3D Anatomy Scan Reformatting With Deep-Learning Landmark Masks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current clinical imaging techniques, such as MRI and CT, require repetitive scanning and manual reformatting to obtain contiguous views of anatomical landmarks, which is time-consuming and prone to errors, especially when technicians lack familiarity with the landmarks.
Innovation Solution
A system employing a pre-trained neural network model to generate deep-learning estimated scan prescription masks, allowing for automated alignment and reformatting of 3D image data to provide continuous views of anatomical landmarks without the need for repeated scans, using low-resolution calibration images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If repetitive scanning is performed to obtain contiguous views of anatomical landmarks, then image quality and completeness are improved, but scan time increases significantly
Solution Approach 1:
The system performs preliminary action by acquiring a single 3D volumetric dataset that contains all anatomical landmarks, then uses deep learning to automatically generate reformatted views of multiple landmarks from this single acquisition. This eliminates the need for repetitive scanning while maintaining the ability to obtain contiguous views of all required anatomical structures.
Solution Approach 2:
The single 3D volumetric acquisition serves multiple functions simultaneously - it captures all anatomical landmarks needed for different clinical assessments in one scan, rather than requiring separate scans for each landmark. The deep learning system then extracts multiple reformatted views from this universal dataset.
2Adaptability or versatility
If manual reformatting is performed by technicians to obtain views of anatomical landmarks, then flexibility and adaptability are improved, but time consumption and error rate increase
Solution Approach 1:
The system performs self-service by automatically identifying anatomical landmarks and generating reformatted views without requiring technician intervention. The deep learning model autonomously processes the 3D volumetric data to produce clinically useful images, eliminating manual reformatting while maintaining adaptability to different landmark requirements.
Solution Approach 2:
The manual mechanical process of technician-operated reformatting is replaced with an automated computational system using deep learning algorithms. This substitution maintains the flexibility to generate various landmark views while dramatically reducing time consumption and eliminating human error.
3Ease of operation
If graphical prescription methods are used to align scans with anatomical landmarks, then ease of operation is improved, but accuracy and reliability deteriorate due to small errors
Solution Approach 1:
The manual graphical prescription method is replaced with automated deep learning-based landmark identification and alignment. This substitution maintains ease of operation as the system is user-friendly while dramatically improving accuracy by eliminating human error in angle and position specification.
Solution Approach 2:
The system incorporates feedback mechanisms where the deep learning model continuously refines its identification of anatomical landmarks and optimization of scan parameters based on the acquired 3D data, ensuring high accuracy in aligning reformatted views with actual anatomical structures.
Data Source
AI summary
Techniques are described for generating reformatted views of a three-dimensional (3D) anatomy scan using deep-learning estimated scan prescription masks. According to an embodiment, a system is provided that comprises a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory. The computer executable components comprise a mask generation component that employs a pre-trained neural network model to generate masks for different anatomical landmarks depicted in one or more calibration images captured of an anatomical region of a patient. The computer executable components further comprise a reformatting component that reformats 3D image data captured of the anatomical region of the patient using the masks to generate different representations of the 3D image data that correspond to the different anatomical landmarks.