Surgical Video Energy Tool Activation Detection for Post-Op Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surgical video analysis systems lack the ability to accurately detect and quantify energy tool activations without relying on internal data logs, which are often incomplete, inaccessible, or prone to timing errors.
Innovation Solution
Develop a deep-learning-based energy tool activation detection model that processes surgical videos to identify activation events, duration, and count, using a high-quality training dataset prepared through two-level annotation and labeling, and applies sequence sampling windows to enhance detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If internal data logging systems are used to record energy tool usage data, then data collection is automated, but the data becomes inaccessible, incomplete, or prone to timing errors
Solution Approach 1:
The patent introduces an intermediary system consisting of computer vision models and video analysis algorithms that mediate between the surgical video feed and the energy tool usage data. This intermediary approach bypasses the unreliable internal data logging system by independently analyzing video frames to detect tool activations, thereby resolving the contradiction between automated collection and data reliability
Solution Approach 2:
The system creates a virtual copy of the energy tool usage data by analyzing video footage rather than relying on the physical data logs from the device. This copying approach generates alternative data representations (activation events, timestamps, durations) from visual information, ensuring accessibility and accuracy without depending on the original logging system
2Measurement precision
If deep learning models are trained with high-quality annotated datasets, then detection accuracy improves, but data preparation time and complexity increase
Solution Approach 1:
The patent applies preliminary action by performing manual annotation and labeling of training data before model training. Experts annotate surgical videos with ground truth energy tool activation events in advance, creating a prepared training dataset that enables the deep learning model to learn accurate detection patterns without requiring complex real-time processing during surgery
Solution Approach 2:
The data preparation process is segmented into distinct phases: video collection, expert annotation, labeling, validation, and model training. This segmentation allows each step to be optimized independently and facilitates quality control at each stage, managing overall complexity while maintaining high detection accuracy
Data Source
AI summary
Embodiments described in this disclosure include a process for collecting energy tool usage data from surgical videos and using such data for post surgery analysis. The process can begin by receiving a plurality of surgical videos of a surgical procedure involving an energy tool. For each surgical video in the plurality of surgical videos, the process detects a set of activation events in the surgical video, wherein each detected activation event includes an identified starting timestamp and a duration. The process further extracts a set of energy tool usage data based on the set of detected activation events, and then stores the extracted set of energy tool usage data in a database indexed based on a set of energy tool usage metrics. Next, in response to a user search request, the process returns the stored energy tool usage data that matches the search request from the database.