Browsing by Author "Assaduzzaman, Md."

Now showing 1 - 5 of 5

A Deep Learning Approach to Detect and Classification of Lung Cancer
(IEEE, 2023-01-26) Khatun, Mst. Farhana; Ajmain, Moshfiqur Rahman; Assaduzzaman, Md.
Cancer is a name of fear to people in the world. Ev- ery year millions of people dead of cancer in the world and lung cancer is one of them. Lung cancer is classified by our research. Non-small cell lung cancer (NSCLC) is the most common of the two main types of lung cancer. Here we have classified our model NSCLC into 2 subtypes Adenocarcinoma and Squamous Cell Carcinoma and non-cancerous benign tumors. The CNN model is utilized here for classification (VGG19, ResNet50, EfficientNetB7 and MobileNetV2). We used 15 thousand image data. The Augmentor package was utilized to enhance to 15 thousand from 250 benign lung tissue, 250 lung adenocarcinomas, and 250 lung squamous cell carcinomas. In comparison to other models, ResNet50 has the best accuracy of 98% among our proposed models. By putting this model into practice, medical experts will be able to create an accurate, automatic method for diagnosing different forms of lung cancer
A Machine Learning Approach to Predict SEER Cancer
(Springer, 2022-07-27) Abid, DM. Mehedi Hasan; Islam, Tariqul; Zaman, Zahura; Yusuf, Fahim; Assaduzzaman, Md.; Hossain, Syed Akhter; Jabiullah, Md. Ismail
The SEER database is among the persuading stores regarding malignancy pointers inside us. The SEER list helps impact investigation for the gigantic measure of patients’ bolstered viewpoints for the most part ordered as an insightful segment and impact. Assistant careful proof nearly the carcinoma dataset is ordinarily started on the site of the National Cancer Institute. The main point of this work is that depending on the individual’s manifestations, and we will foresee whether individuals are in danger of malignant growth or not. Perseverance and desire for the benefit of malignant growth patients have the option to upsurge prophetic exactitude and limit in the end cause better-educated decisions. To the current end, various amendments smear AI to disease data of the surveillance, epidemiology, and end results database. It may be used to better forecast cancer in the medical sector, and these studies can give a good chance to enhance existing models and build new models for uncommon cancers of minority groups in particular. In this paper, the authors contribute to getting more predicted accuracy for SEER cancer and use it to better forecast cancer in the medical sector.
An Explainable Ai-based Blood Cell Classification Using Optimized Convolutional Neural Network
(Elsevier, 2024-07-02) Islam, Oahidul; Assaduzzaman, Md.; Hasan, Md Zahid
White blood cells (WBCs) are a vital component of the immune system. The efficient and precise classification of WBCs is crucial for medical professionals to diagnose diseases accurately. This study presents an enhanced convolutional neural network (CNN) for detecting blood cells with the help of various image pre-processing techniques. Various image pre-processing techniques, such as padding, thresholding, erosion, dilation, and masking, are utilized to minimize noise and improve feature enhancement. Additionally, performance is further enhanced by experimenting with various architectural structures and hyperparameters to optimize the proposed model. A comparative evaluation is conducted to compare the performance of the proposed model with three transfer learning models, including Inception V3, MobileNetV2, and DenseNet201.The results indicate that the proposed model outperforms existing models, achieving a testing accuracy of 99.12%, precision of 99%, and F1-score of 99%. In addition, We utilized SHAP (Shapley Additive explanations) and LIME (Local Interpretable Model-agnostic Explanations) techniques in our study to improve the interpretability of the proposed model, providing valuable insights into how the model makes decisions. Furthermore, the proposed model has been further explained using the Grad-CAM and Grad-CAM++ techniques, which is a class-discriminative localization approach, to improve trust and transparency. Grad-CAM++ performed slightly better than Grad-CAM in identifying the predicted area's location. Finally, the most efficient model h
An Explainable Artificial Intelligence Model for Multiple Lung Diseases Classification from Chest X-ray Images Using Fine-tuned Transfer Learning
(Elsevier, 2024-07-02) Mahamud, Eram; Fahad, Nafiz; Assaduzzaman, Md.; Zain, S.M.; Goh, Kah Ong Michael; Morol, Md. Kishor
Traditional deep learning models are often considered “black boxes” due to their lack of interpretability, which limits their therapeutic use despite their success in classification tasks. This study aims to improve the interpretability of diagnoses for COVID-19, pneumonia, and tuberculosis from X-ray images using an enhanced DenseNet201 model within a transfer learning framework. We incorporated Explainable Artificial Intelligence (XAI) techniques, including SHAP, LIME, Grad-CAM, and Grad-CAM++, to make the model’s decisions more understandable. To enhance image clarity and detail, we applied preprocessing methods such as Denoising Autoencoder, Contrast Limited Adaptive Histogram Equalization (CLAHE), and Gamma Correction. An ablation study was conducted to identify the optimal parameters for the proposed approach. Our model’s performance was compared with other transfer learning-based models like EfficientNetB0, InceptionV3, and LeNet using evaluation metrics. The model that included data augmentation techniques achieved the best results, with an accuracy of 99.20%, and precision and recall of 99%. This demonstrates the critical role of data augmentation in improving model performance. SHAP and LIME provided significant insights into the model’s decision-making process, while Grad-CAM and Grad-CAM++ highlighted specific image features and regions influencing the model’s classifications. These techniques enhanced transparency and trust in AI-assisted diagnoses. Finally, we developed an Android-based system using the most effective model to support medical specialists in their decision-making process.
Social Media Hate Speech Detection Using Machine Learning Approach
(Springer Nature, 2023-07-22) Haider, Farhatul; Dipty, Ismotara; Rahman, Fiaj; Assaduzzaman, Md.; Sohel, Amir
Humanity has profited enormously from the interchange of information and the expanding use of social media but it has also raised a number of challenges, such as the persistence of hate speech. This growing problem on social media platforms, latterly studies used a different type of point engineering system and machine literacy algorithms to automatically descry hate comments on numerous data. As we know, several studies have been done so far and compared several point engineering strategies with machine literacy algorithms to discover which strategy is the most productive. This investigation aims to examine the performance of multiple engineering approaches with five machine literacy algorithms. The data sets contain the class orders hate speech, not hate speech and offensive comments independently. These social media posts are split into these two groups. To recognize the particular traits of hate speech text messages, the appropriate n-gram feature sets are extracted. The n-gram TF-IDF weights provide the foundation for these feature models. The main aspiration of this research work is to analyze, and resolve the above problem and compare algorithms and features used in machine learning to automatically detect hate speech and specified them like labeling into various classes like hate speech, offensive, and neither, etc. After using different classifiers, “Random Forest” has come up with better accuracy, precision, and recall compared to SVM (Support Vector Machine), Naive Bayes, Logistic Regression, Ada Boost, and Gradient boost algorithms. This system achieved an accuracy of 90.26% using a Random Forest. The experimental result showed that the “Random Forest” provided the best all-around accuracy from the model that has been made and it is more accurate than compare to other work done in recent times on this. So, the result obtain from the model, based on the resulting intensity of the comments can be extracted.