Electrocardiogram (ECG) analysis is a critical aspect of cardiovascular health assessment, yet the manual processing of paper-based ECGs remains a bottleneck in healthcare systems. Our team, Berru's Widows, proposes a novel approach to address this challenge through the digitization and classification of 12-lead ECG images. In this work, we present our methodology and initial results from the unofficial phase, along with our proposed enhancements for the official phase. Our methodology begins with a comprehensive pre-processing stage, encompassing grayscaling, noise reduction, and scaling techniques to prepare the images for analysis. We then employ a U-Net architecture for grid removal, adapting it to operate on patches of the original image. This phase aims to produce gridless versions of the ECGs, facilitating subsequent analysis. For the recognition of Regions of Interest (ROI), we developed a rule-based algorithm that assumes a specific layout and distribution of leads. However, we acknowledge the need to enhance this algorithm to accommodate diverse distributions. The digitization process involves extracting signal information from the ROI, followed by scaling to align with standard ECG grid parameters. In our preliminary evaluation, our method yielded promising but unsatisfactory results, with an F-measure score of 0.00 for classification and an SNR of 0.00 for signal reconstruction. To address these limitations, we propose several enhancements. Firstly, we aim to bolster the robustness of our pre-processing stage by augmenting the dataset and implementing noise reduction post-inference. Additionally, we plan to enhance the ROI recognition algorithm to accommodate varying distributions and incorporate dynamic sizing into the digitization process to adapt to different grid sizes. Our work represents a significant step towards automating ECG analysis, offering the potential to streamline healthcare workflows and improve patient care. In the official phase, we aim to implement these enhancements, with the ultimate goal of achieving clinically viable performance metrics.