iToBoS: Clinical Data Acquisition Study (iToBoS)

iToBoS: Intelligent Total Body Scanner for Early Detection of Melanoma - Clinical Data Acquisition Study

Universitat de Girona, Fundacion Clinic per a la Recerca Biomédica, University of Trieste, Trilateral Research Limited, Robert Bosch Espana Fabrica, IBM Israel - Science and Technology Ltd

The (overarching) iToBoS Project involves 18 research partners spanning the European Union (including UK and Israel), and 1 Australian partner. The overall aim is to develop an AI assisted diagnostic platform for the early detection of melanoma. The Clinical Data Acquisition Study (this study) will recruit 600 participants across 3 international sites (Brisbane, Italy, and Spain). The primary objective is to compare the quality and resolution of conventional dermoscopic images of skin lesions with the full-body images captured by the iToBoS imaging system. Secondary objectives are to collect imaging, clinical and genetic data across the three sites, to create labelled datasets for use in training the iToBoS AI component. Also, to refine and develop a holistic melanoma risk score method to be used for the iToBoS system. Lastly, to assess safety of the iToBoS system. At study site we will aim to recruit 200 participants stratified by risk (of melanoma) categories (low/normal, high, and ultra-high). Participants will be required to attend 3 study visits (months 0, 6 and 12), for total body imaging with the iToBoS system, and dermoscopic images of individual moles. Genetic research and clinical testing are an optional part of the study.

Synopsis The overall aim of the wider iToBoS projectis to build a new imaging and diagnostic tool for the early detection of melanoma, incorporating all the available information of the participant. This holistic assessment tool should incorporate the specific characteristics of every participant in order to enable a personalised, early detection of melanoma. The aim of this prospective clinical study is to collect imaging, phenotypic, clinical and genetic data necessary to develop the iToBoS System as an integrative diagnostic platform for melanoma detection. This will be completed by recruiting 600 participants across three study sites (200 each at Barcelona Spain, Trieste Italy, and Brisbane Australia), to be imaged using the iToBoS system and conventional dermoscopy. The data acquired in this study will be used to achieve three key objectives; To evaluate the resolution and usability (for diagnosis) of the iToBoS images compared with traditional dermoscopic images. (Primary Objective) To collect an image database for training the AI component of the iToBos System. To refine previous melanoma risk assessments to develop a holistic melanoma risk score. A subsequent clinical study (separate protocol), will be conducted to validate the developments from this study using a refined iToBoS System and AI Cognitive Assist Tool. Study design This is a prospective, multi-site, cohort study, recruiting participants with various medical and melanoma history. All participants will be followed for a 12-month period, with study visits scheduled every 6 months (Baseline, months 6 and 12). Participants will be imaged using the iToBoS full body photography system, and individual lesion images will be taken using a handheld dermatoscope. Relevant medical, phenotypic and sun behaviour data will be collected at baseline, and saliva samples collected (optional) for genetic analysis. Enrolment People who expressed interest to participate, were referred to participate, or identified through records (where consent to contact for future research has been granted) as potentially eligible to participate, will be contacted via telephone. During this contact, participants will be provided with study information verbally, and sent a copy of the Participant Information and Consent Form (PICF) by email (or post), and given adequate time to read, discuss with family members, and ask any questions about the PICF and study participation. If the participant would like to proceed, the participant will be assessed for eligibility and their risk for melanoma will be evaluated. Written informed consent will be obtained at the baseline clinic visit. Both the participant and the investigator will sign and date the PICF. All participants who have singed the PICF will be listed on the Participant Screening and Enrolment Log and the Participant Identification Log (or similar). Study Procedures Baseline Clinic Visit (Visit 1, Day 0) At the baseline clinic visit, participants will be provided with detailed information about the study and what will be involved, including any foreseen risks or benefits (e.g., risk relating to data privacy, and inconvenience of time and travel for appointments, and benefits relating to the close monitoring of skin, and imaging). A clinic staff member will thoroughly go through the Participant Information and Consent Form (PICF) and will give the participant an opportunity to ask any questions. The participant will have received a copy of the PICF at least one week prior to the appointment, giving them sufficient time to read in advance and be prepared with any questions they have. A member of the trial staff will review the inclusion and exclusion criteria with the participant and once consent is obtained, the participants will complete a few questionnaires, undergo iToBoS Total Body Photography, dermoscopy and provide a biological saliva sample. Clinical History and Phenotypic Data Capture At baseline a member of the research team will record information from the participant on personal and family medical history (including number of melanomas and non-melanoma skin cancers and ages at diagnoses, number of other skin excisions, solarium use, history of severe sunburns, history of UV-exposure (e.g. chronic, intermittent, constant), medication use, co-morbidities, family history of melanomas), demographic data (including age, date of birth, gender and sex, place of birth, number of years living in current town, marital status, educational level, ancestry) and phenotypic data such as skin and eye colour. Clinical information on any suspicious lesions will be recorded including recent changes, previous biopsy, or previous ablative treatments. This may include requesting previous or future pathology reports. Consent for the collection of pathology reports and relevant medical history will be included in the PICF. All data will be entered directly into an Electronic Data Capture (EDC) system to ensure consistency and confidentiality of participant's information. iToBoS Total Body Photography Participants will undergo Total Body Photography using the iToBos whole-body scanner. The iToBoS System consists of a framework of 15 high-resolution industrial cameras equipped with liquid lenses that arch over a horizontal bed that the participant will lay on. Participants will undress to their underwear with hair tied up if applicable. Participants will be offered the option of wearing disposable underwear to increase the area of skin to be imaged, which may otherwise be covered by their own underwear. Participants will be instructed to lay on a bed, with shoulders and elbows in line, in an angle 90 degrees to the body, with hands placed slightly over the head. The framework of cameras will systematically move along rails of the framework capturing several images at different focus points to capture the whole skin surface. Participants will then be asked to change position to laying on their stomach. The imaging process for each position will take less than 5 minutes. Software is then used to merge images into a single composite picture and generate a 3D body map of the participant's skin. To protect participant privacy, the 3D reconstruction of the participant will be semi-anonymised by creating a 3D avatar, which allows mapping of precise location of moles. Dermoscopy Dermoscopic images will be taken, using a handheld device (standardised between the sites), of any skin lesions that are of concern to the study participant or the medical practitioner conducting the imaging and skin examination. In addition, any naevus 5mm or greater in diameter will also be dermoscopically imaged up to a total of 30 lesions. If the participant has more than 30 lesions, that the clinician believes should be imaged for monitoring, then more than 30 lesions will be imaged for dermoscopy. If participants have less than 30 lesions with a ≥ 5mm diameter, then images of smaller naevi will be taken to make up the total. The clinician will attempt to take dermoscopy images from a range of body sites. These lesion images will be mapped to the total body images taken with the iToBoS System. Additionally, up to 10 dermoscopy images per participant will be taken of benign lesions (e.g. seborrhoeic keratosis, dermatofibroma, angioma) from various body sites, for the purpose of assessing image capabilities at different body locations and for training the AI component of the iToBoS System. Clinical Examination Dermatologically-trained clinicians will conduct a clinical skin examination and record phenotype details on a standard questionnaire. Eye colour and freckling density on the face, dorsum of right hand and shoulders will be recorded (none = 1, mild = 2, moderate = 3, severe = 4). Skin colour will be clinically assessed by Individual Topology Angle (measured using a spectrophotometer) on the ventral upper arm (unexposed) and dorsal forearm (exposed). Height and weight of participants will be measured. Clinical information on any suspicious lesions will be recorded including recent change in size, shape, elevation, bleeding, itch, previous biopsy, previous cryotherapy or other ablative treatments. Management of skin lesions of concern During the study period, the investigators may identify skin lesions that are suspicious for malignancy that require further medical attention. For sites where a dermatologist sees participants face-to-face, the dermatologist will manage the treatment of any suspicious lesions as per standard care (i.e. monitoring, further imaging with Reflectance Confocal Microscopy, biopsy, topical or cryotherapy as per standard care). For the other sites, imaged lesions will be reviewed within a fortnight by the consultant dermatologist together with dermatologically-trained clinicians using the available clinical information, 3D avatar and dermoscopic images to identify any skin lesions suspicious for malignancy that require referral to the participant's regular medical practitioner for management. A study report will be generated including images of any lesions requiring action. This will be sent to the participant's nominated treating practitioner. Any referrals will be followed-up and reports and pathologies will be requested. Collection of Biospecimens Participants will be asked to give a saliva sample using an Oragene DNA self-collection kit (or similar). Samples will be sent to an accredited facility for processing and will be genotyped using an array which will detect common variants in known low to moderate risk susceptibility genes. This genetic information will be combined with clinical risk factors to create a holistic risk score. This method builds on previously validated risk scoring methods, and this study will be valuable in validating this modified algorithm. Due to time and financial constraints, individual holistic risk scores will not be returned to participants in this study. It is the investigators intention to return the results in a separately funded study, under a separate protocol, where usefulness and psychosocial impact of returning risk assessment results can be evaluated. A subset of participants with family or medical history suggestive of an inherited mutation will be offered additional panel testing of familial melanoma genes. It is predicted that approximately 10% of participants may meet criteria for familial melanoma. In this instance, participants will receive appropriate pre-test genetic counselling to inform them of the benefits and limitations of genetic testing, and the likely outcomes of testing. This is to allow participants to provide informed consent for additional genetic testing. The results of genetic testing for high penetrance familial melanoma genes will be returned to the participant by a genetic counsellor, or a clinician trained by a genetic counsellor. Providing a saliva sample for genetic analysis in this study is optional. Participants may still take part in the study if they do not wish to provide a genetic sample. A separate consent page will be used for participants to provide consent for genetic analysis of saliva sample. Follow-up Clinic Visits Participants will be asked to attend a follow-up clinic visit every 6 months, for a 12-month period (baseline, 6, and 12 months). Follow-up visits will include iToBoS Total Body Photography, and Dermoscopy. Any lesion that was photographed using dermoscopy at baseline, will be repeated at the 12 month follow up visit. At the 6 month follow up visit only suspicious lesions will be photographed using dermoscopy. Furthermore, any new lesions that are concerning will additionally be imaged using dermoscopy. Any suspicious lesions identified will be either followed-up by an onsite dermatologist, or referred to the participant's nominated treating. There will be a ± 8-week window for participants to complete visits 2 and 3.

The iToBoS intervention is a total body imaging device, to image the total skin surface in order to detect and monitor for signs of skin cancer. The imaging process involves laying down on a bed that has a framework of cameras arched over it. The imaging process takes less then 10 minutes, and requires the participant to lay in two different positions (face-up and face-down). The study visit also includes individual dermoscopic images taken of certain moles on the skin. This is done in combination with a clinical skin examination. Participants are given the option of providing a saliva sample for genetic research. Participants are then asked to complete a series of questionnaires. There a three visits in total (month 0, 6 and 12), in which these procedures are repeated (except for saliva sample).

The intelligent total body scanner (iToBoS) device will be an AI diagnostic platform for early detection for melanoma, which includes a novel total-body high-resolution scanner and a Computer Aided Diagnostics (CAD) tool. The prototype iToBoS imaging device used in this study, will not have the integrated CAD system. The device consists of a horizontal bed on which the participant would lie on. The bed will slide under a series of 5 arc shaped rails that have a total of 15 vision units mounted on rails (3 vision units per arc). The vision units incorporate high-resolution cameras and LED lighting system. When imaging is initiated by an operator, the vision units systematically move along the arch rails taking numerous images capturing majority of the skin surface. The imaging process will take approx. 10 minutes. Purpose built software will be used to stitch images together to create a body avatar on which skin lesions can be viewed and monitored with.

The primary outcome measure of this study is the assessment of image quality of total-body photography taken with the iToBoS system. A clinical and quantitative comparison of conventional dermoscopy images of individual lesions with iToBoS images will be systematically assessed by a panel of dermatologically trained clinicians and research assistants. The panel will be asked to independently assess the iToBoS image and dermoscopic images as acceptable or not acceptable. Three panel members will assess each image, with consensus being considered with agreement of 2 or more. The order of the images presented to each panel member will be randomised.

The secondary outcome measure will include classification, by field experts, of dermoscopic images into 9 diagnostic categories (e.g., naevus, melanoma, basal cell carcinoma, squamous cell carcinoma, actinic keratosis & intraepidermal squamous cell carcinoma, benign keratosis, dermatofibroma, vascular tumour, and other). The field experts will also be asked to provide a binary classification of suspicious or benign. Sequential dermoscopy images of the same lesion taken over different visits will additionally be assessed for clinically significant phenotypic change (yes/no). This dataset of labelled lesion images will be used to train the iToBoS artificial intelligence algorithms for development of computer aided diagnosis.

An additional secondary outcome measure is the holistic risk score algorithm developed as part of the wider iToBoS project, which will be tested on the complete prospective dataset from the initial visits of all 600 individuals in three centres. The ability of the model to discriminate between case and control individuals will be assessed. In addition, the net reclassification improvement index will be calculated to quantify the sensitivity and specificity of model classification. These factors will enable assessment as to the relative contribution of individual risk factors in accurate risk stratification. The goal of this task will be to refine the clinical protocol for stratification of participants into low/average, high and ultra-high-risk categories to be used in a subsequent clinical validation study (separate protocol).

A protocol for recording any Adverse Events, and Serious Adverse Events is established. Adverse events are catogrised by the nature of the event, the intensity (mild/moderate/severe), the duration, and the relatedness to the investigational device (probably/possibly/not related), and the outcome is recorded (recovered/resoved, recovering/resolving, not recovered/not resolved, recovered with sequelae, fatal, unknown). AEs and SAEs are not expected in the study, and therefore any reported AEs and SAEs will be reviewed by the trial investigator within 48 hours.

Images of the same skin lesions taken at baseline, compared to the participant's last study visit at 12 months will be used to detect any changes in the lesion. This will be important in training the iToBoS AI component to detect changes in skin lesions. An expert panel of clinicians will classify subsequent images as 'not changed', or 'changed'. Cohen's Kappa will be used to compare agreement between the lesion classification of change vs. no change.

Inclusion Criteria: Aged 18 years or older Able to provide written informed consent Willing to attend 3 clinical visits over a 12-month period Willing to provide a genetic sample (optional) Willing to follow the clinical procedures (e.g. no physical restrictions from imaging process) Have a BMI between 18.5 to 40 kg/m2 Have a height between 140 - 190 cm Have a thorax height (participant lying face up) of 20 - 45 cm Have an elbow to elbow width (breadth) of 40-50cm Exclusion Criteria: Have a pacemaker Are pregnant, or planning to become pregnant Any condition in which the investigator's opinion may adversely affect the participant's ability to complete the study, or its measures, or which may pose significant risk to the participant.

The iToBoS consortium will organise two competitive 'skin image analysis' challenges where groups can participate in solving new problems on: 1) lesion detection and boundary segmentationTotal Body Photography images, and 2) on lesion classification using minimal (non-identifying) participant clinical data, genotyping results, and the lesion images extracted from Total Body Photography images. To facilitate these challenges we are sharing limted patient data (non-identifiable images, minimal clinical information, a genetic risk score, and minimal genotype information).

The datasets will be shared in international conferences, for example MICCAI (Medical Image Computing and Computer Assisted Intervention) or similar, where the challenge and corresponding workshop will run. The details are not finalised at this stage.