scispace - formally typeset
Search or ask a question
Author

Parvez Ahmad

Bio: Parvez Ahmad is an academic researcher from Huazhong University of Science and Technology. The author has contributed to research in topics: Convolutional neural network & Segmentation. The author has an hindex of 6, co-authored 21 publications receiving 133 citations. Previous affiliations of Parvez Ahmad include Indian Institute of Technology Delhi & Shenzhen University.

Papers
More filters
Journal ArticleDOI
TL;DR: Various Data Mining techniques such as classification, clustering, association, regression in health domain are reviewed and applications, challenges and future work of Data Mining in healthcare are highlighted.
Abstract: Data mining is gaining popularity in disparate research fields due to its boundless applications and approaches to mine the data in an appropriate manner. Owing to the changes, the current world acquiring, it is one of the optimal approach for approximating the nearby future consequences. Along with advanced researches in healthcare monstrous of data are available, but the main difficulty is how to cultivate the existing information into a useful practices. To unfold this hurdle the concept of data mining is the best suited. Data mining have a great potential to enable healthcare systems to use data more efficiently and effectively. Hence, it improves care and reduces costs. This paper reviews various Data Mining techniques such as classification, clustering, association, regression in health domain. It also highlights applications, challenges and future work of Data Mining in healthcare.

78 citations

Journal ArticleDOI
TL;DR: A novel combined architecture of dense connection, residual connection, and inception module is proposed that contains three stages, namely the densely connected stage, a residual inception stage, and an up-sampling stage to extract the volumetric contextual information of medical data.

43 citations

Journal ArticleDOI
TL;DR: In this paper, a hierarchical block is introduced between the encoder-decoder for acquiring and merging features to extract multi-scale information in the proposed architecture, which achieves state-of-the-art performance on four challenging MICCAI datasets.
Abstract: UNet and its variations achieve state-of-the-art performances in medical image segmentation. In end-to-end learning, the training with high-resolution medical images achieves higher accuracy for medical image segmentation. However, the network depth, a massive number of parameters, and low receptive fields are issues in developing deep architecture. Moreover, the lack of multi-scale contextual information degrades the segmentation performance due to the different sizes and shapes of regions of interest. The extraction and aggregation of multi-scale features play an important role in improving medical image segmentation performance. This paper introduces the MH UNet, a multi-scale hierarchical-based architecture for medical image segmentation that addresses the challenges of heterogeneous organ segmentation. To reduce the training parameters and increase efficient gradient flow, we implement densely connected blocks. Residual-Inception blocks are used to obtain full contextual information. A hierarchical block is introduced between the encoder-decoder for acquiring and merging features to extract multi-scale information in the proposed architecture. We implement and validate our proposed architecture on four challenging MICCAI datasets. Our proposed approach achieves state-of-the-art performance on the BraTS 2018, 2019, and 2020 Magnetic Resonance Imaging (MRI) validation datasets. Our approach is 14.05 times lighter than the best method of BraTS 2018. In the meantime, our proposed approach has 2.2 times fewer training parameters than the top 3D approach on the ISLES 2018 Computed Tomographic Perfusion (CTP) testing dataset. MH UNet is available at https://github.com/parvezamu/MHUnet .

25 citations

Proceedings ArticleDOI
01 Sep 2018
TL;DR: 3D hyper-dense Convolutional Neural Network (Cnn) is developed to segment tumors, in which it captures the global and local contextual information from two scales ofglobal and local patches along with theTwo scales of receptive field.
Abstract: Glioma is one of the most widespread and intense forms of primary brain tumors. Accurate subcortical brain segmentation is essential in the evaluation of gliomas which helps to monitor the growth of gliomas and assists in the assessment of medication effects. Manual segmentation is needed a lot of human resources on Magnetic Resonance Imaging (MRI) data. Deep learning methods have become a powerful tool to learn features automatically in medical imaging applications including brain tissue segmentation, liver segmentation, and brain tumor segmentation. The shape of gliomas, structure, and location are different among individual patients, and it is a challenge to developing a model. In this paper, 3D hyper-dense Convolutional Neural Network(Cnn)is developed to segment tumors, in which it captures the global and local contextual information from two scales of global and local patches along with the two scales of receptive field. Densely connected blocks are used to exploit the benefit of a CNN to boost the model segmentation performance in Enhancing Tumor (ET), Non-Enhancing Tumor (NET), and Peritumoral Edema (PE). This dense architecture adopts 3D Fully Convolutional Network (FCN) architecture that is used for end-to-end volumetric prediction. The dense connectivity can offer a chance of deep supervision and improve gradient flow information in the learning process. The network is trained hierarchically based on global and local patches. In this scenario, the both patches are processed in their separate path, and dense connections happen not only between same path layers but also between different path layers. Our approach is validated on the BraTS 2018 dataset with the dice-score of 0.87, 0.81 and 0.84 for the complete tumor, enhancing tumor, and tumor core respectively. These outcomes are very close to the reported state-of-the-art results, and our approach is preferable to present 3D-based approaches when it comes to compactness, time and parameter efficiency on MRI brain tumor segmentation.

21 citations

Journal ArticleDOI
TL;DR: Zhang et al. as discussed by the authors proposed an encoder-decoder-based CNN for skin lesion segmentation, based on the widely used UNet architecture, which exploited the benefit of combining densely connected network (DenseNet) and residual network (ResNet)-based encoderdecoder architectures.
Abstract: Melanoma is one kind of dangerous cancer that has been increasing rapidly in the world. Initial diagnosis is essential to survival, but often the disease is diagnosed in the fatal stage. The rapid growth of skin cancers raises a huge demand for accurate automatic skin lesion segmentation. While deep learning techniques, i.e., convolutional neural network (CNN), have been widely used for precise segmentation, the existing densely connected network (DenseNet) and residual network (ResNet)–based encoder-decoder architectures used non-biomedical features for skin lesion tasks. The complexity of tuned parameters, small information in the pre-trained features, and the lack of multi-scale information degrade the performance of skin lesion segmentation. To address these issues, we present encoder-decoder–based CNN for skin lesion segmentation, based on the widely used UNet architecture. We exploit the benefit of combining DenseNet and ResNet to improve the performance of skin lesion segmentation. In the encoder path, atrous spatial pyramid pooling (ASPP) is used to generate multi-scale features from different dilation rates. We used dense skip connection to combine the feature maps of both encoder and decoder paths. We evaluate our approach on ISIC 2018 dataset and achieve competitive performance as compared to other state-of-the-art approaches. Compared to the previous UNet approaches, our method gains a high Jaccard index, Dice, accuracy, and sensitivity. We think that this progress is mainly due to the combined architecture of DenseNet, ResNet, ASPP, and dense skip connection that preserve the contextual information in the encoder-decoder paths. We utilized the combined benefits of both recent DenseNet and ResNet architectures. We used ASPP to exploit multi-scale contextual information by adopting multiple dilation rates. We also implemented dense skip connections for better recovery of fine-grained information of target objects. In the future, we believe that this approach will be helpful to other medical image segmentation tasks.

20 citations


Cited by
More filters
01 Jan 2002

9,314 citations

Journal ArticleDOI
TL;DR: A literature review of the usage of process mining in healthcare and the most commonly used categories and emerging topics have been identified, as well as future trends, such as enhancing Hospital Information Systems to become process-aware.

453 citations

Journal ArticleDOI
TL;DR: This work constructs a Deep Recurrent Q-Network model which is a Recurrent Neural Network deep learning algorithm over the Q-Learning reinforcement learning algorithm to forecast the crop yield, outperforming existing models by preserving the original data distribution.
Abstract: Predicting crop yield based on the environmental, soil, water and crop parameters has been a potential research topic. Deep-learning-based models are broadly used to extract significant crop features for prediction. Though these methods could resolve the yield prediction problem there exist the following inadequacies: Unable to create a direct non-linear or linear mapping between the raw data and crop yield values; and the performance of those models highly relies on the quality of the extracted features. Deep reinforcement learning provides direction and motivation for the aforementioned shortcomings. Combining the intelligence of reinforcement learning and deep learning, deep reinforcement learning builds a complete crop yield prediction framework that can map the raw data to the crop prediction values. The proposed work constructs a Deep Recurrent Q-Network model which is a Recurrent Neural Network deep learning algorithm over the Q-Learning reinforcement learning algorithm to forecast the crop yield. The sequentially stacked layers of Recurrent Neural network is fed by the data parameters. The Q- learning network constructs a crop yield prediction environment based on the input parameters. A linear layer maps the Recurrent Neural Network output values to the Q-values. The reinforcement learning agent incorporates a combination of parametric features with the threshold that assist in predicting crop yield. Finally, the agent receives an aggregate score for the actions performed by minimizing the error and maximizing the forecast accuracy. The proposed model efficiently predicts the crop yield outperforming existing models by preserving the original data distribution with an accuracy of 93.7%.

126 citations

Journal ArticleDOI
TL;DR: This paper is specific to reviewing upcoding fraud analysis and detection research providing an overview of healthcare, upc coding, and a review of the current data mining techniques used therein.
Abstract: From its infancy in the 1910s, healthcare group insurance continues to increase, creating a consistently rising burden on the government and taxpayers. The growing number of people enrolled in healthcare programs such as Medicare, along with the enormous volume of money in the healthcare industry, increases the appeal for and risk of fraudulent activities. One such fraud, known as upcoding, is a means by which a provider can obtain additional reimbursement by coding a certain provided service as a more expensive service than what was actually performed. With the proliferation of data mining techniques and the recent and continued availability of public healthcare data, the application of these techniques towards fraud detection, using this increasing cache of data, has the potential to greatly reduce healthcare costs through a more robust detection of upcoding fraud. Presently, there is a sizable body of healthcare fraud detection research available but upcoding fraud studies are limited. Audit data can be difficult to obtain, limiting the usefulness of supervised learning; therefore, other data mining techniques, such as unsupervised learning, must be explored using mostly unlabeled records in order to detect upcoding fraud. This paper is specific to reviewing upcoding fraud analysis and detection research providing an overview of healthcare, upcoding, and a review of the current data mining techniques used therein.

71 citations

Journal ArticleDOI
TL;DR: A novel combined architecture of dense connection, residual connection, and inception module is proposed that contains three stages, namely the densely connected stage, a residual inception stage, and an up-sampling stage to extract the volumetric contextual information of medical data.

43 citations