Compute system with explainable ai for skin lesions analysis mechanism and method of operation thereof
A method of operation of a compute system includes: segmenting a skin lesion in a patient image; constructing a normalized image by cropping the patient image and adding padding to position the skin lesion at a center of the normalized image, identifying a skin lesion classification, a skin lesion sub-class, and a risk level assessment by analyzing a symmetry axis, a border, color variation, and dermoscopic structures, and generating a skin lesion display including the normalized image, the skin lesion classification, the skin lesion sub-class, and the risk level assessment for displaying on a device.
1 . A method of operation of a compute system comprising:
segmenting a skin lesion in a patient image, which is an uploaded image accepted based on a photo guideline;
constructing a normalized image by cropping the patient image and adding padding to position the skin lesion at a center of the normalized image;
identifying a skin lesion classification, a skin lesion sub-class, and a risk level assessment by analyzing a symmetry axis, a border, color variation, and dermoscopic structures, wherein a potential malignancy is determined by a discrepancy between a convex hull and the border of the skin lesion; and
generating a skin lesion display including the normalized image, the skin lesion classification, the skin lesion sub-class, and the risk level assessment for displaying on a device.
2 . The method as claimed in claim 1 further comprising predicting a medical element by calculating a total loss (EQ2) to establish a confidence of accuracy of the skin lesion classification.
3 . The method as claimed in claim 1 further comprising predicting a medical element including a pigment network, a streak, pigmentation, a regression structure, dots and globules, a blue whitish veil, a vascular structure, or a combination thereof.
4 . The method as claimed in claim 1 further comprising generating a negative grayscale format by replacing pixel intensities based on background intensity (EQ1) to assess asymmetry, the border, the color variation, and the dermoscopic structure of the skin lesion.
5 . The method as claimed in claim 1 further comprising assessing a difference between the border and the convex hull, wherein the assessing includes a highlight showing asymmetry of the skin lesion.
6 . The method as claimed in claim 1 further comprising identifying as malignant the skin lesion based on a low number of symmetry axes.
7 . The method as claimed in claim 1 further comprising identifying as benign the skin lesion based on a low number of colors.
8 . A compute system comprising:
a control circuit, including a processor, configured to:
segment a skin lesion in a patient image, which is an uploaded image accepted based on a photo guideline;
construct a normalized image by cropping the patient image and adding padding to position the skin lesion at the center of the normalized image;
identify a skin lesion classification, a skin lesion sub-class, and a risk level assessment by analysis of a symmetry axis, a border, color variation, and dermoscopic structures, wherein a potential malignancy is determined by a discrepancy between a convex hull and the border of the skin lesion; and
generate a skin lesion display including the normalized image, the skin lesion classification, the skin lesion sub-class, and the risk level assessment for displaying on a device.
9 . The system as claimed in claim 8 wherein the control circuit is configured to predict a medical element by calculating a total loss (EQ2) to establish a confidence of accuracy of the skin lesion classification.
10 . The system as claimed in claim 8 wherein the control circuit is configured to predict a medical element including a pigment network, a streak, pigmentation, a regression structure, dots and globules, a blue whitish veil, a vascular structure, or a combination thereof.
11 . The system as claimed in claim 8 wherein the control circuit is further configured to generate a negative grayscale format by replacing pixel intensities based on background intensity (EQ1) to assess asymmetry, the border, the color variation, and the dermoscopic structure of the skin lesion.
12 . The system as claimed in claim 8 wherein the control circuit is further configured to assess a difference between the border and the convex hull, wherein the assessment includes a highlight showing asymmetry of the skin lesion.
13 . The system as claimed in claim 8 wherein the control circuit is further configured to identify as malignant the skin lesion based on a low number of symmetry axes.
14 . The system as claimed in claim 8 wherein the control circuit is further configured to identify as benign the skin lesion based on a low number of colors.
15 . A non-transitory computer readable medium including instructions for a compute system which, when executed by a processor, cause the processor to execute the instructions comprising:
segmenting a skin lesion in a patient image, which is an uploaded image accepted based on a photo guideline;
constructing a normalized image by cropping the patient image and adding padding to position the skin lesion at a center of the normalized image;
identifying a skin lesion classification, a skin lesion sub-class, and a risk level assessment by analyzing a symmetry axis, a border, color variation, and dermoscopic structures, wherein a potential malignancy is determined by a discrepancy between a convex hull and the border of the skin lesion; and
generating a skin lesion display including the normalized image, the skin lesion classification, the skin lesion sub-class, and the risk level assessment for displaying on a device.
16 . The non-transitory computer readable medium as claimed in claim 15 further comprising predicting a medical element by calculating a total loss (EQ2) to establish a confidence of accuracy of the skin lesion classification.
17 . The non-transitory computer readable medium as claimed in claim 15 further comprising predicting a medical element including a pigment network, a streak, pigmentation, a regression structure, dots and globules, a blue whitish veil, a vascular structure, or a combination thereof.
18 . The non-transitory computer readable medium as claimed in claim 15 further comprising generating a negative grayscale format by replacing pixel intensities based on background intensity (EQ1) to assess asymmetry, the border, the color variation, and the dermoscopic structure of the skin lesion.
19 . The non-transitory computer readable medium as claimed in claim 15 further comprising assessing a difference between the border and the convex hull, wherein the assessing includes a highlight showing asymmetry of the skin lesion.
20 . The non-transitory computer readable medium as claimed in claim 15 further comprising identifying as malignant the skin lesion based on a low number of symmetry axes.