PASI AI Demo Guide
- Go to https://demo.belle.ai and select the AI module.
- Upload a corresponding image. You will have the option to add multiple images (look for the plus sign that will appear after the first image is uploaded).
- Then, the AI will analyze the image.
-
The model will produce an AI-predicted PASI score and clinicians may move the slider bars to create a Clinician Adjusted Score.
-
The Oops Score checks if the image contains human skin. An Oops score of 1 means the image does not contain any skin.
-
The Image Quality Score evaluates image blur and sharpness. The larger the score, the better (0.0 - 1.0).
-
The AI then grades Induration (Thickness), Erythema (Redness), and Desquamation (Scaling) on a scale between 0 and 4. The AI calculates the percentage of affected skin as a percentage of the total skin shown.
-
The computation of PASI score is as follows: bodypart_factor * affected_area * (redness+thickness+scaling). In the example image, it is a local computation of PASI as we do not have information about the body part, so we replace this factor by 6 (which would be the maximum). Then for this example it would be: 6 * 0.22 * (2.8 + 2.7 + 2.5) = 10.56 rounded to 11.
-
You may click on the Segmentation or Heatmap checkboxes to visualize the outputs that the model used to calculate the PASI score. The different color contours depict differing levels of severity based on the pixelwise heatmap; this model generates individual heatmaps for each criterion (redness, thickness, or scaling), enabling the calculation of a severity score specific to each feature.
-
You may add notes or comments under “What do you think about the result?” You may check the box of what condition you believe is present. Your input will not change any scoring.
- You may revisit previous analyses via the History tab (located on the sidebar) which automatically saves previous results. Note: comments on such analyses must be saved manually by clicking the “Save" button.
