Toan Nang Do

Prof. Dr. Toan Nang Do


Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 August 2014 in Annals of Mathematics and Artificial Intelligence
Verified
Journal Article
Closed access
PDF file
TPThe-Anh Pham
The-Anh Pham
HLHong-Ha Le
Hong-Ha Le
TDToan Nang Do
Toan Nang Do

1 January 2023 in Computers, Materials & Continua
Verified
Journal Article
Open access
PDF file
HTHa Manh Toan
Ha Manh Toan
LHLam Thanh Hien
Lam Thanh Hien
NVNgo Duc Vinh
Ngo Duc Vinh
TDToan Nang Do
Toan Nang Do
Deep learning created a sharp rise in the development of autonomous image recognition systems, especially in the case of the medical field. Among lung problems, tuberculosis, caused by a bacterium called Mycobacterium tuberculosis, is a dangerous disease because of its infection and damage. When an infected person coughs or sneezes, tiny droplets can bring pathogens to others through inhaling. Tuberculosis mainly damages the lungs, but it also affects any part of the body. Moreover, during the period of the COVID-19 (coronavirus disease 2019) pandemic, the access to tuberculosis diagnosis and treatment has become more difficult, so early and simple detection of tuberculosis has been more and more important. In our study, we focused on tuberculosis diagnosis by using the chest X-ray image, the essential input for the radiologist’s profession, and researched the effectiveness of the transfer learning approach in the case study of Vietnamese chest X-ray images. We proposed four strategies to clarify our hypothesis in different ways of applying transfer learning and different training set types. We also prepared a Vietnamese X-ray image dataset with the support of the VRPACS team to provide the basis for training and testing deep learning models. Our experiments were carried out by applying three different architectures, Alexnet, Resnet, and Densenet, on international, Vietnamese, and combined X-ray image datasets. After training, all models were verified on a pure Vietnamese X-rays set. The results show that transfer learning is suitable in the case study of Vietnamese chest X-ray images with high evaluating metrics in terms of AUC (Area under the Receiver Operating Characteristic Curve), sensitivity, specificity, and accuracy. In the best strategy, most of the scores were more than 0.93, and all AUCs were more than 0.98.

14 January 2025 in Diagnostics
Verified
Journal Article
Open access
PDF file
LHLam Thanh Hien
Lam Thanh Hien
HTHieu Pham Trung
Hieu Pham Trung
TDToan Nang Do
Toan Nang Do
Introduction: Cancer is a highly lethal disease with a significantly high mortality rate. One of the most commonly used methods for treatment is radiation therapy. However, cancer treatment using radiotherapy is a time-consuming process that requires significant manual work from planners and doctors. In radiation therapy treatment planning, determining the dose distribution for each of the regions of the patient’s body is one of the most difficult and important tasks. Nowadays, artificial intelligence has shown promising results in improving the quality of disease treatment, particularly in cancer radiation therapy. Objectives: The main objective of this study is to build a high-performance deep learning model for predicting radiation therapy doses for cancer and to develop software to easily manipulate and use this model. Materials and Methods: In this paper, we propose a custom 3D convolutional neural network model with a U-Net-based architecture to automatically predict radiation doses during cancer radiation therapy from CT images. To ensure that the predicted doses do not have negative values, which are not valid for radiation doses, a rectified linear unit (ReLU) function is applied to the output to convert negative values to zero. Additionally, a proposed loss function based on a dose–volume histogram is used to train the model, ensuring that the predicted dose concentrations are highly meaningful in terms of radiation therapy. The model is developed using the OpenKBP challenge dataset, which consists of 200, 100, and 40 head and neck cancer patients for training, testing, and validation, respectively. Before the training phase, preprocessing and augmentation techniques, such as standardization, translation, and flipping, are applied to the training set. During the training phase, a cosine annealing scheduler is applied to update the learning rate. Results and Conclusions: Our model achieved strong performance, with a good DVH score (1.444 Gy) on the test dataset, compared to previous studies and state-of-the-art models. In addition, we developed software to display the dose maps predicted by the proposed model for each 2D slice in order to facilitate usage and observation. These results may help doctors in treating cancer with radiation therapy in terms of both time and effectiveness.

1 January 2024 in Intelligent Automation and Soft Computing
Verified
Journal Article
Open access
PDF file
TDToan Nang Do
Toan Nang Do
LHLam Thanh Hien
Lam Thanh Hien
HTHa Manh Toan
Ha Manh Toan
NVNguyen Trong Vinh
Nguyen Trong Vinh
PHPham Trung Hieu
Pham Trung Hieu
Cancer is one of the most dangerous diseases with high mortality. One of the principal treatments is radiotherapy by using radiation beams to destroy cancer cells and this workflow requires a lot of experience and skill from doctors and technicians. In our study, we focused on the 3D dose prediction problem in radiotherapy by applying the deep-learning approach to computed tomography (CT) images of cancer patients. Medical image data has more complex characteristics than normal image data, and this research aims to explore the effectiveness of data preprocessing and augmentation in the context of the 3D dose prediction problem. We proposed four strategies to clarify our hypothesis in different aspects of applying data preprocessing and augmentation. In strategies, we trained our custom convolutional neural network model which has a structure inspired by the U-net, and residual blocks were also applied to the architecture. The output of the network is added with a rectified linear unit (Re-Lu) function for each pixel to ensure there are no negative values, which are absurd with radiation doses. Our experiments were conducted on the dataset of the Open Knowledge-Based Planning Challenge which was collected from head and neck cancer patients treated with radiation therapy. The results of four strategies show that our hypothesis is rational by evaluating metrics in terms of the Dose-score and the Dose-volume histogram score (DVH-score). In the best training cases, the Dose-score is 3.08 and the DVH-score is 1.78. In addition, we also conducted a comparison with the results of another study in the same context of using the loss function.

8 April 2026 in Computers & Education: X Reality
Verified
Journal Article
Open access
LTLe Son Thai
Le Son Thai
DADuong Minh Anh
Duong Minh Anh
HTHieu Pham Trung
Hieu Pham Trung
TDToan Nang Do
Toan Nang Do

1 February 2026 in IAES International Journal of Artificial Intelligence (IJ-AI)
Verified
Journal Article
Open access
TTTung Nguyen Trong
Tung Nguyen Trong
NVNgo Duc Vinh
Ngo Duc Vinh
HTHa Manh Toan
Ha Manh Toan
TDToan Nang Do
Toan Nang Do
Autism is a serious mental health problem with long-term effects on life. Therefore, early diagnosis is a topical issue for effective treatment. This study proposes a novel facial landmark transformation-based data augmentation method that allows for the generation of geometric transformations related to facial geometry. This method increases the generalizability and provides a perspective on the role of facial regions in autism detection. The proposed augmentation method ensures the generation of variants that are consistent with the facial image structure and the nature of the facial image. Next, conduct a comprehensive and comparative study with EfficientNet-B0, EfficientNet-B4, ResNet-18, ResNet-50, ResNet-101, MobileNet-V2, DenseNet-121 and DenseNet-201. Also analyze the model's attention over the main regions of the face that are related to facial landmarks. The results clearly show that the models trained with the proposed method outperform the default augmentation method. Specifically, when averaging the measures across the tested models, the results are 0.905417 for accuracy, 0.962133 for area under the curve (AUC), 0.9198 for precision, 0.888333 for recall, and 0.903678 for F1-score. Furthermore, when analyzing the gradient-weighted class activation mapping (Grad CAM) heatmaps, the high-value regions are clearly concentrated on the main areas of the face. Source code is published on GitLab platform.