Pedicle screw placement (PSP) is crucial for managing symptomatic spinal disorders, but its success hinges on accurate placement to avoid pedicle breaches and nerve damage. These breaches can occur in a concerning range of 1.1 up to 29.0 percent of procedures. The ball-tip technique is initially introduced by using a ball-shaped metal tip with a metal semi-flexible shaft. The use of a ball-tip feeler at various checkpoints could assist the surgeon in confirming pedicle wall integrity and re-planning the screw trajectory. While the smart ball-tip feeler improves surgical time and accuracy, it remains heavily reliant on surgeon experience. To tackle this problem, recent studies have explored various sensors to automatically detect the breach during the drilling procedure, such as drilling force or electrical impedance. Preliminary results show promising potential but are limited by cost and complexity. An affordable, user-friendly, and highly accurate breach detection solution would be preferential. Inertial measurement units (IMUs) and load cells are cost-effective options. However, traditional data processing methods can be unreliable. Deep learning, with its ability to identify patterns in complex data, presents a promising alternative for robust breach detection in PSP. This paper proposes a low-cost, deep learning-based technique using the ball-tip feeler to provide breach alerts via audible beeps. The developed system evaluated its performance on synthetic phantoms