APASI
What is APASI?
The Automatic Psoriasis Area and Surface Index (APASI) represents a significant advancement as the automated iteration of the most extensively employed scoring system for evaluating psoriasis severity in clinical trials. However, the complexity, substantial effort, and time demands associated with completing the assessment have led to the prevalent use of PASI in day-to-day practice. An additional constraint lies in the issue of inter-observer variability, which has been reported at 17% in our own retrospective study.
Addressing these challenges, APASI emerges as a transformative solution by harnessing the capabilities of smartphone-captured images. The algorithm adeptly identifies the extent and intensity of visual signs such as erythema, induration, and desquamation, enabling the computation of the PASI score within mere seconds.
Body zones
The PASI defines 4 body zones:
- Head: scalp and neck.
- Upper Extremities: arms, hands, palms.
- Trunk: armpits, torso, and genitals.
- Lower Extremities: buttocks, legs, and feet.
Each body zone contributes a different percentage to the final score: head 10%, upper extremities 20%, trunk 30%, and lower extremities 40%.
Local APASI
The classic PASI is calculated with the patient in consultation, which means it is based on observation by the dermatologist. However, when it comes to clinical images taken with a smartphone, we rely on the photographer and the limitations of the capturing device.
The local APASI is calculated similarly to the PASI, but it is given for a single image. It is calculated using the intensity of the visual signs on a scale from 0 to 4 and the extent of the surface. By utilizing an AI Marker, a color sticker used for calibration and resolution measurement, a Body Surface Area (BSA) formula, and the percentage of the body zone, we can estimate the extent of the surface on that specific body zone.
General APASI
The general APASI is computed by adding multiple photos of the patient into the equation. Only photos of affected areas are required, and the procedure for acquisition is explained in the next section.
When several pictures of the same zone are captured (for example, right and left arms, both part of the upper extremities category), the maximum intensity is the one used for that zone. This follows the definition of the PASI, where a representative area of psoriasis is selected for each body region.
How to take pictures
Ensuring the capture of high-quality photos is paramount to achieving optimal algorithm performance. A comprehensive guide featuring essential tips for image acquisition is available to assist in this endeavor. This section is exclusively dedicated to detailing the photo capture process, with the singular objective of attaining the finest outcomes in APASI computation.
How does APASI work?
APASI utilizes cutting-edge convolutional neural networks that have been trained on a vast dataset comprising thousands of images meticulously annotated by numerous experts.
Estimating Body Surface Area (BSA)
The extent of psoriasis-affected skin is assessed across four distinct body regions. Within each region, the affected area is classified into categories: absent (0), 1-9% (score 1), 10-29% (score 2), 30-49% (score 3), 50-69% (score 4), 70-89% (score 5), or 90-100% (score 6). APASI employs pixel-level detection to ascertain disease extent, leverages the calibration marker to derive surface measurements in square centimeters, and employs a formula for estimating overall body surface area, thereby determining the percentage of the body affected.
Assessing Visual Sign Intensity
The algorithm quantifies the intensity of psoriasis characteristics—redness, thickness, and scaling—based on a scale of absence (0), mild (1), moderate (2), severe (3), or very severe (4). This evaluation enables APASI to comprehensively understand the severity of these visual signs for accurate scoring.
Performance
APASI has undergone thorough evaluation through both retrospective and prospective clinical trials, consistently demonstrating superior performance when compared to individual dermatologists. The ensuing tables present a comprehensive overview of the variability, expressed in terms of Root Mean Absolute Error (RMAE), observed between the specialists and the algorithm:
| Visual sign | RMAE dermatologists | RMAE APASI |
|---|---|---|
| Erythema | 13.32 | 9.97 |
| Induration | 18.62 | 9.81 |
| Desquamation | 17.61 | 11.5 |
| Average | 16.51 | 10.42 |
Request APASI score
{
"requestId": "90925097-820b-403d-a75d-4cd989903df1",
"data": {
"type": "image",
"modality": "clinical",
"operator": "Practitioner",
"bodySite": "ARM_LEFT",
"knownConditionForThisImage": {
"conclusion": "Psoriasis"
},
"previousMedia": [
{
"content": "base64 image",
"date": "2022-02-22T12:16:59+01:00"
}
],
"subject": {
"identifier": "6ec724a0-6fa3-11eb-a15f-0242ac160004",
"gender": "m",
"height": 175,
"weight": 71,
"birthdate": "1986-10-21",
"generalPractitioner": {
"identifier": "44f89a8c-6f8a-11eb-9c8a-0242ac160004"
},
"managingOrganization": {
"identifier": "b13cd636-327b-11ec-86b0-0242ac180004",
"display": "Hospital Central"
}
},
"scoringSystems": ["APASI_LOCAL", "PASI_LOCAL", "DLQI", "PURE4"],
"questionnaireResponse": {
"DLQI": {
"question1": 2,
"question2": 3,
"question3": 1,
"question4": 2,
"question5": 1,
"question6": 2,
"question7": 2,
"question8": 2,
"question9": 3,
"question10": 2
},
"APASI_LOCAL": {
"surface": 1
},
"PASI_LOCAL": {
"surface": 1,
"erythema": 1,
"induration": 2,
"desquamation": 2
},
"PURE4": {
"question1Pure": "0",
"question2Pure": "1",
"question3Pure": "0",
"question4Pure": "0"
}
},
"content": "base64 image"
}
}
Questionnaire for the scoring system
{
"code": "APASI_LOCAL",
"questions": [
{
"code": "surface",
"label": "Affected area",
"answers": [
{ "label": "0", "value": "0" },
{ "label": "0-10%", "value": "1" },
{ "label": "10-30%", "value": "2" },
{ "label": "30-50%", "value": "3" },
{ "label": "50-70%", "value": "4" },
{ "label": "70-90%", "value": "5" },
{ "label": "90-100%", "value": "6" }
],
"openField": false
}
]
}
Response
"evolution": {
"domains": {
"APASI_LOCAL": {
"explainabilityMedia": {
"content": "base 64 image",
"detections": null
},
"facets": {
"desquamation": {
"intensity": 0,
"value": 0
},
"erythema": {
"intensity": 0,
"value": 0
},
"induration": {
"intensity": 0,
"value": 0
},
"surface": {
"intensity": 1,
"value": 10
}
},
"grade": {
"category": "None",
"score": 0.0
}
}
}
}