Drop-in gamma probes are widely used in robotic-assisted minimally invasive surgery (RAMIS) for lymph node detection. However, these devices only provide audio feedback on signal intensity and lack the visual feedback necessary for precise localization. Previous work attempted to predict the sensing area location using laparoscopic images, but the prediction accuracy was unsatisfactory. We propose a three-branch deep learning framework that leverages stereo laparoscopic images as input to a Nested ResNet main branch, with depth estimation (via transfer learning) and orientation guidance (through probe axis sampling) as additional guidance. Our approach has been evaluated on a publicly available dataset, demonstrating superior performance over previous methods. Quantitatively, our method yields a 22.10% decrease in 2D mean error and a 41.67% reduction in 3D mean error. Qualitative analyses further confirm the enhanced precision of our solution. This advancement enables real-time visual feedback during gamma probe use in RAMIS, improving surgical localization accuracy and reliability. Our code is available at: https://github.com/Songyu-Xu/Nested-ResNet .