necessitates the advancement of innovative strategies to combat localized bacterial infections. Antibacterial photodynamic therapy (APDT) emerges as a promising alternative, utilizing light-activated photosensitizers (PS) to target pathogens without inducing resistance. This study investigates the potential of electrospun polyvinyl alcohol (PVA) and sodium carboxymethyl cellulose (CMC) nanofibers having different degrees of substitution (DS) for delivering photosensitizers in APDT. This study addresses the critical yet unexplored question of how DS influences the electrospinnability characteristics, and morphology of PVA/CMC nanofibers. Additionally, the efficacy of cationic (methylene blue, toluidine blue) as well as anionic (rose bengal) photosensitizers incorporated into the nanofibers via one-step electrospinning was compared. The developed nanofibers were characterized using SEM, TGA, tensile testing, swelling, and weight loss tests. It was found that CMC of DS 1.2 produced better microarchitecture, swelling properties and improved hydrolytic activity, while DS 0.7 produced better tensile modulus due to enhanced crosslinking. In vitro dye release study showed burst release of PS and showed saturation within 4 hours. Antibacterial efficacy of the PS with 2000 lumen light exposure was analyzed against gram (–) Escherichia coli bacteria as well as gram (+) Staphylococcus aureus bacteria. Cationic PS showed promising results and better antibacterial efficacy in combating both the bacterial strain after 30 minutes of light exposure. This study effectively demonstrates the potential of photosensitizer loaded electrospun nanofibers as biomaterials with antimicrobial properties, offering a promising approach for antibiotic-free infection control.
BurnFlow: A Comprehensive Analysis Of Burn Fluid Management Using An Interactive Mobile App
Student: A S M Anas Ferdous, Sk Shamiur Rahman
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
Burns, caused by flames, ultraviolet radiation, hot liquids, electricity, lightning, or specific chemicals, require immediate first-aid intervention. Partial and full-thickness burns, in particular, necessitate urgent medical attention. Full-thickness burns often necessitate skin graft surgery. In- terestingly, due to compromised nerve endings, simple analgesics (e.g., ibuprofen, acetaminophen) and opioids (e.g., morphine) are typically ineffective for pain management in third-degree burns. Accurate assessment of the burn area is crucial for optimal patient care. The modified Lund- Browder chart remains a widely used and reliable tool for estimating Total Body Surface Area (TBSA) in burn patients. This information plays a vital role in determining fluid resuscitation requirements. The initial 24 hours following a burn injury, known as the “golden hour,” are critical. During this period, precisely calculated fluid resuscitation is crucial. The modified Parkland formula guides the calculation of this essential fluid volume to prevent complications. For pediatric burn patients, additional maintenance fluids are necessary to account for their ongoing growth and development needs. Furthermore, regardless of age, nutritional support is essential to combat potential nutritional deficiencies that can arise from burn injuries. The amount of nutritional fluid is calculated using Curreri’s formula.
Patient-Specific Breast Cancer Treatment Optimization with Radio Frequency Hyperthermia Therapy
Students: Jannatul Ferdous Anyotoma, Mrinmoy Nandi Bappa
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
Radio-frequency ablation (RFA) has gained a lot of interest recently as a promising minimally invasive technique that uses high ablative temperatures to treat deep- seated malignant tumors, such as breast cancer. RFA treatment outcomes might vary depending on how well the probes are positioned into the tumor and surrounding tissues, as well as the expertise and expertise of the treating doctor. To reflect this, there needs to be more well-established strategies for optimum outcome forecast before the ablative procedure. The purpose of this study is to determine the optimal conductive length, diameter, insertion angle of the probe, and the applied voltage, as well as duration of the procedure in order to assess the possibility of assisting the surgeon in developing the most efficient surgical plan using patient-specific geometry. Following the reconstruction of the breast and tumor geometry from MRI data, we optimized the outcome by running several Finite-Element method (FEM) simulations with different probe and probe-placement settings. The tumor’s size, shape, and location are all taken into consideration throughout the optimization process, along with the surrounding tissues’ thermal properties and perfusion rates. Based on four patient-specific geometries, this led to an exceptionally reliable forecast of the ablation procedure, which produced maximal destruction of tumor cells and minimal healthy cell damage. The application of our work to the development of RFA planning techniques personalized to individual patients will result in promising increases in treatment effectiveness and a decrease in related risks.
Arterial Tissue Drug Uptake Augmentation Using Disintegrating Film for Balloon-Based Drug Delivery
Students: Basnin Musfirat Mohiuddin, Nawshin Jannat
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
Neointimal hyperplasia is a pathological condition marked by the growth and movement of vascular smooth muscle cells (VSMCs) and the accumulation of extracellular matrix in the innermost layer of a blood vessel, known as the intima. This phenomenon typically arises in response to vascular injury, such as that caused by procedures like angioplasty or the placement of vascular grafts and stents. The resulting thickening of the vessel wall can cause restenosis, a re-narrowing of the vessel that reduces blood flow. Addressing neointimal hyperplasia is a major challenge in cardiovascular interventions, driving the need for continuous research and the development of innovative treatment approaches. A drug-eluting balloon (DEB) is an advanced medical device used in the treatment of vascular diseases, particularly to combat restenosis caused by neointimal hyperplasia. The balloon is coated with an anti- proliferative drug that, when the balloon is inflated at the site of a narrowed or blocked artery, is delivered directly to the arterial wall. This localized drug delivery inhibits the proliferation of VSMCs and prevents the excessive tissue growth characteristic of neointimal hyperplasia. DEBs offer several advantages over traditional treatments, such as drug-eluting stents, including the absence of a permanent implant and a reduced risk of long-term complications. They represent a promising approach in the ongoing effort to improve the outcomes of cardiovascular interventions. A drug-eluting balloon, paired with an innovative disintegrating film, represents a sophisticated and effective alternative to conventional stents. This advanced film enhances drug uptake, thereby maximizing therapeutic efficacy. Its performance has been meticulously validated through a range of characterizations, including High-Performance Liquid Chromatography (HPLC), Field Emission Scanning Electron Microscopy (FESEM), surface pH measurement, moisture content analysis, disintegration time assessment, and Fourier Transform Infrared Spectroscopy (FTIR). Each of these comprehensive evaluations underscores the potential of the disintegrating film as a superior solution to current vascular intervention challenges, highlighting its promise in improving patient outcomes and advancing cardiovascular treatment methodologies.
Assessment of GF120918 as a Potential Inhibitor of P-gp to Increase Cancer Drug Uptake: A Molecular Dynamics Study
Students: Mahbuba Ferdaous, Shawkat Osman Shishir
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
The development of multidrug resistance (MDR), which is primarily brought on by the overexpression of P-glycoprotein (P-gp/ABCB1/MDR1), is the primary reason why chemotherapy treatments for carcinomas fail. The lack of an experimentally determined 3D structure for the P-gp transporter until recently hindered the development of potential P-gp inhibitors through the use of in vitro/in vivo methods. So, to understand the atomic level interaction we focused our thesis on the molecular dynamics simulation of drug-protein only, with Gromacs and Autodock and vina. In vitro/In vivo research shows that MDR inhibitor like Gf-120918 binding with P-glycoprotein allows anti-carcinoma drugs (Taxol, doxorubicin etc.) to pass through the tissue to work on the affected cells. Our molecular dynamic simulation and docking scores also shows the same result as in silico methods validating the molecular/atomic level properties like rmsd, rmsf, radius of gyration, H-bond, Bonding Energy to be legitimate to use in further study.
Deep Learning Model for ECG Signal Classification
Student: Maksudul Hoque Rafi
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
Cardiovascular diseases (CVDs) are a leading cause of death worldwide, making early and accurate diagnosis critical for effective treatment and management. This thesis presents a deep learning model for classifying electrocardiogram (ECG) signals, leveraging the combined capabilities of Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BLSTM) networks. The model was trained using the MIT-BIH Arrhythmia Database, which consists of 68,000 ECG heartbeats. The study explored two training approaches: one using raw ECG signals and the other using spectrograms of the ECG signals. When trained with raw ECG signals, the model achieved an accuracy of 99.54%, precision of 99.55%, and recall of 99.54%. In contrast, training with spectrograms resulted in an accuracy of 97.91%, precision of 97.98%, and recall of 97.91%. The superior performance of the model trained with raw ECG signals highlights the importance of preserving temporal and morphological features present in the time domain for effective ECG classification. The findings show the potential of the proposed deep learning model to significantly enhance the diagnostic accuracy of CVDs, facilitating early detection and timely intervention. The implementation of such models in clinical practice can lead to improved patient outcomes and reduced healthcare costs.
Development of a Personalized 3D-Printed Guide for Dental Implant Surgery
Students: Atkia Atia, Md. Iftekharul Haque Fahim
Supervisor: Dr. Jahid Ferdous, Associate Professor
Abstract:
Dental implant surgery involves the precise placement of implants to achieve the best possible functional and aesthetic results for the missing teeth. It includes both corrective and cosmetic procedures. But free-handed surgeries are often less accurate and poorly predictable. A digitally planned surgical guide can increase the precision and quality in implant placement surgery. Though digital dentistry is still relatively new in Bangladesh, it has a lot of potential to improve treatment efficiency and accuracy. The purpose of this thesis is to develop and evaluate the use of preplanned, 3D-printed surgical guides for dental implant procedures performed in Bangladesh. A clinical trial is conducted as a part of the study to evaluate the efficacy of our surgical guide by comparing the accuracy of fully guided and free-handed dental implant surgeries. The surgical guides will be developed using 3D printing technology and Cone Beam CT data. In each case, the basis of evaluation is comparing the preoperative digital plan with the actual postoperative status. The study also explores the design considerations and procedural efficacy of surgical guides required to generalize the implementation of 3D-printed surgical guides in dental implantology. By integrating digital tools and advanced imaging techniques, this research provides a comprehensive analysis of the potential benefits and challenges associated with the widespread adoption of such technology. The findings of this study are intended to serve as a guideline for future practitioners, potentially revolutionizing the practice of dental implant surgery in Bangladesh and contributing to the digital advancements in dental healthcare.
Platelet-Rich Plasma Lysate Incorporated Gelatin/Sodium Alginate Hydrogel for Chronic Diabetic Wound Healing
Student: Fairooz Nawer
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Diabetic ulcers have been one of the burning issues in recent years due to their severe wound- healing-related complications, which can lead to amputation. Hyperglycemic conditions in diabetes cause severe growth factor deficiency, which directs the wound-healing process towards chronic inflammation, impairing its effective healing. Platelet-rich plasma (PRP) lysate, a combination of multiple growth factors, is explored for addressing the growth-factor compromised chronic diabetic wound treatment due to its tendency to accelerate wound healing by providing all the required growth factors and cytokines. Among different biomaterial-based delivery systems for PRP lysate, hydrogels with interpenetrating polymer networks have drawn attention due to their tunable swelling and degradation properties, which can be exploited for the controlled delivery of PRP lysate to the wound. Although natural polymer-based hydrogels have exhibited advantageous properties such as enhanced biocompatibility, efficient loading, and wound healing responses, pure natural polymer-based hydrogels still fall behind due to their low mechanical properties and uncontrolled release of biomolecules. This study fabricated a multiple natural polymer-based composite hydrogel (Gel/SA@PL) with interpenetrating networks composed of gelatin and sodium alginate where gelatin offered enhanced wound healing responses and sodium alginate provided efficient loading and sustained release of PRP lysate. Moreover, this composite hydrogel exhibited efficient, strong hydrogel formation loaded with PRP lysate and excellent hemostatic properties, along with enhanced wound healing process characterized by early initiation of re-epithelialization, granulation tissue formation, hair follicle production, and angiogenesis. Overall, Gel/SA@PL hydrogel holds the potential for treating diabetic wounds enhancing the quality of life by introducing personalized healthcare treatment.
LCC-Net: A Lightweight Deep Learning Framework for Lung Cancer Classification Using Lung CT Scans
Students: Adhora Madhuri, Nusaiba Sobir, Tasnia Binte Mamun
Supervisor: Dr. Taufiq Hasan, Professor
Abstract:
Lung cancer is the leading cause of cancer deaths worldwide, with nearly 2.5 million new cases diagnosed in 2022. Early detection is crucial for improving the survival rate of lung cancer patients, as it allows for effective and timely treatment interventions. But in developing countries like Bangladesh, where there is a severe shortage of qual- ified radiologists, the situation is a bit complicated. The manual identification and classification of pulmonary nodules are not only time-consuming but also demand a high level of expertise, leading to potential misdiagnoses or missed diagnoses due to variability in clinicians’ skills and the high volume of scans.Deep learning-based appli- cations play a significant role in this endeavor. Numerous publicly accessible datasets have inspired researchers to develop deep-learning algorithms for early lung cancer diagnosis. To aid in this process, we develop a lightweight deep learning framework LCC-Net for lung cancer classification using publicly available LUNA16 dataset. Our novel approach focused on intensity range branching using Hounsfield Units (HU) as a learning parameter. This method involved dividing inputs into various HU ranges, with each range processed through separate models. The Multi-Branched LCC-Net model demonstrated superior performance compared to the standard LCC-Net, par- ticularly with user-defined intensity ranges. Although our study did not achieve the highest accuracy, it provides a strong foundation for future improvements. This study underscores the potential of deep learning models to aid radiologists by pro- viding consistent, accurate lung nodule classification, thereby improving early lung cancer detection and reducing mortality rates in resource-limited settings.
Synthesis and Evaluation of Natural Gum Modified HPMC-Based Oral Disintegrating Film Incorporated with Folic Acid
Student: Munira Joshone
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Oral disintegrating film (ODF) is an effective alternative to the conventional oral dosage forms for geriatric and pediatric patients. To enhance patient compliance and ensure a pleasant mouthfeel, HPMC based ODFs were modified adding natural gums incorporated with folic acid. Gum acacia, tamarind gum, and guar gum were added as natural gums. The effect of the modification by adding gums with HPMC on disintegration time, folding endurance, mechanical strength, surface pH, ATR-FTIR spectra, XRD, TGA, contact angle, moisture content, transparency, and surface morphology was evaluated. 0.2% gum acacia added formulation (H_0.2A) showed more mechanical strength along with higher folding endurance and lower disintegration time among the formulations. In vitro drug release study was conducted for folic acid where the gum-added ODF formulations performed a more sustained and controlled release profile compared to the commercial tablet. Hence, natural gum modified HPMC based ODF can be advantageous for achieving rapid drug release.
AI-Powered Chest X-ray Report Writing Assistance for Radiologists
Students: Mahmud Wasif Nafee, Tasmia Rahman Aanika
Supervisor: Dr. Taufiq Hasan, Professor
Abstract:
Deciphering chest X-rays is crucial for diagnosing a range of disorders, including common ones like pneumonia and more serious problems like lung cancer and car- diomegaly. Radiologists dedicate much time and expertise to carefully examine these pictures in order to assure precise diagnoses, often managing extensive amounts of data in the face of increasing patient needs and limited staff resources. This rig- orous procedure not only burdens resources but also contributes to the exhaustion and fatigue experienced by radiologists. In order to address these difficulties, this study suggests utilizing advanced deep learning methods to provide automated aids for chest X-ray reports. The main objective is to improve the efficiency and accuracy of diagnoses by utiliz- ing AI-driven technologies that can conduct initial screens, maintain consistency in outcomes, and speed up the diagnostic process. The system attempts to successfully detect and highlight anomalies by incorporating models such as CheXNet for visual feature extraction and YOLO for illness localization. These developments have the dual purpose of decreasing the workload of radiologists and serving as teaching tools to train and improve diagnostic skills. The goals involve creating a system that can effectively identify diagnostic labels, produce reports based on impressions, and ensure that the language used in the reports adheres to clinical standards. Moreover, the system is specifically engineered for rapid processing in order to reduce the amount of time that physicians are not actively working, and it is also adaptable to accept different imaging modules. The hypothesis suggests that combining multi-label classifiers and advanced object identification algorithms will enhance the accuracy of localising objects and the semantic significance in generating reports. This, in turn, will assist radiologists in making well-informed therapeutic judgments. Moreover, the future endeavours entail augmenting the dataset by incorporating specific demographic information from Bangladesh to improve the effectiveness of the model. Additionally, there is a plan to modify and utilise more advanced language model architectures such as GPT-3 to provide reports with greater subtlety and sophistication. Additionally, the inclusion of systematic reviews conducted by radiologists will be incorporated to authenticate the clinical usefulness and precision of AI-generated reports. This comprehensive strategy seeks to create a strong structure for automated interpretation of chest X-rays that not only improves diagnostic capabilities but also promotes a more efficient healthcare delivery system. This thesis seeks to optimize the diagnostic workflow, alleviate radiologist burnout, and eventually improve patient care outcomes by utilizing AI technologies. The suggested system is a notable advancement towards creating a healthcare ecosystem that is more productive and efficient, benefiting both healthcare practitioners and patients.
Synthesis and Evaluation of Phosphorylated Chitosan as a Hemostatic Powder for Efficient Blood Clotting
Student: Raisa Islam
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Hemorrhaging is a significant cause of illness and death globally. Developing an efficient hemostat is essential for controlling bleeding and improving patient outcomes. For this purpose, phosphorylated chitosan (ChiP), a chitosan derivative, was used to synthesize microparticle hemostats. Based on the synthesis process, M_ChiP and H_ChiP are the two types of phosphorylated chitosan. Fourier-transform infrared spectroscopy (FTIR) and energy-dispersive X-ray spectroscopy (EDX) were used to validate the chemical structure of phosphorylated chitosan. The in vitro hemostatic effects of phosphorylated chitosan were assessed using a blood clotting index and erythrocyte adhesion. At the same time, its in vivo efficacy was evaluated using a mice liver laceration model. The in vivo hemostatic study revealed that M-ChiP and H_ChiP, had clotting times of 10.22 s and 11.14 s, respectively, and blood loss of 65 mg and 60 mg, discretely for mouse liver laceration. The developed hemostat showed high hemocompatibility. Adding polyphosphate to ChiP powder makes it a potentially effective hemostatic agent.
Vitamin E Loaded Nanoemulgel as a Topical Formulation
Student: Fahmida Akhtar
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
In this study, vitamin E-loaded nanoemulsion and nanoemulgels, containing biocompatible ingredients, were explored as potential formulations for the controlled and sustained topical delivery of antioxidants and moisturizing agents. Olive oil was used as the oil phase of the nanoemulsion, while various anionic natural gums, along with glycerol, were utilized as moisturizing and gelling agents in the nanoemulgels. The formulations were prepared via a low-energy method and had excellent stability during storage, and under thermodynamic stresses. The particle diameters of the vitamin E loaded nanoemulsion, the 2% xanthan gum-glycerol-based nanoemulgel and the 2% gellan gum-glycerol-based nanoemulgel were 84.83±5.64 nm, 241.14±2.58 nm and 286.85±2.67 nm, respectively, indicating promising transdermal penetration through intercellular and follicular routes. The nanoemulsion displayed 88.45% free radical scavenging in DPPH assay, highlighting its significant potential as topical antioxidant. In vitro occlusion studies revealed substantial occlusion potential of the nanoemulgels, with xanthan gum-based formulation reducing transepidermal water loss (TEWL) by as much as 33.33% and gellan gum-based formulation by 15.13%. Drug release studies demonstrated controlled and sustained release profiles, with the nanoemulsion releasing 90.95% of vitamin E over 18 days, compared to 79.80% and 13.08% for gellan and xanthan gum-based nanoemulgels, respectively. This suggests the potential for tailoring release rates by varying gum type and concentration. The release patterns fitted the Korsmeyer-Peppas model for all the formulations. Additionally, the nanoemulgels were found to be light, spreadable, non-greasy and hemocompatible, with a pH range suitable for topical use (4-6). These nanoformulations can be particularly useful in the repair of UV-damaged or moisture-deficient skin barriers and can be incorporated in photoprotective cosmetics .
ColpoSense: Development and Evaluation of an Artificial Intelligence Framework for Cervical Cancer Detection from Colposcopic Images
Students: Asfina Hassan Juicy, Raiyun Kabir
Supervisor: Dr. Taufiq Hasan, Professor
Abstract:
Cervical cancer currently poses a significant health threat due to late-stage diagnosis, leading to higher mortality rates and limited access to timely screenings, especially in underserved populations. In context, colposcopy is a screening method to visually examine the cervix using a colposcope for signs of cancerous or precancerous cervix tissue. Colposcopy, used for further examination of women with abnormalities from the cytology or HPV test, is based on the interpretation of cervical images from the colposcope. Four kinds of cervical images are used for visual inspection: saline, acetic acid, green filter, and iodine images. Although official organizations have released standards and quality control for colposcopy practice, the diagnoses are mainly dependent on the experience of doctors, which is subjective across variabilities. The gold standard for diagnosing cervical lesions is colposcopy-directed biopsy with histopathological confirmation. However, it is invasive and may cause complications such as bleeding or infection. Thus, an objective and accurate cervical screening approach based on existing clinical practices is needed. As traditional vision screening from colposcopy risks misdiagnosis due to a lack of trained experts in rural regions and is low accessibility due to time and personnel constraints, the primary focus is on leveraging advanced computational techniques to act as decision-support tools for the analysis of colposcopy images, addressing the limitations inherent in traditional vision screening approaches. Our work in this thesis utilizes deep learning methods to create artificial intelligence models that aid vision assessments, provide a reliable and efficient means of identifying transformation zones, and detect early signs of cervical abnormalities by predicting Swede scores. Central to the aim of this research is the recognition of the pressing need for scalable and cost-effective solutions in low-income settings. The proposed deep learning model seeks to mitigate the shortage of trained personnel by automating the intricate process of colposcopic image analysis. This streamlines the screening process and enables its widespread implementation, making cervical cancer screening more accessible to a larger population.
Delivery of Antioxidant Coenzyme Q10 Using Mucoadhesive Nanocarriers for the Treatment of Dry Eye
Student: Samina Nishat Binte Akram
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Novel Alginate-Mucin nanoparticles were prepared by ionotropic gelation method using Ca2+ as a crosslinker for application as a delivery agent for dry eye therapeutics. Alginate-Mucin nanoparticles exhibited significant mucoadhesive characteristics compared to pure alginate nanoparticles across all different ratios of alginate to mucin. The optimal ratio of 20% Mucin 80% alginate nanoparticles demonstrated stable nanoparticle configuration with spherical morphology and was chosen as an appropriate vehicle for actives. The final nano formulation exhibited shear thinning behavior, preferrable for ocular applications. The nanoparticles were non hemolytic and did not exhibit any signs of corneal edema or cell damage in vivo. Antioxidant coenzyme Q10 was encapsulated on the optimized nanoparticles with an average encapsulation efficiency of 96.3933%. Coenzyme Q10 loaded nanoparticles showed pseudo Fickian release pattern on appropriate release medium. In-vivo dry eye rabbit model was established and efficacy of nanoparticles as a suitable carrier for coenzyme Q10 was evaluated. In-vivo data demonstrated significant increase in tear production in 3 days under regular dosing compared to untreated rabbits and also indicated prevention of corneal damage from histological analysis.
Evaluation of Human Chorionic Membrane (HCM) Decellularization Methods for Wound Healing
Students: Purnopama Saha, Arpa Dhar
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Decellularization of biological biomaterials for tissue engineering is a critical process aimed at removing cellular components while preserving the extracellular matrix (ECM) components like collagen, GAGs(glycosaminoglycan) etc. Human placentas, specifically the chorion layers, are particularly attractive for this decellularization purpose due to their abundance post- birth and rich ECM composition. However, establishing effective decellularization protocols specific to human chorionic membrane (HCM) is essential. This study undertook a comprehensive analysis to determine optimal decellularization methods for human chorionic membrane (HCM) and prepare a scaffold that facilitates wound healing. Initially, preliminary tests were conducted to evaluate various parameters, including detergent types, freezing temperatures, and perfusion techniques. These tests aimed to assess the efficacy of different conditions in achieving complete cell removal while maintaining ECM integrity and ultrastructure. Among the detergents tested, sodium dodecyl sulfate (SDS) emerged as the most effective for cell removal, indicating its suitability for decellularization purposes. The impact of freezing placentas prior to decellularization was also investigated, revealing that freezing necessitated longer incubation periods in detergents to achieve effective decellularization. Both perfusion and immersion methods were found to be capable of removing cells from the placentas, providing flexibility in the decellularization process. Based on the preliminary tests, three protocols were selected for further analysis, involving histology, scanning electron microscopy (SEM), Fourier Transform Infrared Spectroscopy (FTIR), DNA quantification etc. To check the effectiveness of this human chorionic membrane derived collagen-based scaffold, it was applied in full thickness wound of mice and in vivo histology was performed for 3,7,14,21 days. In summary, this study compares different protocols and checks which method retains the ultrastructure and facilitates wound healing. These findings provide valuable insights for the development of optimization of decellularization protocols for human chorionic membrane, facilitating its potential applications in tissue engineering and regenerative medicine, for example in wound healing.
Exploring Deep Learning Algorithms for Aedes Mosquito Detection from Smartphone Images
Students: Tonmoy Chandro Saha, Mahian Kabir Joarder
Supervisor: Dr. Taufiq Hasan, Professor
Abstract:
Add abstract here.The outbreak of dengue fever in recent years has become a grave public health concern as it has spread to 20 countries in South America and South Asia. As vectors of the flavivirus, several mosquito species belonging to the Aedes genus are responsible for transmitting dengue fever. Effective vector surveillance and control are essential in reducing dengue outbreaks. However, due to their minute variations in anatomical structure, it is challenging to identify Aedes mosquitos without expert entomologists using a microscope. In this regard, deep learning algorithms can play a vital role in identifying mosquitoes using smartphone-captured images and pave the way for deskilling automated vector surveillance, provided that sufficient training examples are available. In this study, we develop the “Aedes Mosquito Image Dataset (AMID)” consisting of smartphone-captured mosquito images consisting of 8 class labels: Aedes aegyptie, Aedes koreicus, Aedes albopictus, Culex pipiens, Armigeres subalbatus, Culex quinquifasciatus, Aedes japonicus and some other unknown species. The images are collected by trapping mosquitos in several locations in Dhaka, followed by image capturing and expert annotations. Additional image data is collected from open- access online repositories. Data augmentation is used for increasing the sample size, and a 5-fold cross-validation experiment is set up. In the later steps, several pre- trained deep learning models, namely, EfficientNetB0, DenseNet121, MobileNet, and ResNet101, were employed to classify the different mosquito species. The results show that EfficientNetB0 has achieved the best overall accuracy of 92.06 (±0.34)% while DenseNet121, MobileNet and ResNet101 have achieved an accuracy of 90.67 (±0.36)%, 89.64 (±0.38)% and 85.67 (±0.44)%. We have also developed a prototype smartphone web application that can identify mosquito species in real time from uploaded mosquito images.
A Tough Gelatin and Silk Fibroin-Based Unoxidized Tannic Acid Modified Bioadhesive for Topical Use
Student: Tahmed Ahmed
Supervisor: Dr. Muhammad Tarik Arafat, Professor
Abstract:
Tissue adhesives are essential in surgery because they seal tissues, heal wounds, and stop bleeding. However, traditional tissue adhesives are expensive, toxic, or have poor tissue adhesion. To improve the current limitations of bioadhesive, a gelatin (GA)-silk fibroin (SF) based unoxidized tannic acid (TA) modified bioadhesive (PG-TA-SF-HCl) was synthesized with superior adhesion and hemostasis with reasonable settling time. Unoxidized environment gave the advantage of controlling the amount of protonation in GA and TA. The adhesion strength (89±2.7 kPa) and toughness (~554 kJ/m3 ) were significantly higher than previously reported tannic acid-gelatin based bioadhesives. FTIR spectroscopy proved the integration of tannic acid and presence of silk fibroin and formation of hydrogen bonds. PG-TA-SF-HCl maintained adhesive strength underwater for ten days and longer and could also join two cut parts of mice’s internal organs. Moreover, PG-TA-SF-HCl shows strong antibacterial properties against S. aureus, hemostatic abilities since it contains TA, a natural and antibacterial cross-linker abundant in the hydroxyl group with the capability of non-covalent bonding in an unoxidized environment and SF, a fascinating natural biomaterial that exhibits remarkable mechanical properties and biocompatibility. PG-TA-SF-HCl was able to stop bleeding in mice liver in 12.9 ± 5 seconds and also showed excellent blood clotting ability. The cross-linking among 20% (w/v) GA, 8% (w/v) SF, and 20% (w/v) TA is done by centrifugation process at room temperature. Hydrochloric acid was used to create the unoxidized environment. The novelty lies not only in the enhanced adhesion strength and toughness but also in the single-step facile fabrication process of the adhesive. This study results in a tough, antibacterial adhesive with increased adhesion strength. Its properties can be tuned by varying pH of the medium and applied to topical region of the body for wound closure. It also has the potential to be a successful tissue adhesive with a variety of uses, including controlled drug delivery at wound sites and other biomedical applications.