Artificial Intelligence (AI) for Peripheral Blood Film Analysis




The aim of the software is not to disrupt, but to be integrated within the laboratory workflow. The AI powered software aims to facilitate: blood cell identification, classification, and diagnosis of peripheral blood films (the standard screening and reporting of peripheral blood film duties by a laboratory technologist or haematologist).  The software will sit on the dry bench of the haematology laboratory and act together with a digital microscope as an assistant to the medical laboratory technician or haematologist. Critical blood films will be flagged out earlier which means a shorter turnaround time for the formal reporting of blood films. This means that clinicians will be alerted sooner than later. A longer clinical lead time is an advantage for diseases which are time sensitive (ie. Malaria, acute promyelocytic leukemia) with better clinical outcomes achieved with prompt initiation of definitive treatment and shorter inpatient stay. By assisting with the standard screening and reporting of peripheral blood film duties by a laboratory medical technologist or haematologist, it would help reduce the total bench time required per staff. Gallenco Science Pte Ltd (in partnership with the National Healthcare Group) is looking for interested AI companies or companies with laboratory based solutions for further collaboration in the development, marketing and commercialization of the software.


Advantages 1.    Reduced manhours, shortened turnaround time for peripheral blood film reporting The software helps in reducing the need for physical/manual slide review under light microscopy for slides. This is achieved by the software analysing the digital image captured by a digital microscope and flagging out slides with abnormal cells to the attention of the medical technologist or haematologist. The software can separate out the critical blood films that warrant urgent review from those that can be reported routinely. Slides that have been screened as negative for abnormal or atypical cells then do not need to be physically reviewed. 2.    24-hour operation time The software is not being susceptible to fatigue and can be run continuously. This is a key advantage during night shifts in laboratory when the manpower during non-office hours is significantly reduced. 3.    Automated blood cell identification and diagnosis Accurate identification and detailed analysis of blood cell type and differential count allows the software to arrive at an accurate diagnosis (and give differential diagnoses as well) for peripheral blood film. This aids the reporting technologist or haematologist, to arrive at the correct diagnosis faster. 4.    Rare cells not missed Computer vision is unbiased and thorough. It has ability to detect rare cells amongst the rest of cells in the film and demonstrates quicker processing power.


Laboratory Use: Potential to be integrated within the Lab Information System (LIS), with storage of digitalized peripheral blood film images on the electronic inpatient medical record for medical professionals to access and view. NHG laboratories aims to be the main reference laboratory to the rest of the satellite laboratories in the NHG, and this could possibly be extended to the rest of the hospital clusters and private hospital groups in Singapore, and healthcare systems in the Asia Pacific region. Report Generation: Current peripheral blood film reports are exclusively in word only format. The diagnostic software will be able to incorporate images of blood cells of interest in the report, improving the value of the report to the clinician. Education: Act as a tool for educating and in training laboratory medical technologists-in-training, haematology senior residents and medical students. Smart Phone Application: Be able to be transposed to different platforms such as smart phone applications. Tele-Haematology: Potential joint collaboration with NHG polyclinic laboratories (each polyclinic has its own organic laboratory on site- which reports their own peripheral blood films). Polyclinic based medical laboratory technologists can tap on the software as a point of care diagnostic tool. Images of peripheral blood films can be digitally obtained and sent to a secure cloud-based platform where the software will be able to provide a diagnosis or to screen suspicious cells of interest.


Faster turnaround times for peripheral blood film reporting and improved reporting accuracy and screening for atypical or abnormal films will assist in reducing laboratory man hours for routine bench work as well as improve clinician decision making capacity, translating into improved patient experience and outcomes. It is estimated that the software will help to save 50-60 man-hours of bench time per week (using Khoo Teck Puat laboratory as a model for a medium sized hospital) if it is validated for screening of blood films (assuming 5 minutes of time taken to report each peripheral blood film). This would translate to $20,000 -$30,000 cost savings per month ($150,000-250,000 per year) for a medium sized hospital and ($200,000-$400,000 per year) for a large sized hospital (ie. Tan Tock Seng Hospital). In addition to that, improving the workflow to flag up more serious cases that might otherwise have gone undetected would allow the clinician to intervene earlier, which would improve patient outcome and reduce length of stay, resulting in lower healthcare costs.