<p>Clinical documents contain a wealth of information in terms of structured and unstructured data presented in diverse formats, varying with medication records, laboratory reports, and discharge summaries. The layouts differ, integrating text, tables, and pictures, rendering them difficult to interpret and analyze automatically. In such cases, the complex structure of these document images necessitates specialized extraction tools adept at information extraction, facilitating search and analysis for disease diagnosis and training AI models with clinical data. This work proposes an end-to-end framework for automatic information extraction from scanned clinical document images. This framework encompasses a set of algorithms collectively called STRIDE (Structure Identification) to identify the structure of tabular data through pattern recognition techniques. A dataset was created for experimental purposes, containing different kinds of clinical document images collected from various patient volunteers of different hospitals. The proposed framework works in three phases: (1) Document Layout Analysis using YoloV3-DLA Deep learning model to segment and identify the clinical document layout (2) Recognizing the structure of each identified region using a pack of algorithms called STRIDE that uses pattern recognition techniques (3) Extraction of data from each cell using appropriate OCRs and storing it as formatted text (JSON Format). Each phase was independently evaluated, and the experimental results show that the proposed framework performs better than the current state-of-the-art techniques in information extraction as per the standard evaluation metrics. The YoloV3-DLA model attained F1 score of 96.16% (IoU= 0.5) for document layout analysis, and the proposed STRIDE algorithms achieved an F1 score of 96.69% for table structure recognition.</p>