Personalized Body Surface Diagnosis Using CNN Baseline Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated solutions for assessing medical conditions on the human body surface are limited in their ability to handle diverse conditions across the entire body surface, including skin and cavities like the ear and mouth, and often rely on parameter estimation or training from fixed generic data, failing to account for variations in image and sound data from specific devices and individuals.
Innovation Solution
A system and method for assessing medical conditions using image and sound data from the human body surface, which involves guiding users to acquire images from specific locations, selecting optimal images, and utilizing convolutional neural networks to generate primary and secondary classification vectors. The system maintains a normal model of classification output vectors for healthy body surfaces and provides personalized diagnoses based on the Mahalanobis distance of potentially abnormal output vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If automated solutions use fixed generic data for training, then the system can be implemented with standard data, but it fails to account for variations in image and sound data from specific devices and individuals
Solution Approach 1:
The system performs preliminary acquisition and storage of normal baseline data for each individual before diagnosis is needed. This preliminary action creates a personalized reference model that accounts for individual variations, allowing accurate detection of abnormalities without requiring extensive training data for each specific individual.
Solution Approach 2:
The system uses each individual's own normal baseline data to create their personalized reference model, allowing the system to adapt to individual variations using the individual's own characteristics. This self-service approach eliminates the need for extensive external training data for each individual.
2Reliability
If manual assessment is performed, then individual variations can be accounted for, but the process is time consuming and susceptible to human errors
Solution Approach 1:
The system replaces manual visual and auditory assessment with automated image and sound processing using convolutional neural networks. This substitution maintains the ability to account for individual variations through personalized baseline models while eliminating human errors and reducing assessment time significantly.
Solution Approach 2:
The system performs automated self-assessment by comparing current images and sounds against the individual's own normal baseline data, eliminating the need for manual assessment while maintaining high reliability through personalized comparison.
3Productivity
If existing automated solutions are used, then assessment speed is improved, but they cannot handle diverse conditions across the entire body surface including skin and cavities
Solution Approach 1:
The system is designed to handle diverse conditions across the entire body surface including skin, throat, and ear by using a unified approach of guided image and sound acquisition followed by automated processing. The system can assess multiple body locations and condition types using the same fundamental methodology, achieving both speed and versatility.
Solution Approach 2:
The system segments the assessment process into guided acquisition at specific locations of interest, automated processing through neural networks, and personalized comparison against baseline data. This segmentation allows efficient handling of diverse conditions at different body locations while maintaining high assessment speed through automation.
Data Source
AI summary
A system and method for personalized diagnosis of human body surface conditions from images acquired from a mobile camera device. In an embodiment, the system and method is used to diagnose skin, throat and ear conditions from photographs. A system and method are provided for data acquisition based on visual target overlays that minimize image variability due to camera pose at locations of interest over the body surface, including a method for selecting a key image frame from an acquired input video. The method and apparatus may involve the use of a processor circuit, for example an application server, for automatically updating a visual map of the human body with image data. A hierarchical classification system is proposed based on generic deep convolution neural network (CNN) classifiers that are trained to predict primary and secondary diagnoses from labelled training images. Healthy input data are used to model the CNN classifier output variability in terms of a normal model specific to individual subjects and body surface locations of interest. Personalized diagnosis is achieved by comparing CNN classifier outputs from new image data acquired from a subject with a potentially abnormal condition to the healthy normal model for the same specific subject and location of interest.