• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

New Method Enables Smartphone-Based Diagnosis of Sickle Cell Disease


Patients with sickle cell disease who do not receive treatment can die from the disease, but diagnostic resources are scarce in some parts of the world where the disease is most prevalent. A new smartphone-based method could change that.

A new mobile phone-based method of screening for sickle cell disease (SCD) could dramatically reshape the identification and care of patients, particularly in under-resourced parts of the world.

The method, described in the journal NPJ Digital Medicine, would significantly reduce the cost of such screenings, which the authors say would make it easier to identify and treat children and adults who have the disease.

Millions of people around the world suffer from SCD, but it is most prominent in Africa, where some 200,000-300,000 children are believed to be born with the disease each year. More than half of those children will die from the disease, even though many of those deaths would likely be preventable if the cases were timely identified and the patients received treatment.

Investigators from the University of California, Los Angeles, wanted to improve those stark statistics by leveraging the power of mobile phone technology and deep learning. The team developed a system that uses 2 distinct neural learning networks, as well as a smartphone microscope. The “microscope” in this case is an opto-mechanical attachment that pairs with the phone’s back-facing camera. The external parts cost around $60, the authors said.

The system begins with the collection of a patient blood smear, which is then placed under the digital microscope. The first neural network enhances and standardizes the image so that it is rendered at the quality of a laboratory microscope. The second neural network then performs a semantic segmentation, differentiating between healthy and sickle cells in the blood.

To study their method, corresponding author Yair Rivenson, PhD, and colleagues used blood smears from 96 patients, 32 of whom were known to have SCD. They performed a blind analysis using different smartphones to see whether the system could correctly identify the patients with SCD. The system achieved 98% accuracy; there was just 1 false-negative and 1 false-positive.

“Our results and analysis demonstrate that the presented method, enabled by smartphone microscopy and deep learning, is robust to perform SCD diagnosis by automated processing of blood smears,” Rivenson and colleagues report.

The system also works much more quickly than manual analysis, with a total analysis time of just under 7 seconds. In the current study, the statistical analysis was done on a computer, but the authors note that the computer can be local or remote. A smartphone app could also be used to do the computation, which would result in an even shorter amount of overall processing time

There are a pair of important limitations to their findings. Blood smears are not an appropriate tool for screening infants, Rivenson and colleagues note, due to a significant risk of false-negatives. Thus, the smartphone-based system would not work in a newborn screening setting, but only for older children and adults. Also, the system is unable to distinguish between sickle cell genotypes, so some patients diagnosed with the smartphone system may require follow-up testing.

Still, the method’s low cost and fast results make it an ideal tool for children and adults who have not undergone newborn screening, the authors say.

“We believe that it would be particularly useful in settings, where the existing point of care technologies are not ideal due to cost, need for reagents etc. as the presented SCD screening method is rapid, cost-effective and the required sample preparation is minimal,” they conclude.


de Haan K, Ceylan Koydemir H, Rivenson Y, et al. Automated screening of sickle cells using a smartphone-based microscope and deep learning. NPJ Digit Med. [Published online May 22, 2020]. doi:10.1038/s41746-020-0282-y

Related Videos
Robert K. Massie, Jr, PhD, MA, Society for Progress
All of the panelists for this series
All of the panelists for this series
All of the panelists for this series
Related Content
© 2023 MJH Life Sciences
All rights reserved.