Browsing by Author "Hasan, Md. Zahid"

Now showing 1 - 20 of 48

A Comparative Study of Machine Learning Models with Lasso and Shap Feature Selection for Breast Cancer Prediction
(Elsevier, 2024-06-25) Shaon, Md. Shazzad Hossain; Karim, Tasmin; Shakil, Md. Shahriar; Hasan, Md. Zahid
In recent decades, breast cancer has become the most prevalent type of cancer that impacts women in the world, which shows a significant risk to the death rates of women. Early identification of breast cancer might drastically decrease patient mortality and greatly improve the chance of an effective treatment. In modern times, machine learning models have become crucial for classifying cancer and strengthening both the accuracy and efficiency of diagnostic and medical treatment strategies. Therefore, this study is focused on early detection of breast cancer using a variety of machine learning algorithms and desires to identify the most effective feature selection process with an amalgamated dataset. Initially, we evaluated five traditional models and two meta-models on separate datasets. To find the most valuable features, the study used the Least Absolute Shrinkage and Selection Operator (LASSO) as well as SHapley Additive exPlanations (SHAP) selection methods and analyzed them through a wide range of performance regulations. Additionally, we applied these models to the combined dataset and observed that the mergeddataset was significantly beneficial for breast cancer diagnosis. After analyzing the feature selection strategies, it was demonstrated that the majority of models performed more accurately when utilizing SHAP methodologies. Notably, three traditional models and two meta-classifiers obtained an accuracy of 99.82%, demonstrating superior performance compared to state-of-the-art methods. This advancement holds a crucial role as it lays the foundation for refining diagnostic tools and enhancing the progression of medical science in this field.
A Comparative Study of Machine Learning Models with Lasso and Shap Feature Selection for Breast Cancer Prediction
(Elsevier, 2024-06-25) Shaon, Md. Shazzad Hossain; Karim, Tasmin; Shakil, Md. Shahriar; Hasan, Md. Zahid
In recent decades, breast cancer has become the most prevalent type of cancer that impacts women in the world, which shows a significant risk to the death rates of women. Early identification of breast cancer might drastically decrease patient mortality and greatly improve the chance of an effective treatment. In modern times, machine learning models have become crucial for classifying cancer and strengthening both the accuracy and efficiency of diagnostic and medical treatment strategies. Therefore, this study is focused on early detection of breast cancer using a variety of machine learning algorithms and desires to identify the most effective feature selection process with an amalgamated dataset. Initially, we evaluated five traditional models and two meta-models on separate datasets. To find the most valuable features, the study used the Least Absolute Shrinkage and Selection Operator (LASSO) as well as SHapley Additive exPlanations (SHAP) selection methods and analyzed them through a wide range of performance regulations. Additionally, we applied these models to the combined dataset and observed that the mergeddataset was significantly beneficial for breast cancer diagnosis. After analyzing the feature selection strategies, it was demonstrated that the majority of models performed more accurately when utilizing SHAP methodologies. Notably, three traditional models and two meta-classifiers obtained an accuracy of 99.82%, demonstrating superior performance compared to state-of-the-art methods. This advancement holds a crucial role as it lays the foundation for refining diagnostic tools and enhancing the progression of medical science in this field.
A Multifactor Authentication Model to Mitigate the Phishing Attack of E-Service Systems from Bangladesh Perspective
(Scopus, 2020) Hasan, Md. Zahid; Sattar, Abdus; Mahmud, Arif; Talukder, Khalid Hasan
A new multifactor authentication model has been proposed for Bangladesh taking cost-effectiveness in primary concern. We considered two-factor authentications in our previous e-service models which were proven to be insufficient in terms of phishing attack. Users often fail to identify phishing site and provide confidential information unintentionally, resulting in a successful phishing attempt. As a result, phishing can be considered as one of the most serious issues and required to be addressed and mitigated. Three factors were included to form multifactor authentication, namely, user ID, secured image with caption, and one-time password. Through the survey, the proposed multifactor model is proven to be better by 59% points for total users which comprises 55% points for technical users and 64% points for nontechnical users in comparison to traditional two-factor authentication model. Since the results and recommendations from the user were reflected in the model, user satisfaction was achieved.
A Robust Framework Combining Image Processing and Deep Learning Hybrid Model to Classify Cardiovascular Diseases Using a Limited Number of Paper-Based Complex ECG Images
(Scopus, 22-11-07) Fatema, Kaniz; Montaha, Sidratul; Rony, Md. Awlad Hossen; Azam, Sami; Hasan, Md. Zahid; Jonkman, Mirjam
Heart disease can be life-threatening if not detected and treated at an early stage. The electrocardiogram (ECG) plays a vital role in classifying cardiovascular diseases, and often physicians and medical researchers examine paper-based ECG images for cardiac diagnosis. An automated heart disease prediction system might help to classify heart diseases accurately at an early stage. This study aims to classify cardiac diseases into five classes with paper-based ECG images using a deep learning approach with the highest possible accuracy and the lowest possible time complexity. This research consists of two approaches. In the first approach, five deep learning models, InceptionV3, ResNet50, MobileNetV2, VGG19, and DenseNet201, are employed. In the second approach, an integrated deep learning model (InRes-106) is introduced, combining InceptionV3 and ResNet50. This model is developed as a deep convolutional neural network capable of extracting hidden and high-level features from images. An ablation study is conducted on the proposed model altering several components and hyper parameters, improving the performance even further. Before training the model, several image pre-processing techniques are employed to remove artifacts and enhance the image quality. Our proposed hybrid InRes-106 model performed best with a testing accuracy of 98.34%. The InceptionV3 model acquired a testing accuracy of 90.56%, the ResNet50 89.63%, the DenseNet201 88.94%, the VGG19 87.87%, and the MobileNetV2 achieved 80.56% testing accuracy. The model is trained with a k-fold cross-validation technique with different k values to evaluate the robustness further. Although the dataset contains a limited number of complex ECG images, our proposed approach, based on various image pre-processing techniques, model fine-tuning, and ablation studies, can effectively diagnose cardiac diseases.
A Study on Knowledge and Attitude of HIV/AIDS Among University Students of Bangladesh
(East West University, 5/21/2016) Hasan, Md. Zahid
HIV/ AIDS are the most serious health problem in the world. The global epidemic of HIV/AIDS is now progressing at a rapid rate among young people. University people are at high risk of HIV and AIDS infections. Therefore, awareness is needed to control and prevent the transmission of HIV/AIDS. The purpose of our study was to determine the knowledge level of University students in Bangladesh about HIV/AIDS and to support government efforts to control the spread of HIV and AIDS. The survey was conducted on 1003 different university students in Dhaka city by using a pre-structured questionnaire. All of the respondents have heard about the term HIV/AIDS. The major sources of information were media (63.5%) and educational institute (36.6%). Majority of the students said that there is no treatment (60.10%) and vaccine (63%) available. According to most of respondents the disease can be transmitted by unprotected sex (94.2%), mother to fetus (86.4%), sharing infected needles or syringe (90.2%), blood transfusion (92.1%) and breast feeding (69.1%). Some respondents said it can be transmitted by razor sharing (40.2%) and medical or dental procedure (30.1%). Some students had misconception about the transmission. About 22% students said it can be transmitted by mosquito bite. Regarding the knowledge about control and prevention most of them had better knowledge. Most participants had positive attitude towards infected person. The study suggested that to reduce the misconception, and increase awareness education and intervention programs are needed to increase the level of knowledge and awareness of HIV/AIDS.
A Systematic Review of Graph Neural Network in Healthcare-Based Applications: Recent Advances, Trends, and Future Directions
(Scopus, 2024-01-16) Saha, Arpa; Hasan, Md. Zahid; Noori, Sheak Rashed Haider; Moustafa, Ahmed
Graph neural network (GNN) is a formidable deep learning framework that enables the analysis and modeling of intricate relationships present in data structured as graphs. In recent years, a burgeoning interest has arisen in exploiting the latent capabilities of GNN for healthcare-based applications, capitalizing on their aptitude for modeling complex relationships and unearthing profound insights from graph-structured data. However, to the best of our knowledge, no study has systemically reviewed the GNN studies conducted in the healthcare domain. This study has furnished an all-encompassing and erudite overview of the prevailing cutting-edge research on GNN in healthcare. Through analysis and assimilation of studies, current research trends, recurrent challenges, and promising future opportunities in GNN for healthcare applications have been identified. China emerged as the leading country to conduct GNN-based studies in the healthcare domain, followed by the USA, UK, and Turkey. Among various aspects of healthcare, disease prediction and drug discovery emerge as the most prominent areas of focus for GNN application, indicating the potential of GNN for advancing diagnostic and therapeutic approaches. This study proposed research questions regarding diverse aspects of GNN in the healthcare domain and addressed them through an in-depth analysis. This study can provide practitioners and researchers with profound insights into the current landscape of GNN applications in healthcare and can guide healthcare institutes, researchers, and governments by demonstrating the ways in which GNN can contribute to the development of effective and efficient healthcare systems
A Variable Length Key Based Cryptographic Approach on Cloud Data
(Scopus, 2019-12-21) Ghosh, Pronab; Jabiullah, Md. Ismail; Hasan, Md. Zahid; Atik, Syeda Tanjila
Security has emerged to be a concerning issue in cloud computing as numerous sensitive data are processed and transferred over the cloud servers. In this paper, a variable length key based security mechanism has been designed, developed, and implemented by using the Advanced Encryption Standard (AES) protocol for the secured delivery of cloud data which also works as SaaS (Software as a Services). The length of the used key is given as input by the user followed by the selection of a secret key which is fed into the AES cryptographic system with the desired cloud data message thus producing the Cipher text that is to be transmitted to the destination. In the receiver end, the reverse process is performed with the same key on the received Cipher text and the plaintext is retrieved. Using Python to implement and validate the security process, several messages from cloud users are used and the result for each input is analyzed. The result of our study is then compared with the two existing approaches which clearly shows the advancement of the proposed approach. This process can be applied in any secured electronic message transactions for cloud data.
AMP-RNNpro: a two-stage approach for identification of antimicrobials using probabilistic features
(Scopus, 2024-06-05) Shaon, Md. Shazzad Hossain; Karim, Tasmin; Sultan, Md. Fahim; Ali, Md. Mamun; Ahmed, Kawsar; Hasan, Md. Zahid; Moustafa, Ahmed; Bui, Francis M.; Al-Zahrani, Fahad Ahmed
Antimicrobials are molecules that prevent the formation of microorganisms such as bacteria, viruses, fungi, and parasites. The necessity to detect antimicrobial peptides (AMPs) using machine learning and deep learning arises from the need for efficiency to accelerate the discovery of AMPs, and contribute to developing effective antimicrobial therapies, especially in the face of increasing antibiotic resistance. This study introduced AMP-RNNpro based on Recurrent Neural Network (RNN), an innovative model for detecting AMPs, which was designed with eight feature encoding methods that are selected according to four criteria: amino acid compositional, grouped amino acid compositional, autocorrelation, and pseudo-amino acid compositional to represent the protein sequences for efficient identification of AMPs. In our framework, two-stage predictions have been conducted. Initially, this study analyzed 33 models on these feature extractions. Then, we selected the best six models from these models using rigorous performance metrics. In the second stage, probabilistic features have been generated from the selected six models in each feature encoding and they are aggregated to be fed into our final meta-model called AMP-RNNpro. This study also introduced 20 features with SHAP, which are crucial in the drug development fields, where we discover AAC, ASDC, and CKSAAGP features are highly impactful for detection and drug discovery. Our proposed framework, AMP-RNNpro excels in the identification of novel Amps with 97.15% accuracy, 96.48% sensitivity, and 97.87% specificity. We built a user-friendly website for demonstrating the accurate prediction of AMPs based on the proposed approach
Amp-rnnpro: A Two-stage Approach for Identification of Antimicrobials Using Probabilistic Features
(Springer Nature, 2024-06-05) Shaon, Md. Shazzad Hossain; Karim, Tasmin; Sultan, Md. Fahim; Ali, Md. Mamun; Ahmed, Kawsar; Hasan, Md. Zahid; Moustafa, Ahmed; Bui, Francis M.; Al-Zahrani, Fahad Ahmed
Antimicrobials are molecules that prevent the formation of microorganisms such as bacteria, viruses, fungi, and parasites. The necessity to detect antimicrobial peptides (AMPs) using machine learning and deep learning arises from the need for efficiency to accelerate the discovery of AMPs, and contribute to developing effective antimicrobial therapies, especially in the face of increasing antibiotic resistance. This study introduced AMP-RNNpro based on Recurrent Neural Network (RNN), an innovative model for detecting AMPs, which was designed with eight feature encoding methods that are selected according to four criteria: amino acid compositional, grouped amino acid compositional, autocorrelation, and pseudo-amino acid compositional to represent the protein sequences for efficient identification of AMPs. In our framework, two-stage predictions have been conducted. Initially, this study analyzed 33 models on these feature extractions. Then, we selected the best six models from these models using rigorous performance metrics. In the second stage, probabilistic features have been generated from the selected six models in each feature encoding and they are aggregated to be fed into our final meta-model called AMP-RNNpro. This study also introduced 20 features with SHAP, which are crucial in the drug development fields, where we discover AAC, ASDC, and CKSAAGP features are highly impactful for detection and drug discovery. Our proposed framework, AMP-RNNpro excels in the identification of novel Amps with 97.15% accuracy, 96.48% sensitivity, and 97.87% specificity. We built a user-friendly website for demonstrating the accurate prediction of AMPs based on the proposed approach which can be accessed at http://13.126.159.30/.
An Effective Approach To Address Processing Time and Computational Complexity Employing Modified CCT for Lung Disease Classification
(Scopus, 22-11-22) Khan, Inam Ullah; Azam, Sami; Montaha, Sidratul; Mahmud, Abdullah Al; Rafid, A.K.M. Rakibul Haque; Hasan, Md. Zahid; Jonkman, Mirjam
Early identification and adequate treatment can help prevent lung disorders from becoming chronic, severe, and life-threatening. X-ray images are commonly used and an automated and effective method involving deep learning techniques can potentially contribute to quick and accurate diagnosis of lung disorders. However, in the study of medical imaging using deep learning, two obstacles limit interpretability. One is an insufficient and imbalanced number of training samples in most medical datasets. The other is excessive training time. Although training time can be reduced by decreasing the number of pixels in the images, training with low resolution images tends to result in poor performance. This study represents a solution to overcome these impediments by balancing the number of images and reducing overall processing time while preserving accuracy. The dataset used in this research contains an unequal number of images in the different classes. The quantity of data in the classes is balanced by creating synthetic images based on the patterns and characteristics of the original images, using a Deep Convolutional Generative Adversarial Network (DCGAN). Unwanted regions are removed from the X-ray images, the brightness and contrast of the images are enhanced, and the abnormalities are highlighted by using different artifact removal, noise reduction, and enhancement techniques. We propose a Modified Compact Convolutional Transformer (MCCT) model using 32 × 32 sized images for the categorization of lung disorders into four classes. An ablation study of eleven cases is employed to adjust several hyper parameters and layer topologies. This reduces training time while preserving accuracy. Six transfer learning models, VGG19, VGG16, ResNet152, ResNet50, ResNet50V2, and MobileNet are applied with the same image size the performance is compared with the proposed MCCT model. Our MCCT model records the greatest test accuracy of 95.37%, requiring a short training time, 10-12 s/epoch, whereas the other models only reach near-moderate performance with accuracies ranging from 43% to 79% and training times of 80-90 s/epoch. The robustness of the model with regards to the number of training samples is validated by training the model multiple times reducing the number of training images gradually from 49621 images to 6204 images. Results suggest that even with a smaller dataset, the performance is sustained. Our proposed approach may contribute to an effective CAD based diagnostic system by addressing the issues of insufficient and imbalanced numbers of medical images, excessive training times and low-resolution images.
An Efficient Modified Bagging Method for Early Prediction of Brain Stroke
(Scopus, 2019-07-12) Alam, Md. Mahabur; Hasan, Md. Mehadi; Hasan, Md. Zahid
Brain stroke become a serious cardiovascular and cerebral disease causes of human death. Precisely predicting stroke effect from a set of predictive attributes may classify high-risk patients and guide cure approaches, leading to reduce relative incidence. In respect to, we have collected the information regarding brain stroke patient's data from five renowned hospitals in Bangladesh with connectivity in patients with acute thalamic ischemic stroke (melanoma), Atypical Nevus (cancer risk) and Common Nevus (No cancer risk). In this work, we propose an ensemble based Modified Bootstrap Aggregating (Bagging) technique for pattern classification. Existing bagging algorithm, can usually progress the performance of a single classifier. However, they typically need larger space as well as quite time-consuming predictions. However, our proposed accuracy based pruning bagging method can improve the classification performance and reduce ensemble size. In general, our proposed modified bagging technique is more appropriate than traditional bagging technique for the prediction of brain stroke disease patients with greater accuracy of 96%.
An IoT Based System for Printing Braille Letter from Speech
(IEEE, 2020-06) Ahmed, Foysal; Choudhury, Abu Raihan; Rakshit, Aniruddha; Hasan, Md. Zahid
Visually impaired person cannot perform their activities like ordinary person. It becomes more difficult when a person is visually impaired along with deafness. Sometimes, it is quite impossible to communicate with such people whether in speaking or writing format. Therefore, Expert Braille Communicating System (EBCS) is a promising solution which provides braille writing format from voice. Unfortunately, people of underdeveloped countries are not getting the facilities of advanced electronic braille system for overpricing. In this work, an attempt has taken on designing a modified braille device which is economical and more compelling than previous. The device takes voice as input from an android apps and can convert the voice signal into text format of braille and prints the braille letter in paper. EBCS is trained with 1 epoch and the accuracy is achieved of 97.6%. By implementing the considered Braille system, visually impaired person can entirely engage themselves in the society.
Bangla song genre recognition using artificial neural network
(Scopus, 2024-06-24) Hasan, Md. Zahid; Akter, Mariam; Sultana, Nishat; Noori, Sheak Rashed Haider
p>Music has a control over human moods and it can make someone calm or excited. It allows us to feel all emotions we experience. Nowadays, people are often attached with their phones and computers listening to music on Spotify, Soundcloud or any other internet platform. Music Information retrieval plays an important role for music recommendation according to lyrics, pitch, pattern of choices, and genre. In this study, we have tried to recognize the music genre for a better music recommendation system. We have collected an amount of 1820 Bangla songs from six different genres including Adhunik, Rock, Hip hop, Nazrul, Rabindra and Folk music. We have started with some traditional machine learning algorithms having K-Nearest Neighbor, Logistic Regression, Random Forest, Support Vector Machine and Decision Tree but ended up with a deep learning algorithm named Artificial Neural Network with an accuracy of 78% for recognizing music genres from six different genres. All mentioned algorithms are experimented with transformed mel-spectrograms and Mean Chroma Frequency Values of that raw amplitude data. But we found that music Tempo having Beats per Minute value with two previous features present better accuracy.
Bayesian Optimized Machine Learning Model for Automated Eye Disease Classification from Fundus Images
(Scopus, 2024-09-16) Zannah, Tasnim Bill; Kafi, Md. Abdulla-Hil-; Shuva, Taslima Ferdaus; Bhuiyan, Touhid; Sheakh, Md. Alif; Hasan, Md. Zahid; Rahman, Md. Tanvir; Khan, Risala Tasin; Kaiser, M. Shamim; zaman, Md Whaiduz
Eye diseases are defined as disorders or diseases that damage the tissue and related parts of the eyes. They appear in various types and can be either minor, meaning that they do not last long, or permanent blindness. Cataracts, glaucoma, and diabetic retinopathy are all eye illnesses that can cause vision loss if not discovered and treated early on. Automated classification of these diseases from fundus images can empower quicker diagnoses and interventions. Our research aims to create a robust model, BayeSVM500, for eye disease classification to enhance medical technology and improve patient outcomes. In this study, we develop models to classify images accurately. We start by preprocessing fundus images using contrast enhancement, normalization, and resizing. We then leverage several state-of-the-art deep convolutional neural network pre-trained models, including VGG16, VGG19, ResNet50, EfficientNet, and DenseNet, to extract deep features. To reduce feature dimensionality, we employ techniques such as principal component analysis, feature agglomeration, correlation analysis, variance thresholding, and feature importance rankings. Using these refined features, we train various traditional machine learning models as well as ensemble methods. Our best model, named BayeSVM500, is a Support Vector Machine classifier trained on EfficientNet features reduced to 500 dimensions via PCA, achieving 93.65 ± 1.05% accuracy. Bayesian hyperparameter optimization further improved performance to 95.33 ± 0.60%. Through comprehensive feature engineering and model optimization, we demonstrate highly accurate eye disease classification from fundus images, comparable to or superior to previous benchmarks.
Breastnet18
(Biology, 2021-11-13) Montaha, Sidratul; Azam, Sami; Muhammad Rakibul Haque Rafid, Abul Kalam; Ghosh, Pronab; Hasan, Md. Zahid; Jonkman, Mirjam; De Boer, Friso
Background: Identification and treatment of breast cancer at an early stage can reduce mortality. Currently, mammography is the most widely used effective imaging technique in breast cancer detection. However, an erroneous mammogram based interpretation may result in false diagnosis rate, as distinguishing cancerous masses from adjacent tissue is often complex and error-prone. Methods: Six pre-trained and fine-tuned deep CNN architectures: VGG16, VGG19, MobileNetV2, ResNet50, DenseNet201, and InceptionV3 are evaluated to determine which model yields the best performance. We propose a BreastNet18 model using VGG16 as foundational base, since VGG16 performs with the highest accuracy. An ablation study is performed on BreastNet18, to evaluate its robustness and achieve the highest possible accuracy. Various image processing techniques with suitable parameter values are employed to remove artefacts and increase the image quality. A total dataset of 1442 preprocessed mammograms was augmented using seven augmentation techniques, resulting in a dataset of 11,536 images. To investigate possible over fitting issues, a k-fold cross validation is carried out. The model was then tested on noisy mammograms to evaluate its robustness. Results were compared with previous studies. Results: Proposed BreastNet18 model performed best with a training accuracy of 96.72%, a validating accuracy of 97.91%, and a test accuracy of 98.02%. In contrast to this, VGGNet19 yielded test accuracy of 96.24%, MobileNetV2 77.84%, ResNet50 79.98%, DenseNet201 86.92%, and InceptionV3 76.87%. Conclusions: Our proposed approach based on image processing, transfer learning, fine-tuning, and ablation study has demonstrated a high correct breast cancer classification while dealing with a limited number of complex medical images.
Burst Header Packet Flood Detection in Optical Burst Switching Network Using Deep Learning Model
(Elsevier B.V., 2018-11-19) Hasan, Md. Zahid; Hasan, K.M. Zubair; Sattar, Abdus
The Optical Burst Switching (OBS) network is mostly victimized to the Denial of Service (DOS) attack, referred as Burst Header Packet (BHP) flooding attack can prevent reasonable traffics from keeping the necessary resources at transitional core nodes. The attack scenario is to flood the malicious BHP without acknowledging Data Bursts (DB) which can affect low bandwidth utilization, degrade network performance, high data loss rate and ultimately DOS. Therefore, machine predicted analysis has become very promising in recent decades that can effectually identify the attack in the optical switching network. However, due to a very small number of samples of the datasets, traditional machine learning approaches such as Naïve Bayes, K-Nearest Neighbor’s (KNN) and Support Vector Machine (SVM) cannot analyse the data efficiently. In this regard, we intend a Deep Convolution Neural Network (DCNN) model to automatically detect the edge nodes at an early stage. Finally, presented that proposed deep model is working enhanced rather than any other traditional model (e.g. Naïve Bayes, SVM and KNN).
Classification of Succulent Plant Using Convolutional Neural Network
(Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST, Springer, 2020-07-30) Das, Ashik Kumar; Iqbal, Md. Asif; Paul, Bidhan; Rakshit, Aniruddha; Hasan, Md. Zahid
Machine learning methods such as deep neural networks have remarkably improved plant species classification in recent years. It is very challenging task to classify plant species based on their categories. In this work, deep learning approach is explained to identify and classify succulent plant species using VGG19, three layers CNN and five layers CNN network on our dataset. The proposed architecture achieved a significant result from VGG19 and three layers CNN model. In succulent plant image dataset, there are 10 different classes of succulent and non-succulent plants. The dataset consists of 3632 succulent plant images and 200 non-succulent plant images. The model achieved 99.77% accuracy which performs better than VGG19 and three layers CNN model
Classification of Succulent Plant Using Convolutional Neural Network
(Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST, Springer, 2020-07-30) Hasan, Md. Zahid; Rakshit, Aniruddha; Das, Ashik Kumar; Iqbal, Md. Asif; Paul, Bidhan
Machine learning methods such as deep neural networks have remarkably improved plant species classification in recent years. It is very challenging task to classify plant species based on their categories. In this work, deep learning approach is explained to identify and classify succulent plant species using VGG19, three layers CNN and five layers CNN network on our dataset. The proposed architecture achieved a significant result from VGG19 and three layers CNN model. In succulent plant image dataset, there are 10 different classes of succulent and non-succulent plants. The dataset consists of 3632 succulent plant images and 200 non-succulent plant images. The model achieved 99.77% accuracy which performs better than VGG19 and three layers CNN model.
Deep Learning-Based Analysis of COVID-19 X-Ray Images: Incorporating Clinical Significance and Assessing Misinterpretation
(SAGE Publications, 2023-11-06) Bhuiyan, Md. Rahad Islam; Azam, Sami; Montaha, Sidratul; Jim, Risul Islam; Karim, Asif; Khan, Inam Ullah; Brady, Mark; Hasan, Md. Zahid; Boer, Friso De; Mukta, Md. Saddam Hossain
COVID-19, pneumonia, and tuberculosis have had a significant effect on recent global health. Since 2019, COVID-19 has been a major factor underlying the increase in respiratory-related terminal illness. Early-stage interpretation and identification of these diseases from X-ray images is essential to aid medical specialists in diagnosis. In this study, (COV-X-net19) a convolutional neural network model is developed and customized with a soft attention mechanism to classify lung diseases into four classes: normal, COVID-19, pneumonia, and tuberculosis using chest X-ray images. Image preprocessing is carried out by adjusting optimal parameters to preprocess the images before undertaking training of the classification models. Moreover, the proposed model is optimized by experimenting with different architectural structures and hyperparameters to further boost performance. The performance of the proposed model is compared with eight state-of-the-art transfer learning models for a comparative evaluation. Results suggest that the COV-X-net19 outperforms other models with a testing accuracy of 95.19%, precision of 96.49% and F1-score of 95.13%. Another novel approach of this study is to find out the probable reason behind image misclassification by analyzing the handcrafted imaging features with statistical evaluation. A statistical analysis known as analysis of variance test is performed, to identify at which point the model can identify a class accurately, and at which point the model cannot identify the class. The potential features responsible for the misclassification are also found. Moreover, Random Forest Feature importance technique and Minimum Redundancy Maximum Relevance technique are also explored. The methods and findings of this study can benefit in the clinical perspective in early detection and enable a better understanding of the cause of misclassification.
Development of Oatmeal Cookies and Quality Assessment
(Daffodil International University, 2019-12-09) Hasan, Md. Zahid
The present study aimed to develop oatmeal cookies with changed wt% of rolled oats as 20% (sample 1), 40% (sample 2), and 60% (sample 3) along with other ingredients like flour, sugar, butter, ghee, milk powder, baking soda, vanilla extract and to specify the quality parameters of the products by proximate analysis. Moisture content of the products were increased as the percentage of the oats increased and were found as 3.72%, 4.14%, 6.28%f for the samples 1, 2 and 3 respectively. Ash content of the samples also showed positive variation with the change of the weight percentage of oats in products and the value for sample 1 was 2.75%, sample 2 was 3.60% and sample 3 was 4.15%. Protein content of the developed samples were increased as percentage of oats and the value of the sample 1 was 3.98, sample 2 was 5.72, sample 3 was 6.56. Fat content were also increased as the changed of oats %, sample 1 was 2.58%, sample 2 was 3.44%, sample 3 was 4.3%. All the three samples were analyzed to find out pH and the values obtained sample 1 was 2.54, sample 2 was 3.18 and sample 3 was 3.72.