Modified Grasshopper Optimization Algorithm (Mgoa) Based Feature Selection And Convolutional Long Short-Term Memory (Conv-Lstm) For Diagnosis Of Thyroid Cancer

Abstract

One prevalent endocrine carcinoma that affects the thyroid gland is Thyroid Cancer (TC). Thyroidectomy is still the principal therapeutic option, despite significant efforts to improve diagnosis. An effective procedure with no unnecessary side effects depends on an accurate preoperative diagnosis. A precise preoperative diagnosis may not always be ensured by the current human evaluation of Thyroid Nodule (TN) malignancy, because this TN maligancy is prone to errors. Medical Data Analysis (MDA) problems can be easily solved with the use of Data Mining (DM) algorithms. To find every problem in the dataset, DM helps with complex DA. With several techniques for classification, clustering, association, etc., DM offers significant assistance with thyroid datasets. Though there is enormous scope for clinical utility in applying Deep Learning (DL) techniques to predict and identify TC, their development and application present a number of difficulties. In order to find critical features and remove irrelevant, redundant, or noisy features, data pre-processing is necessary for DL approaches. This reduces the size of the feature space. Data collection (DC), data pre-processing, feature selection (FS), data classification, and performance evaluation are the five main stages of the suggested task. Initially, the Kaggle online repository is used to gather the DC, TC risk prediction dataset. This dataset mimics real-world TC risk factors and includes 212,691 records with 23 features. Then, issues with Data Encoding (DE), data resampling, data normalisation (DN), and handling missing data are addressed by using data pre-processing techniques. Min-Max normalisation (MMN), also known as Min-max scaling, is used to perform normalisation. Boosting, the Synthetic Minority Over-sampling Technique (SMOTE), and Tomek links (BST) are effective data balancing techniques for addressing data imbalance problems. Finally, the Modified Grasshopper Optimisation Algorithm (MGOA) is used to choose appropriate features. For TC detection, the most pertinent features (an optimal reduced feature subset) are chosen using MGOA. The search space (SS) can be effectively explored and exploited by MGOA, a nature-inspired algorithm (NIA). The TN malignancy is then predicted by classifying the data using Convolutional- (LSTM) Long Short-Term Memory (Conv-LSTM).  This will enhance estimation and makes the parameters easier to understand. The final metrics utilised for assessing efficiency are Accuracy (AC), Specificity (SP), Precision (PR), Recall (R) or Sensitivity (SE), and the Area Under Receiver Operating Characteristic (AUROC) curve. According to the study of the results, the suggested model outperformed the other methods currently in use in terms of AC