User interface
For your users to interact with the device, it's very likely that you may need to develop a user interface (UI) that allows them to upload images, view reports, and interact with the device's features. The UI should be designed to be user-friendly and intuitive, allowing users to easily navigate through the various functionalities of the device.
This section describes the elements that can be displayed in the user interface. It also explains the purpose of each element and how it can be used to help HCPs make decisions about the next steps in a patient's care.
The user interface can be divided into two main sections:
- the instructions for capturing images
- the report generated by the device.
Instructions
In this section, we describe the instructions that can be displayed to users to help them take good images of skin lesions. The instructions are designed to be easy to understand and follow, ensuring that users can capture images that the device can analyse effectively.
Select problem type
Instructions are different for different types of conditions. For example, the instructions for localised lesions are different from the instructions for other types of conditions.
For that reason, the instructions can be divided into two sections: one for localised lesions and another for other types of conditions. This can be done by clicking on a button or a link that takes them to the appropriate section of the instructions.
Type of issue
Select the type of issue
Localized Lesion
For example, a mole or a pigmentation.
Something else
Any problem that is not on a fixed spot.
Instructions for localised lesions
Localised lesions, such as moles, require close-up images where the lesion is the main element in the image. The following instructions are recommended for taking images of localised lesions. Users should follow these instructions to ensure that the device can analyse the images effectively.
You can include both the image and the text in the instructions. This way, the user can see both the image and the text at the same time.
Image | Text |
---|---|
Capture the photo with your mobile phone | |
Center the photo in the lesion. | |
You can take up to 3 images of the same lesion | |
Crop the image |
The device also admits images taken with a dermatoscope. In this case, the instructions are similar to the ones for localised lesions, but they are not really necessary because the dermatoscope is already designed to take close-up and standardised images of skin lesions.
Instructions for other types of conditions
For other types of conditions, such as skin infections, inflammatory skin diseases, or rashes, the instructions are different. These conditions have a lot of variability, so it is difficult to provide a single set of instructions that will work for all cases.
Report
Metadata
This information can be displayed to all users who need to understand the context of the report. It contains basic information such as the time the report was created, the time the device took to process the image, the image modality and the image quality.
Report information
Information about the report.
- TimestampFeb 01, 2024, 10:06 PM
- Analysis performed in0.65 seconds
- Image modalityDermatoscopic
- Visual image quality82%
- Sensitivity92.22%
- Specificity89.88%
Explanation
- Visual image quality: Parameter that describes focus, lighting, resolution and other qualities that AI needs to function properly.
- Image modality: The type of image that was used to create the report, such as dermoscopy, or clinical.
- Analysis performed in: The time it took for the device to process the image and generate the report.
- Sensitivity: Probability that a sick patient will have a positive result.
- Specificity: Probability that a healthy patient will have a negative result.
Images
Since the image is the most important part of the report, it is typically displayed in a dashboard or a summary view for all users.
Original Image
The original image is typically displayed in a dashboard or a summary view for all users. It is usually the first thing that a user sees when they open a report. It is a good way to give the user a quick overview of the patient's skin lesion and help them understand the context of the report.
Image with annotations
The image with annotations as output from the device helps the dermatologist understand the location of the skin lesion and the areas of interest. It is typically displayed in a dashboard or a summary view for dermatologists. It is focused on helping the dermatologist make a decision about the possible diagnosis of a skin lesion. It also helps superimpose the predictions on the image.
Top-5 predictions
The top-5 predictions are typically displayed in a dashboard or a summary view. It is focused on helping the dermatologist make a decision about the possible diagnosis of a skin lesion.
Select a condition
The algorithm has detected the following conditions.
- Condition A67.51%
- Condition B26.24%
- Condition C1.19%
- Condition D0.99%
- Condition E0.53%
Select other condition
The rest of the classes that do not appear in the top-5 predictions can be found in the Select other condition
button or a similar function. However, showing the top-5 predictions is a good way to give the dermatologist a quick overview of the most likely diagnoses.
Entropy
The entropy is a measure of the uncertainty of the predictions. It is typically displayed in the report next to the top-5 predictions, to help the dermatologist understand how confident the device is about its predictions.
The entropy of this report is 86%. This means that the algorithm is not confident in its predictions.
You may show this message if the entropy is above a certain threshold. This threshold can be set by the organisation, but a good starting point is 0.5. This means that if the entropy is above 0.5, the device is not very confident about its predictions and the dermatologist should be careful when interpreting the results.
Quick insights
This information is typically displayed in a dashboard or a summary view for primary care physicians who need to quickly understand the most important information in a report. It is focused on the most important information that a primary care physician needs to know to make a decision about the next steps in a patient's care.
The pathology Nevus has a probability of around 67.51%. The suspicion of malignancy is 26.46%.
Insights
- Refer to specialistMeets referral criteria
- Condition AIt could also be Condition B.
- Probably malignant or pre-malignant62% suspected malignancy
Malignancy suspicion
This information helps dermatologists assess the risk of a lesion being malignant. It is typically displayed in a dashboard or a summary view for dermatologists. It is focused on helping the dermatologist make a decision about the possible diagnosis of a skin lesion.
Suspicion of malignancy
62%
Predicted by the algorithm
Evalution of the severity
This information helps dermatologists assess the severity of a lesion. It is typically displayed in a dashboard or a summary view for dermatologists. It is focused on helping the dermatologist make a decision about the possible treatment courses.
Example a of full DiagnosticReport
Here's an example of a report that combines various elements.
BodySite
Date of DiagnosticReport
The entropy of this report is 86%. This means that the algorithm is not confident in its predictions.
Select a condition
The algorithm has detected the following conditions.
- Condition A67.51%
- Condition B26.24%
- Condition C1.19%
- Condition D0.99%
- Condition E0.53%
Select other condition
The pathology Nevus has a probability of around 67.51%. The suspicion of malignancy is 26.46%.
Insights
- Refer to specialistMeets referral criteria
- Condition AIt could also be Condition B.
- Probably malignant or pre-malignant62% suspected malignancy
Suspicion of malignancy
62%
Predicted by the algorithm
Report information
Information about the report.
- TimestampFeb 01, 2024, 10:06 PM
- Analysis performed in0.65 seconds
- Image modalityDermatoscopic
- Visual image quality82%
- Sensitivity92.22%
- Specificity89.88%