Introduction

Nowadays people are constantly expose to sun rays, which rises the incidence of cutaneous lesions. In order to help the specialists on performing the diagnose of skin lesions, specially skin cancer, ATONUS has developed a computational vision system to acquire, store and analyze images of the surface of the human skin. The intention is to decrease the subjectivity on skin lesion analysis through both the use of the computational tools available on the system and the remote transference of the acquired images between specialists of the medical area, via internet,  for a remote diagnosis.

The Problem Increasing incidence of cutaneous lesions Subjectivity in the cutaneous lesions analysis and diagnoses Deficit in the exchange of information between experts in the subject

The system described on this document uses computational vision techniques to classify the cutaneous lesions trough a morphologic analysis of the digital images. The automatic analysis is done by using as much as the rule ABCD of dermatology as the method "seven point-checklist". The rule  ABCD  extracts the following informations about the cutaneous lesion:   Asymmetry between two axis, Border regularity in 8 axis, presence of several  Colors, presence of D ifferential structures over the lesion. The method "seven point-checklist" computes the potential of the lesion be malignant trough the following characteristics: itchy feeling or other smaller change on the sensibility , lesion bigger than 1 cm of diameter, growth historical or other change in the lesion, defective contour, defective and varied colors, inflammation near or on the lesion contour, bleeding.

Scope

The current document presents general characteristics of the cutaneous lesions analysis system and the system specifications to the development and implantation of this project.

The system architecture proposal describes the fuctional requirements of each subsystem to be developed, user requirements, accomplishment and trust.

Glossary

•  LC - Cutaneous Lesions.
• CLC - Cutaneous Lesions Center treatment.
• SVD - Video-Episcope Subsystem and cutaneous lesions analysis.
• SAP - Pathological Anatomy Subsystem (NOT IMPLEMENTED ON THIS PROJECT).
• STM - Telemedicine Subsystem.

General Description of the System

The Coputational System for cutaneous lesions analysis provided by ATONUS is composed by three subsystems:

• Video-Episcope Subsystem and cutaneous lesions analysis (SVD): it is composed by the video-episcope equipments, computer with monitor, ethernet network adapter, image grabber, scanner and printer. The software available allows you to register the patients data, acquire and process images of cutaneous lesions and do the lesions morphologic analysis (rule ABCD of dermatology). Others exams are: "seven-point check list", clinical ABCD and detailed characteristics. It is part of the system a manual episcope (no video) allowing the doctor to compare visually the acquired image with the real image of the lesion.
• Pathological Anatomy Subsystem (SAP): it is composed by an appropriate microscope to the pathological anatomy analysis, conected to a video-camera, image grabber, computer with a monitor, scanner, printer and ethernet network adapter. The software available in this subsystem will have functions like capture, process and store the images in a database of removed tissue of cutaneous lesions. Not available yet.
• Telemedicine Subsystem (STM): it is composed by a fax/modem board conected to the computer of the  SVD subsystem and a software to do the exchange of informations and images via local network (LAN) or via Internet (WAN). This subsystem has a site in the internet to store the exchange informations between several remote SVD  subsystems, performing the work methodology of information exchange between doctors trough the telemedicine. This way it is used the experience of several professionals giving a more reliable diagnoses.

System Architecture

The Cutaneous Lesions analysis (LC) can be accomplished by the doctor trough 6 distinct ways:

• visual analysis of the LC,
• visual analysis of the LC with the manual episcope,
• visual analysis of the LC with the video-episcope,
• computational analysis of images of the LC in a workstation,
• remote analysis trough information exchange (images and data) between doctors in remote workstations (telemedicine).

To the accomplishment of the LC visual analysis the doctor can or can't dispose of an episcope. In the case of LC analysis without the episcope the doctor use the rule clinical ABCD and the method "seven-point checklist ". In the case of LC analysis with the episcope the doctor use the rule ABCD of  dermatology . The episcope can or can't dispose of a video-camera.

On the other hand, the last two types of LC analysis, which correspond to the Atonus products, need the images to be captured by the computer.

The following picture shows the system of cutaneous lesions analysis suggested, emphasizing the subsystems of the system.

                                                                         Click on the Image to  enlarge.

Functional requirements of the system

Video-Episcope Subsystem and Cutaneous Lesions Analysis (SVD)

Video-episcope Module

• Visualization of LC images,
• Capture of LC images trhough the equipment composed by a video-episcope (camera, light ring and font) connected to the computer with a image grabber,
• Images recorded on the format JPG and MPEG,
• Video-episcope calibration to enable the quantitative measures from the captured images,
• Map tabulation of the lesions and the video-episcope images to the selected patient,
• Share the images acquired on the exams trough the local network or Intranet,
• Lesions map to inform the localization of the lesion in the patient body.

Module of Cutaneous Lesions Analysis

• Visualization of the LC images acquired in the Video-episcope Module ,
• Compare two or more images to pursue the lesion evolution (reclaim the image orientation if it's necessary),
• Capture of image in the scanner,
• Manipulation of image of any size,
• Images recorded on the format JPG and MPEG,
• Image processing (contrast, straining, etc),
• Save the exams results in disk,
• Automatic analysis of LC images based in radiometric parameters,
• Classify LC images by it's nature: benignant (nevi or keratosis) or malignant (dysplastic nevi or melanoma),
• Lesions map to inform the localization of the lesion in the patient body,
• Seven-point checklist: mutiple option questions that need to be field by the expert to help to categorize the lesion,
• Use of the diagnoses of pathological anatomy proceeding of the captured image analysis from the microscope to help in the LC classification,
• Remote access of the store informations, via Local Network (LAN) or Intranet,
• Register of general data from the patients, anamnese from the exam and from the images captured in the scanner.

Pathological-Anatomy Subsystem (SAP)

• Visualization of the pathological anatomy images of the LC,
• Capture of pathological anatomy images of the LC trough a miscroscope with a video camera conected to a computer with image grabber,
• Images recorded on the format JPG and MPEG,
• Register of pathologial anatomy images of the LC from the patient showing the lesion localization in the patient body using the patient lesion map,
• Share the images acquired on the exams trough a local network or Intranet.

Telemedicine Subsystem (STM)

• VISUALMED site to access the data sent by the workstations,
• Inclusion of images and texts sent by registered doctors,
• Registered doctors can remove the images and texts from the site,
• Exchange of messages between registered doctors,
• Register of doctors with respective access level,
• Register of informative messages of relevant events to the users comunity.
• Visualization, in the remote workstations, of LC and pathological anatomy images store in the site and released according to the user access level,
• Transference, to remote workstations, of images to sent to remote diagnoses.

Gratefulness

This project has been suported by FAPESP, trial 97/07390-4, in the Program of Technolonogy Innovation in Small Business.