Skin abnormality monitoring systems and methods
A process for assessing image quality metrics includes the steps of receiving image data from an image capture device, wherein the image data includes a skin portion; extracting, via a machine learning system, a first set of information from the image data; extracting, via the machine learning system, a second set of information from the image data; scoring the first and second sets of information of the image data to reflect image quality metrics; identifying a threshold score reflecting acceptable image quality metrics; when the score is greater than the threshold score, submit the image data; and when the score is less than the threshold score, prompt the user to retake the image data.
1. A process for determining one or more outlier lesions, the process comprising the steps of:
receiving a plurality of image data from an image capture device, wherein each image data includes a skin portion, wherein each image data of the plurality of image data comprises a different skin image of the skin portion, the skin portion including a plurality of lesions;
extracting, via a machine learning system, a first set of information associated with each image data in the plurality of image data;
extracting, via the machine learning system, a second set of information associated with each image data in the plurality of image data;
determining a plurality of vector representations for each set of information associated with each image data in the plurality of image data;
determining an average vector based on the plurality of vector representations across the plurality of image data of the skin portion;
calculating a normalized distance metric based on the plurality of vector representations and the average vector; and
analyzing the plurality of lesions to identify one or more outlier lesions within the plurality of lesions based on the normalized distance metric.
2. The process of claim 1 , wherein a subset of the plurality of image data includes a lesion of the plurality of lesions.
3. The process of claim 2 , further comprising the step of cropping each image of the subset around the respective lesion of the image.
4. The process of claim 1 , wherein the plurality of vector representations for each set of information includes 1024 numerical values.