Brain MRI detection and classification: Harnessing convolutional neural networks and multi-level thresholding Rasool Reddy Kamireddy, Rajesh N. V. P. S. Kandala, Ravindra Dhuli, Srinivasu Polinati, Kamesh Sonti, et al. Plos One, 2024 Brain tumor detection in clinical applications is a complex and challenging task due to the intricate structures of the human brain. Magnetic Resonance (MR) imaging is widely preferred for this purpose because of its ability to provide detailed images of soft brain tissues, including brain tissue, cerebrospinal fluid, and blood vessels. However, accurately detecting brain tumors from MR images remains an open problem for researchers due to the variations in tumor characteristics such as intensity, texture, size, shape, and location. To address these issues, we propose a method that combines multi-level thresholding and Convolutional Neural Networks (CNN). Initially, we enhance the contrast of brain MR images using intensity transformations, which highlight the infected regions in the images. Then, we use the suggested CNN architecture to classify the enhanced MR images into normal and abnormal categories. Finally, we employ multi-level thresholding based on Tsallis entropy (TE) and differential evolution (DE) to detect tumor region(s) from the abnormal images. To refine the results, we apply morphological operations to minimize distortions caused by thresholding. The proposed method is evaluated using the widely used Harvard Medical School (HMS) dataset, and the results demonstrate promising performance with 99.5% classification accuracy and 92.84% dice similarity coefficient. Our approach outperforms existing state-of-the-art methods in brain tumor detection and automated disease diagnosis from MR images.
Diagnosis of glaucoma from retinal fundus images using disc localization and sequential DNN model Kamesh Sonti, Ravindra Dhuli International Journal of Imaging Systems and Technology, 2024 Deep learning is an emerging trend with enormous applications over the past few years. Ophthalmology is one such area in medical applications where early disease detection is required to avoid loss of vision. Glaucoma is a rapidly growing disorder related to human eye, which arises due to the increase in pressure inside the eye. The medical diagnosis methods available for glaucoma have some limitations; hence, computer‐aided design (CAD) approach is preferred using images. In the context of image processing, convolution neural networks (CNNs) are preferred for classification because of their ability to grasp highly discriminate features from raw pixel intensities. In our approach, diagnosis of glaucoma is implemented by extracting the region of interest (ROI) by splitting the coefficients into recurrence decays and will improve the possibility of identifying even poorly differentiated exudates and upgrading the normal recurrence ranges. Later, a sequential deep neural network (DNN) model with a rectified linear unit (ReLU) and sigmoid function is designed to train the data with effective features matching from training and testing samples. The proposed model is implemented on two publicly available datasets (Drishti‐GS1 and ACRIMA) using 10‐fold cross validation (CV), 60:40 and 70:30 split ratio approaches, and performance is assessed using the metrics and plotted the region of convergence curves. The model is also tested on two more datasets (ORIGA and Refuge) to validate the robustness of the proposed model. The obtained simulation results and the evaluated performance metrics prove that our proposed model diagnose glaucoma from retinal fundus images effectively compared with other existing models.
