Automated Ultrasound Analysis for Deep Vein Thrombosis Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing deep vein thrombosis (DVT) are inefficient due to high false-positive rates from D-dimer blood tests, leading to excessive referrals for ultrasound scans, increased costs, and potential inappropriate prescriptions of anticoagulants, particularly for patients who cannot access specialist radiologists.
Innovation Solution
A system utilizing machine learning algorithms with ultrasound equipment, including handheld devices, to assist non-specialists in performing accurate ultrasound examinations and diagnosing DVT by providing real-time guidance and image analysis, allowing for effective pressure application and blood vessel obstruction determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If D-dimer blood test is used to screen for DVT, then sensitivity for detecting PE risk is improved, but false positive rate increases leading to excessive ultrasound referrals
Solution Approach 1:
The patent introduces an automated ultrasound analysis system as an intermediary between the D-dimer test and specialist interpretation. This system processes ultrasound images automatically to provide objective measurements of vein compressibility and thrombus presence, serving as a mediator that reduces false positives while maintaining sensitivity
Solution Approach 2:
The system enables non-specialist healthcare providers to perform and interpret DVT ultrasound examinations independently through automated image analysis and standardized protocols. This self-service capability eliminates the need for specialist radiologist interpretation while maintaining diagnostic accuracy, thereby reducing false positives and excessive referrals
2Measurement precision
If ultrasound scans are performed by specialist DVT radiologists, then diagnostic accuracy is improved, but workload and costs increase due to high volume of referrals
Solution Approach 1:
The system transfers the diagnostic capability from specialist radiologists to non-specialist providers through automated image analysis. The system performs measurements, generates reports, and provides diagnostic recommendations automatically, enabling primary care providers to conduct DVT screening without specialist involvement while maintaining high diagnostic accuracy
Solution Approach 2:
The patent replaces the mechanical system of specialist radiologist interpretation with an automated computer-based image analysis system. The system uses standardized protocols and automated measurements to assess vein compressibility and detect thrombi, substituting human specialist expertise with algorithmic analysis that maintains accuracy while reducing workload
3Reliability
If D-dimer test returns positive result, then PE risk detection is improved, but patient anxiety and unnecessary anticoagulant prescriptions increase
Solution Approach 1:
The system provides immediate feedback through automated ultrasound analysis that objectively confirms or rules out DVT based on vein compressibility measurements. This rapid feedback loop reduces patient anxiety by providing definitive results quickly, preventing unnecessary anticoagulant prescriptions while maintaining high sensitivity for PE risk detection
Solution Approach 2:
The system enables immediate on-site diagnosis through automated analysis, allowing patients to receive definitive results during the same visit without waiting for specialist appointments. This self-service capability provides closure to patients quickly, reducing anxiety and preventing inappropriate treatment decisions while maintaining diagnostic sensitivity
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system enables non-specialists to perform accurate DVT diagnoses with high reliability, reducing unnecessary tests and prescriptions, thereby decreasing costs and improving patient safety by providing precise diagnostic tools for remote or underserved areas.
Implementation Method 1
an imaging device, such as an ultrasound probe, to transmit and receive real-time images of a patient's blood vessels
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a method and system for diagnosing blood vessel obstruction, such as would occur in deep vein thrombosis (DVT). More specifically, the present invention relates to a software tool for use with sensor hardware to allow users to perform standardised and repeatable testing of patients to assist with diagnosis of deep vein thrombosis or related conditions. The present invention further relates to an apparatus and method for conducting the diagnosis. According to a first aspect of the present invention, there is provided an apparatus for diagnosing whether a blood vessel is obstructed, comprising: an interface for communication with an imaging device which imaging device comprises a transmitter and a receiver,a user interface, and a processor, which processor is programmed to perform the following steps: inform the user, via the user interface, where to place the imaging device on a patient's body, instruct the imaging device to transmit radiation from the probe, receive reflected radiation from the imaging device,interpret the reflected information fully automatically using a learned algorithm, determine from the reflected radiation whether the imaging device is located correctly relative to a blood vessel of the patient,inform the user, if the imaging device is not located correctly, to reposition the imaging device and to repeat the instruct, receive and determine steps until the imaging device is correctly located,inform the user to apply pressure to the patient's blood vessel using the imaging device,instruct the imaging device to transmit further radiation, receive further reflected radiation from the imaging device, and determine from at least the reflected further radiation whether the blood vessel is obstructed. The system may be arranged to work with any other real-time imaging device by retraining and/or relabelling of new images. It may also work with other non-real-time methods, for example CT and MRI. For example, with a pressure cuff: scan uncompressed, scan under compression, evaluate difference. The problem may be simpler and the setup more expensive, but the same pipeline arrangement could still work. In relation to CT and MRI scans a skilled medical professional may still be required to be present doing the scan. The learned algorithm may comprise one or more elements of machine learning, and/or have been developed using machine learning.