Dr Humera Shaziya
@nizamcollege.ac.in
humerashaziya@nizamcollege.ac.in
Nizam College
Dr. Humera Shaziya has been working as Assistant Professor in the department of Informatics, Nizam College, an autonomous and constituent college of Osmania University since July 2004. She has served as the head of the department from July 2018 to July 2020 and as chairperson board of studies for BCA program from 2019 to 2020. She has been teaching for over 19 years to PG programme. She has been handling courses on algorithms, artificial intelligence, machine learning, deep learning, data science and several other core courses of computer sciences. She has been guiding students for carrying out major projects. Furthermore, she has been mentoring students on their academic and personal aspects. She has delivered 10 extension lectures and seminars in various colleges. Additionally, she has obtained 8 certificates from Coursera platform. She was the member of several committees including IQAC, NAAC criterion committee, Women Empowerment Cell (WEC), and syllabus revision committee.
EDUCATION
She has received M.Sc(IS), M.Tech(CSE) and BCA each with distinction from Osmania University and has qualified UGC-NET and AP-SET eligibility tests for Lecturership in the year 2012. She has been awarded the PhD(CSE) on the topic “Automatic Detection and Classification of Lung Cancer in Pulmonary CT Images using Deep Learning” during March 2023 under Visvesvaraya PhD scheme for Electronics and IT, Ministry of Electronics and IT, Government of India from department of CSE, University College of Engineering, Osmania University.
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Science Applications, Artificial Intelligence, Computer Vision and Pattern Recognition, Computer Science
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
L K Suresh Kumar, Humera Shaziya, and Raniah Zaheer
CRC Press
Humera Shaziya and Shyamala Kattula
IEEE
Lung cancer detection at an early stage would be life saving. Usually it is diagnosed at a later stage which leads to increase in the mortalities. Detection of malignant lung nodules from CT images is a challenging task, given several factors that impact the detection and classification. In this work, we are proposing a convolutional neural network (CNN) based deep learning model that improves the accuracy of the nodules classification into benign and malignant types. Lung imaging database consortium-image database resource initiative (LIDC-IDRI), a publicly available lung CT scans dataset have been chosen for experiments. The proposed method come up with an approach to patchify the image to include the nodules segments of the image thus reducing the size of CT image drastically by extracting the nodule patches. Computational overhead is decreased due to the presented strategy. 6691 images containing both nodules and non-nodules are subsequently loaded into a 4-layered 2D CNN. Apparently two convolutional and two dense layers form the four layered CNN. Twenty filters having size of 5x5 is employed with relu activation function for first convolutional layer and 40 filters with size 3x3 has been specified for the second one. The model has been trained and validated on 70% and 10% respectively and tested on 20% of dataset. The verification performed on evaluation data resulted in 93.58% accuracy, 95.61% sensitivity and 90.14% specificity.
Humera Shaziya and Raniah Zaheer
Springer Singapore
Humera Shaziya and Raniah Zaheer
Springer Singapore
Humera Shaziya and K. Shyamala
IEEE
Image Segmentation performs segregation of distinct segments of an image. Lung segmentation separate different elements of thoracic region. It is an essential prerequisite to several analysis tasks performed on the Computed Tomography (CT) images of lungs. Computational complexity is greatly reduced only when the required area is segregated from the entire CT image. Automated segmentation facilitates quick processing since it requires relatively less time to process more images. Conventional computer based segmentation methods require extensive support for determining the features. Users develop the features and provide to the system which then utilize those features to delineate the required regions. Recent advancements in deep learning showed optimal results in solving numerous image recognition and segmentation problems. The significant characteristic of deep learning is that the model itself learns the features from the input images and then apply the learned features to process new images. The most successful model of deep learning is Convolutional Neural Network (CNN) has outperformed earlier techniques for image recognition, object and face detection and is considered to be the most successful architecture of deep learning. CNN has also been applied for segmentation tasks. In this proposed work, CNN and UNet models have been implemented to evaluate the processing of medical images. The focus of the work is on CT images of lungs. Results obtained on the lungs dataset of 267 images on CNN is 81.34% and UNet is 82.61%. Thus U-Net has improved the dice coefficient by 1.27%. The experiments show that UNet model outperforms CNN model to segment the lung fields in CT images.
Humera Shaziya, K. Shyamala, and Raniah Zaheer
IEEE
Segmentation is the process of partitioning an image into distinctive subsets that share similar characteristics. Segmentation is an important prerequisite to semantic image analysis. Segmentation in general is useful in many different applications such as object and face detection and recognition. Particularly in medical image analysis segmentation plays a vital role in efficient processing of images. Segmentation is used to determine the volume of mass, planning of radiotherapy, and detection of artifacts in various organs. In lung cancer diagnosis, segmentation of lungs is the crucial step. Segmenting lungs from nearby structures significantly reduce the execution time of nodule detection and helps improve its efficiency. Lung segmentation is challenging and difficult task considering the heterogeneous nature of lung fields, closeness in gray level of different soft tissues, anatomical variability, and differences in scanners and scanning protocols and dose of radiation. Various automatic and semi-automatic approaches are presented for lung or nodule segmentation. The proposed study is a review of numerous techniques for lung segmentation. The present work investigated lung segmentation methods starting with conventional methods to machine learning techniques and finally the most remarkable methods of deep learning.
Raniah Zaheer and Humera Shaziya
IEEE
Training the deep learning models involves learning of the parameters to meet the objective function. Typically the objective is to minimize the loss incurred during the learning process. In a supervised mode of learning, a model is given the data samples and their respective outcomes. When a model generates an output, it compares it with the desired output and then takes the difference of generated and desired outputs and then attempts to bring the generated output close to the desired output. This is achieved through optimization algorithms. An optimization algorithm goes through several cycles until convergence to improve the accuracy of the model. There are several types of optimization methods developed to address the challenges associated with the learning process. Six of these have been taken up to be examined in this study to gain insights about their intricacies. The methods investigated are stochastic gradient descent, nesterov momentum, rmsprop, adam, adagrad, adadelta. Four datasets have been selected to perform the experiments which are mnist, fashionmnist, cifar10 and cifar100. The optimal training results obtained for mnist is 1.00 with RMSProp and adam at epoch 200, fashionmnist is 1.00 with rmsprop and adam at epoch 400, cifar10 is 1.00 with rmsprop at epoch 200, cifar100 is 1.00 with adam at epoch 100. The highest testing results are achieved with adam for mnist, fashionmnist, cifar10 and cifar100 are 0.9826, 0.9853, 0.9855, 0.9842 respectively. The analysis of results shows that adam optimization algorithm performs better than others at testing phase and rmsprop and adam at training phase.
Humera Shaziya, K. Shyamala, and Raniah Zaheer
IEEE
Convolutional Neural Network (CNN) or ConvNet is the leading-edge deep learning model that has achieved phenomenal successes in the tasks of image classification, object recognition, speech recognition and natural language processing. ConvNets are inherently complex architectures and training ConvNets requires significant amount of computation. There is a need to determine whether GPU or CPU provide the effective method for implementation of ConvNets. Very few studies have come up with the comparison of the implementation of ConvNets on both CPU and GPU. The present work examines the impact of GPU on the implementation of ConvNets. ConvNet is trained on MNIST dataset to perform classification of handwritten digits. The experiments have been performed on both CPU and GPU and observed that there is a performance improvement of 5 times in terms of training time speedup on GPU. The proposed work also investigates the effect of regularization and the results show that regularization indeed reduces the problem of overfitting.
Humera Shaziya, K. Shyamala, and Raniah Zaheer
IEEE
Lung Cancer is the most perilous cancer. Early detection of the disease can improve survival rate. Automation of detection of lung nodules aid radiologists in quickly and accurately diagnosing the disease. Developing computer aided diagnosis (CADx) systems for lung cancer is a challenging task. Several components make up CADx and one of the most significant components is lung segmentation. Segmentation of lungs is an essential prerequisite to efficiently detect and classify lung nodules. Lung segmentation is the process of segregating lungs region from other tissues in the CT image. Conventional methods for lung segmentation either do not accurately segments normal and abnormal lungs or rely heavily on user generated features for the lungs. Deep learning has outperformed other methods in image processing and computer vision tasks. An architecture called U-Net convolutional network has been proposed and implemented exclusively for the segmentation of biomedical images. In this study U-Net ConvNet has been implemented on lungs dataset to perform lungs segmentation. The lungs dataset consists of 267 CT images of lungs and their corresponding segmentation maps. The accuracy and loss achieved is 0.9678 and 0.0871 respectively. Hence U-Net ConvNet can be used for the segmentation of lungs in CT scans.
Raniah Zaheer and Humera Shaziya
IEEE
Activation functions are important components of Convolutional Neural Networks (CNN) that introduces nonlinearity in the model to compute complex functions. There are different types of activation functions used with CNNs in different applications, however it turns out that an effective activation function yields better results and improves performance of the model. In this study four of the widely used activation functions are chosen to analyze and evaluate to figure out their efficiency in terms of the model's accuracy. Sigmoid, hyperbolic tangent, rectified linear unit (ReLU) and exponential linear unit (ELU) activation functions have been used with most of the successful models. A CNN model has been implemented on the MNIST dataset to perform the analysis task. The experiments have been performed on Nvidia GPU 940MX to accelerate the training and testing of the CNN model. It has been observed that ReLU the most popular activation function performs better than sigmoid and tanh and a recent activation function ELU performs better than ReLU.