Can we use artificial intelligence tools for automatic analysis of bone marrow samples?
Not Applicable
- Conditions
- Training of convolutional neural network algorithms for identification and counting of cellular lineages and specific cell types of bone marrowNot Applicable
- Registration Number
- ISRCTN10382623
- Lead Sponsor
- SpotLab S.L.
- Brief Summary
Not available
- Detailed Description
Not available
Recruitment & Eligibility
- Status
- Ongoing
- Sex
- All
- Target Recruitment
- 150
Inclusion Criteria
Patients:
1. Suspected hematological disease
2. Signed informed consent
Bone marrow samples:
1. Good quality BMA sample (with proper staining and lump to provide sufficient quality and quantity)
Professionals/experts:
1. Sanitary professionals of the National Health System (Doctors, Cytologists) working at Hematology Department of the Hospital Universitario 12 Octubre with microscopy experience on hematological diseases
Exclusion Criteria
Patients:
1. Individuals unwilling to participate in the study
2. Unspecified reasons that, in the opinion of the investigator or sponsor, make the subject unsuitable for enrollment
Bone marrow samples:
1. BMA samples that do not have a good quality stain
2. BMA samples with insufficient lump
Study & Design
- Study Type
- Observational
- Study Design
- Not specified
- Primary Outcome Measures
Name Time Method <br> 1. Number of samples analysed by web platform (TeleSpot) and analysis time per sample<br> 2. Professionals' satisfaction measured with the new system measured by a usability report based on the results from a system usability scale (SUS) and AdaptaSpot Usability Questionnaire evaluating the remote analysis process. The SUS and the product questionnaires are completed every three months during the length of the study<br>
- Secondary Outcome Measures
Name Time Method <br> 1. Number of digitized bone marrow aspirate images correctly marked and tagged<br> 2. Accuracy of the AI algorithm developed and the % of agreement among experts and AI algorithm. Cell-type classification performance will be tested by assessing the prediction quality of the algorithm in the validation set compared to the ground truth annotated by the specialist during the labelling phase.<br>