A Systematic Kidney Tumour Segmentation and Classification Framework Using Adaptive and Attentive-Based Deep Learning Networks With Improved Crayfish Optimization Algorithm

Kidney illness constitutes a category of many serious persistent diseases that can affect an individual. Early diagnosis of this condition is critical for effective therapy. Kidney tumours are the 2nd most common type of urological tumour. They come in a variety of forms, the majority of which are cancerous. In comparison to the laborious and lengthy conventional evaluation, deep learning’s autonomous detection techniques may reduce diagnostic time, enhance the precision of tests, lower expenses, and minimize the radiologist’s burden. It is difficult for clinicians to distinguish kidney cancers from renal Computerized Tomography (CT) images. During an operation, the precise division of kidney tumours can assist physicians in determining tumour intricacy and severity. However, due to their variety, segmenting renal tumours mechanically might be challenging. Therefore, an intellectual kidney tumour segmentation and classification model is implemented to recognize benign and malignant tumours at an early stage. To execute this procedure, the input CT images are gathered from standard websites. Then these images are given to the proposed 3D-Trans-Residual DenseUnet++ (3D-TR-DUnet++) network for the segmentation process. With the help of the segmentation process, doctors can identify the normal and abnormal regions in the kidney. The segmented images are then preceded by the classification stage. To classify kidney tumours, a deep learning-based method called Adaptive and Attentive Residual Densenet with Gated Recurrent Unit (AA-RD-GRU) is developed. Here, the parameters from this network are optimized via the recommended Modified Crayfish Optimization Algorithm (MCOA). The precise segmentation and classification of tumours in the kidney help to provide better treatments at the correct time. The segmentation and classification results are contrasted with other deep learning networks as well as various optimization algorithms.